Nominal height of Cerenkov box is 333cm from top of box to bottom of bracket into which adjustable foot is attached. The foot adds an additional ~4cm
This was measured with tape measure by DMW DJL, Jan 2002.
Extra space is required on both sides of box versus nominal dimensions given in Ed's memory dump.
Beam
|
V
+-------------+
| |
| |
+-------------+
19cm 7cm
gas windows to
system access
plumbing
Jim Kilmer High-rise 11 East
phone 2637
pager 630-722-1209
home 879-7661
kilmer@fnal.gov
Support Mirrors in Dave Christian's office
(Hirise 10 West towards North)
Steve Pordes (Por-dez) head of PPD could help finding
cables for bases.
Drawings at Nevis: contact Brian Cole, Columbia
Wilson Hall
N
/
+--+ +--+
| | | |
|W | | E|
| | | |
+--+ +--+
Flags
Summary of May 2002 trip (PMT work)
November 21, 2003
Phototube installation
green = type 2
yellow = type 3
Remaing to install (waiting for wavelength coating)
35=3 38=3
37=3 4=2 8=2 29=2 32=2
3=2 7=1
33=2 36=2 2=2 6=1
1=2 5=1
Base coating person: Eileen Hahn 4164
December 15-18, 2003
tube location=base #
35=91 38=85 39=69 28=86 32=94
4=93 8=95
34=10 37=88 3=96 7=45
2=92 6=80
33=89 36=89 1=90 5=63
Spider box ribbon cables
SB1 X6-7
SB2 X6-8
SB3 X6-6
SB Spare Top X6-5
SB4 X6-2
SB5 X6-3
SB6 X6-4
SB Spare Bottom X6-1
Minibox low voltage cables
MB1 L3
MB2 L4
MB3 L1
MB4 L2
Pierrick, Needs 8 slots in upstream crate
Using LeCroy 3377 TDCs (32 channel 0.5 ns resolution
Newly Coated Tubes, were never tested
28131 93 (base was formerly labelled #55, but was apparently a duplicate)
27068 85
27067 88
26509 87
28216 not used
One can print using:
flpr -q e907_hp4si - single sided
and
flpr -q e907_hp4si_d - double sided printing
January 14-16, 2004
Called Carl Rosenfeld about delay board
250nsec delay
completely passive (no power), chip is just a delay line
chips are symmetric, so can plug in up-side down
Our boards:
Lemo input on chip side
Ribbon output on chip side
lemo output on solder side (just shorted to ribbon output, so lemo
will work, but no impedence match)
Our boards are different from TOF
TOF boards use lemo input (chip side) but split output 90/10 onto two
lemos, ribbon ignored
Solder Side
2 1 18 17 34 33
4 3 20 19 36-35
...
16-15 32-31 48-47
Chip Side
33-34 17-18 1-2
35-36 19-20 3-4
...
47-48 31-32 15-16
Minibox threshold test
term (ohm) measured connected
1 16.8 22.7 5.71 MegaOhm (I believe this is correct)
2 25.2 31.2 5.71 MOhm (I believe this is correct)
3 infinity infinity 206 Ohm (problem)
4 52.2 55.9 1066 Ohm (threshold and low voltage
location 76 not connected) (problem)
use kinit -a dmw to get kerberos tickets that include afs
on fnalu use 'setup netscape' to get access to netscape
CKOV DAQ (all Lecroy except where noted)
ADC 4300 (some 4300B, does it matter)? slots 2,3,4
Disc 4415A, TDC 3377, slots 6,7 9,10 12,13
DAC Bira 5408 slot 15
Output register DSP PR612 slot 16
LV power supply test (supplies were incorrectly labelled! -> fixed by me)
Top MB1 -5.63 V (measured at connector on box, no load)
Top MB2 -5.65
Bot MB3 -5.71
Bot MB4 -5.73
Gas System
5000 sccm flow controller
trip valve settings
.25in top
2in bot
bubbler 0.5 oil top
Minibox test (check resistance across fan-out connector
MB3 most bases are 2.9MOhm, some are 1k-11k (I believe this is bad)
Low voltage connector, female end
A,B=Gnd; C,D=-5.2 V expected (at detector side, under full load)
February 4-6, 2004
Page system works now, press page and then zero
Base LV&threshold resistance test
Connectors are 25 pair, last pair is not used
MB3: Bot (Beam left)
pair base location thresh lv
# # # ohm ohm
1 41 62 11k 9k
20 8 55 10k 8k
Total with pair 1 and 20 disconnected thresh=lv=1.97 Mohm
MB4: Bot (Beam right)
pair base location thresh lv
# # # ohm ohm
2 72 68 8k 7k
4 46 65 53k 56k
7 43 95 10k 15k
9 66 77 2k 200
16 71 72 1.1M 1.1M
18 62 73 5k 3k
Total with pairs 2,4,7,9,16,18 discon thresh=lv=2,04 Mohm
MB1 Top Beam Left Total=5.60 MOhm
pair base loc thresh lv lvconnector pair
# # # ohm ohm # #
7 77 40 53k 52k 7bar 19
8* 69 39 open*
* = fixed by pushing on harder/twisting
total (Ohm) doesn't matter if both lv and thresh. fanout is plugged,
does matter if LV power plugged into box
before pair 8 was fixed
5.60M no LV power
32k no LV power
after pair 8 fixed
200 with LV power
MB2 Top Beam Right Total=2kOhm
because skipped on connector, pair 25 is used on this one
1 open 1bar 25
14 18 2.8M 2bar 24* was open
4 18 46 10k 8k 3bar 23
5 84 43 540k 570k 4bar 22
13* 56 42 open, push on tight, very flakey
14* 76 30 open, push on tight, flakey
18 16 21 26k 25k 17bar 9
20 13 29 16k 24k 19bar 7
23 94 32 5k 4K 22bar 4
25 39 27 577k 500k 24bar 2
open 25bar 1
* = fixed
Connected HV to Lecroy 1440
1443PF (positive) Modules (16 channel)
slot HV cable
----- -------------
6 1-16
5 17-32
4 33-48
3 49-64
2 65-80
1 80-96
Possible leak at fitting of valve on output of flow controller (snoop-it)
Gas system normal operation with CO2
o about 10psi on gauge on supply
o both highflow and lowflow valves open
o both rotameters (high and veryhigh) closed
o flowcontroller actively metering gas
o pressure relief valve at top of detector set at .25 inches
(so it is possible to waste gas if flow controller is set too high
and top valve is venting)
o pressure gauge at bottom of detector reads ~ 0.5 inches
o note that the scale on the top/bottom gaugesare different
February 11-12, 2004
miniboxes now have 2x 3Ampr fuses (one fuse for 12 channels)
Bob Jones put the LV fanout connector back in upside down, redid myself
MB1 by itself
-5.66 Volts no load
-4.60 Volts with all 24 bases
~1.5Amps according to crude gauge on power supply
MB2 by istelf
-5.68 Volts no load
-4.60 Volts with all 24 bases
~1.5Amps according to power supply
MB1&MB2
-4.6 Volts with all 48 bases
~3.5 Amps according to power supply
MB3&MB4
-5.63V at power supply
MB3 -4.67V on bases
MB4 -4.64 V on bases
Current gauge does not appear to work on this supply
Raised both power supplies to -5.68V (max it will go)
Wrote hvc program for controlling hv, in use by Mike Longo and
Carl Rosenfeld
March 8-10, 2004
Fixed branch highway to CKOV crates
Rack 6, VME crate contains beamppc and camac controller
connects to crate 1 Rack 3 -> crate 2 Rack 6 -> crate 3 rack 12 -> crate 4 rack 13
CKOV crates are 3&4
Branch highway from rack 6 to rack 2 is made from two 25 foot cables with 1foot union
the union is in cable tray opposite rack 7
Removed unioin
o one of the pins was not fully inserted, disassembled connector and repaired
o screws on face of connector just barely touch those on the other connector
replaced screws on one end of union with lower profile, now does not touch
Three low bits were always zero. After fiddling with union ok,
but, address high bit (=16) not addressable
push on cable going into crate 4, now everything ok
Testing the branch highway
Logon to e907daq and then ssh to beamppc (use your own account or daq if you
don't have one)
source /usr/local/mipp/setup-mipp-online.csh
initCAMAC 1 [crate]
execCAMAC 1 [crate] [slot] 17 1 [value] # write value to slot in crate
execCAMAC 1 [crate] [slot] 1 1 # read value from slot in crate
Matt, (Grad. student or postdoc?) from Colorado working on TOF
Fixed 1443PF HV in 1440.
Problem was that HV readback was only tens of volts at HV settings of few hundred,
it turns out it was caused by positive current limit (read with RL, set with LI 255)
HV power is supplied by 1442 modules plugged into mainframe. There are two
bays for these modules on the left side of the unit. The left-most module was missing,
there was a face plate with LBL stickers, but no unit!
Got two 1442 modules from prep, installed one and labelled the other as spare. Also
returned one 1442 module to prep that Carl said was bad.
Note that the ribbon connector is keyed, but you plug it in the other way. If
ribbon cable is up-side down, some of the LEDs on the 1445 will flash slowly
instead of being steady on.
The mainframe modules tilt forward. Unscrew only the top and tilt down. There is
a spring loaded stop that holds the unit open. Very convenient.
Hank the prep HV tech said that the left-most 1442 powers slots 0-7, with both positive
and negative. The manual says otherwise, I did not verify this.
Analog signal check on channel 84
At the base, need digital scope to see signal
(can use analog scope but can only measure the rate by looking at trigger light,
won't be able to see the trace)
50 Ohm DC, -30mV trigger at 1800V see 10nsec wide pulse at about a hertz
At the end of the analog cable, the signal is a bit fatter (from the delay line), lower
threshold to -25mV
On output of delay boards signal is really small and very long (tested without ribbon
plugged in, lemo output is on solder side, note that the ribbon and lemo are
just shorted, remember that our boards are different than the TOF boards)
go to .1microsec per division and trigger at -5mV
See lots of ringing in signal now, impedances are not matched. ADC counts should be
the same.
Threshold DAC BiRa 5408
on beamppc run ~lange6/Helpers/bin/test5408, e.g.
test5408 1 4 16 0 -.5
read back should be same as set voltage
Tested output on lemo going into minibox, within .001 of
DAC setting.
Tested output at subminix cable going to base. Note that
with Low voltage on, there was -.4 volts on the threshold
line even with the threshold set to zero.
thresh lv off lv on
0.00 0 -.401
-0.50 -.483 -.590
-0.75 -.724
-1.00 -.960 -1.042
-1.25 -1.195
-1.50 -1.423 -1.490
-1.75 -1.641
-2.00 -1.735* -1.935
* going down down (towards zero) .001 every 15 seconds
Card locations
Crate 4
2,3,4 ADC FERA 4300
6&7, 9&10, 12&13 Disc 4415A, TDC 3377
14 Disc 4413 (lemo in, rats)
15 Scaler 2551
16 DAC 5408
17 NIM to ECL RFD ?
18&19 Gate Delay 2323
Crate 3
2,3,4 ADC FERA 4300
6&7, 9&10, 12&13 Disc 4415A, TDC 3377
Notes from March 15-19, 2004 trip
Rearranged crate 4
14 NIM to ECL
15&16 Gate Delay 2323
17 DAC 5408
18 Scaler 4434
19 Scaler 2551
20 Disc 4415A
21 Pulser
Must initialize crates for 2551 scaler to work!, i.e. initCAMAC 1 4
Crate 4 top address bit dropped again. Touched branch highway cables in rack 6 seemed to fix it
Checked digital output on scope
Trigger on analog output (directly from base) 50 Ohm DC, 50mV, 10nsec, -15 mV threshold
Short (8foot) ribbon cable from spider box to ECL-NIM modules in NIM carte
Look at NIM output of ECL-NIM on digital scope
Channel 84 (bottom)
MB3,
Spider box 4, channel 12
HV slot 1 channel 3 (counting from 0)
digital output has -300 mV DC offset (at 50Ohm) (NB through ECL-NIM module)
digital signal shoots back to zero volts and then down to -2V.
jagged and triangular with a width of about 150 nsec
Digital signal after long ribbon cable (to rack 13) and then to NIM-ECL
-300mV offset, big positive pulse (-300 to 0 mv) width ~40 nsec followed
by big negative pulse (-300 to -500 mV) width ~40 nsec, but 150 nsec after first pulse
Digital lines must be inverted!
Invert ribbon cable (now find Channel 84 on channel 6), DC offset is gone, nice
negative pulse -0.5 volts, primary width 50nsec, but multiple trailing pulses of diminishing
magnitude, typically see two smaller pulses after main one.
Analog signal on channel 84 sometimes disappears. Touching subminex connectors on the base seems to solve this problem.
CAMAC has intermittent problem with highest address bit. Touching rack 6 branch highway appears to fix it.
Digitial output does not appear unless threshold set to -1.8 (For Channel 84 this is MB3 SB4).
With new frankenstein cables here is how to get digital signal
Both + - should have 56Ohm pulldown resistor to -2V
(Note that True+ is white base output labelled -, True - is blue base output labelled +)
Make sure threshold is set to -1.8V and LV and HV are on.
To see signal on scope, pull out + or - signal (with powered pulldowns) and couple
scope with AC 1Mohm. Will see 1.5V by 40 nsec wide pulse (positive going for +
and negavite going for -, note that there is a DC offset that keeps the signals
entirely negative)
Note that there is a smaller (~10 times smaller) signal that is 200nsec after the real
signal. This is probably a reflectiioin somewhere along the cable. This is probably not
due to the ribbon to lemo conversion on the scope because if you put the ribbon
directly into the discriminator you can lower the threshold enough to trigger on both
the big and little signals.
Set all Discriminators (4415A) to minimum (100nsec) and 300mV (on test port). If
truly differential this is equivalent to 150mV.
The frankenstein cable should be male with female close by to accept terminating plug,
then female on the other end. Leave space between the male and female end, because
the female end must duck unto cables near the discriminator input.
LED bench tests
Single 325 Ohm resistor in series with n LEDs in parallel
Green LED is visible at 1.80 Volts, supposed to be 20 mA at 2.2VF
Yellow is 10 mA at 2.2VF
testing green LED
Vbias Vresistor
nLeD 1 2 5
V mV mV mV
1.80 41 58 77
1.90 101 125 148
2.00 179 203 236
2.10 259 286 321
World's greatest pulser: 1340 Pulse generator (from Prep)
has variable output 0-10V as well as pulse width and frequency.
Provided that pulse generator and NIM linear fanout can drive LEDs
Reorganized DAQ crates again
Slot Crate 4 Crate 4
1&2 2323A Gate Delay Empty
3-5 4300 ADC Same
6 RFD04 NIMto ECL Same
7&8 4415A Disc/3377TDC Same
9&10 " "
11&12 " "
13&14 4433 Scaler Same
15 Empty
16 5408 DAC
17&18 Empty
19 2551 Scaler
20 4416B Disc
Note that third ADC and TDC in each crate should be terminated (for gate chain).
Modified jumpers on 2323A so that ECL output is Delayed (jumper inside, see manual)
All Modules (ADC,TDC,Sclaers) are working and can be readout.
4415A discriminator works on base digital output with -2V through 56Ohm pulldowns.
Split analog signal from base, one half to 4416 discriminator, other half to ADC.
See charge on ADC before and after delay boards from South Carolina!
Notes from March 30-April3, 2004 trip (Lange)
o Hardware
Found that I could not control the dac. Did not appear
to be a problem with camac bits being dropped. Put in a
'spare', which worked, but does not have the right
voltage range (need -10-10, spares are 0-10). Took
both the working and broken modules to prep for repair.
Say that early next week will be done. Will ask Doug
to get them.
Plugged in the digital output with swap cable directly
into one tdc. See lots of hits (though threshold is not connected...)
This might indicate that tdc would work for us without kludge
cable
Talked to Mike B. about discriminator inputs...Didn't have a
solution (expected it to work, save our strange ecl signals)
Got from Mike 24 10 pin 56 ohm sips. Basically cleaned them out...
Also found some cable and a few (but not enough) connectors. Did not
make it to the stockroom to see if they have the male connectors or not.
Its really hot near our crates - in fact they are hot to the touch. Should
ask about getting a fan in that area
Got a trigger line .. not labeled yet.
Found that we were not seeing the beam trigger pulse. TOF also was not. Traced
by Andre/Pierrick to be a very short pulse (5ns). Now fixed.
Both 2323a modules ECL for channel 0 is the delay. For the
2323a in crate 3, ECL for channel 1 is not delayed. For
the 2323a in crate 4, it is. I'm not sure how the trigger
timing is working. Given that the minimum delay is 50 ns
we may need to use the non-delayed output of the 2323a.
Kludge cables are made. 1 is connected to 1 4 12 tdc
the other 5 are in the plastic bag in the box in the
counting house.
o Where to find stuff in daq-land
nominal ckov config file ::/ppcboot/common/usr/local/mipp/online/config/ckov_config.xml
log file from current run ::/ppcboot/common/usr/local/mipp/log/ckov.log
log file from previous runs ::/ppcboot/common/usr/local/mipp/log/archive/ckov.log.*.gz
o Software:
--> Got CkovReadout executable working. Defined data format for
the CKOV: Code lives in Detector/Ckov/CkovDetector.cxx, needs
to be consistent with MippIo/MippIoCkovBlock.cxx also.
// 0xADC
// number of ADC words read out
// wordcount*0x800 + crate*0x100 + slot of the ADC
// ADC word 1
// ...
// ADC word N
// ........... for other ADC's
// wordcount*0x800 + crate*0x100 + slot of the ADC
// ADC word 1
// ...
// ADC word N
// 0x2DC0
// number of TDC words read out
// wordcount*0x800 + crate*0x100 + slot of the TDC
// TDC word 1
// ...
// TDC word N
// ........... for other TDC's
// wordcount*0x800 + crate*0x100 + slot of the TDC
// TDC word 1
// ...
// TDC word N
--> Learned how to commit code to production area
--> for some packages there are two areas
/ppcboot/common/usr/local/mipp/online (powerpc code)
/usr/local/mipp/online (e907daq code)
--> in my /ppcboot/common/home/lange6/prod and
/home/lange6/online/
areas I have set up compiling areas to allow for
easily updating production code (ie, make install is
a dangerous command in these areas)
--> Added code to RawData, MippIo, Raw2Root for parsing the
raw data format of the ckov. Most of it is taked from
the tof code, which is nominally working.
--> setup environment:
cd /home/lange6/mippSoft/040402
setenv code_907 /afs/fnal.gov/files/code/e907
source $code_907/mippsoft/bin/setup_linux.csh
--> Command for Raw2Root (convert raw format data into
a root file)
bin/Linux2-GCC/raw2root /data/1/mipp00004938.0000.ra
--> Command to run over raw root file
bin/Linux2-GCC/anamipp -i raw2root.root -x JobControl/xml/test/job.xml -o test1.root -n 100
--> Wrote a simple module to make a th1f of my adc counts in
/home/lange6/mippSoft/040402/JobControl/test/CkovTest.cc
Need to find a home for ckov analysis code.
o Co2
--> Terry gave me the quick tour of the gas computer. Shows a
nominal rate of 4 ft^3/day. Need to check the variation
of atm. pressure. Plot of co2 use vs. time shows significant
variation with time. Rate of usage is 3x projected. May be an
issue given c4f10 cost. Did not manage to re-leakcheck the
ckov. Should ask if we know that mass-flow usage is consistent
with bottle usage. (may not be known...)
o HV
--> Problems with Andrew's HV program. Trips channels off that are
more than 10V from nominal. However, a program setting of 1800
volts results in a reading of 1807-1812, so this is a problem
for us. Turning off the hv_monitor program is a temp. fix.
As of 040403 1am, Andrew has loosened the limits for the ckov
channels.
--> Our HV is still off by default. Can be loaded via ckov.hv
file.
o Data
--> a few 'long' runs with ckov at 1800V and beam. Nominally
run 4938 is a 'good' run with the ckov running. The delay wa
50 ns, delayed width is 300ns. Have not checked that this is 'ok'
or not.
--> Run 4971 is 50K of pulser triggers. Trigger prescales
were 500 for pulse generator prescale (at the top)
20 for pulse gen 1 prescale (at the bottom)
20 for pulse gen 2 prescale (at the bottom)
this gave 10s of Hz, which is about what the
daq seems to handle.
--> Run 4972 is taken just after 4972 with the ckov HV on
is with beam.
data for each pmt plots
email
Notes from April 7-9, 2004 trip (Wright)
Discriminators in slot 17&18 of Crate 4 go with TDCs in Crate 3
In Co2, the pion threshold is about 5 GeV, so in a 40 GeV
beam, at least the primary particles should radiate, except
for protons.
LV was off for Ckov top, switched on at 7:55pm 4-7-04
(bottom was already on)
config file for ckov daq (controls gate delay and width is here)
/ppcboot/common/usr/local/mipp/online/config/ckov_config.xml
Gatelen = gate delay (ns)
DelayWid = gate width (ns)
Run Control FAQ
trigger prescalar is n+1, i.e. 0 = scale of 1, i.e., none
1 = scale of 2 i.e., every other trigger
beam trigger fails to work if nothing running on richppc (richppc is then never ready)
Ckov DAQ takes microseconds 10^3.1
e907 daq setup required to run david's analysis programs (anamipp...)
cd /home/lange6/mippSoft/040402
setenv code_907 /afs/fnal.gov/files/code/e907
source $code_907/mippsoft/bin/setup_linux.csh
/ppcboot/common/usr/local/mipp/online/config/ckov_config.xml
took beam data with various gate delays
width 50ns
run delay events
5068 0 108?
5069 50 122
5070 100 118
5071 150 123
5072 200 22
5075 200
5076 200
5077 250
5079 300
5080 350
5081 400
"fixed" DAC BiRa 5408 from prep don't work at all
neither addressable from CAMAC (red light should come on)
one module puts out 0V by default, the other puts out -2V by default
left -2V DAC installed (Crate 4, slot 15, to drive thresholds)
Fixed trigger by magic hands touching logic module Lecroy 2365, slot 9 of
branch 1, crate 1 (Rack 3)
This module doesn't like being all the way in tight. use testLecroy3265 1 1 9
If you get error syncing it isn't working.
hv crates not responding after reboot of e907daq
minicom couldn't contact either
reset serial port (button on front) there is no power switch
minicom can now access hv but not hvc or hv_control
restore proper settings on the device files:
# stty -F /dev/ttyQ01e6 1:0:8bd:0:3:1c:7f:15:4:5:1:0:11:13:1a:0:12:f:17:16:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0
# stty -F /dev/ttyQ01ec ...
now it works
returned -10 to 10V Bira 5408 DAC to prep, asked them to check ability to read from CAMAC
Notes from April 14-17, 2004 trip (Lange)
o For next trip (Dougs) Items to check during access:
o spider box to tdc map
o output of 5408 module. Still -2.0?
o Are all threshold lemos plugged in at detector end?
o Lam problems in crate 3:
o 1000 event pulser run with ~20Hz rate should show
that some times we don't get all lams. You can see this
in the lower right plot on rc-gui if there are 4 lumps
in the ckov readout time. If not, check the log file as
well.
o Swap cables for adcs. Does problem with lams follow cable?
o Swap 2323as. Does problem with lams follow 2323? (I doubt it)
o Take the terminated adc in slot 5 out of the chain (change
the config file and unplug the ribbon cable for the gate. If
this solves in then unterminate that adc. I don't quite remember
how, likely some resistors in the upper part of the board (inside..)
There is no terminated adc in crate 4.
o Double check that hex bolts on side of ckov are tight. Look
for gas leaks..
o Items to follow up on without access
o 5408 status at prep? Do we need a different solution? Eg, Power supply
in counting house?
Progress:
Fixed readout and reconstruction code for the TDCs. Any
data before run 5222 is junk for the TDC readout. Had truncated
the # of words per tdc at 32. The data is there, but it will
take some kludging to process the raw bits. Starting from 5222
the data can be read out in a sensible way.
Found that the 5408 was reading ~0. Power cycling the
crate returned it to -2V as Doug saw originally.. Verified
that I can not talk to it via camac commands. Prep has
not verified the functionality of the other 5408. Argh.
o Further fixes to adc and tdc readout code. Added map
of mirror #, adc channel, spider box, tdc to database
and now use map in Raw2Root code. Map can be found
at
http://cdcvs0.fnal.gov/cgi-bin/public-cvs/cvsweb-public.cgi/e907/mipp/offline/CkovReco/maps/
note that it has not been checked against the cabling
scheme.
o Have started a CkovReco package for my histo module.
Andre updated several things just after I stopped working,
so need to resync to stay up to date
However, the code I have in ~lange6/mippSoft/040402
is working. The recipe is unchanged with one exception,
run anamipp with a different xml file as input
raw2root -o -n 10000 /data/1/.raw
anamipp -i -x CkovReco/xml/CkovReadout.xml -o -n 100000
o Nice signals in a few ADCs and a few TDCs -- channels
tend to be next to the beam plane, as one would expect.
o Some camac problems evident:
1. Crate 3, slot 3 adc showed strange pedestal
behaviour - a peak with a big low side tail.
Swapped the adc seems to have fixed this. I Failed
to get the swapped out adc to prep. I'll ask someone
to label+ and exchange it for me.
2. Crate 3 adcs do not always see LAM. TDCs do not have
this problem. I've tried both ECL outputs of the
2323 in crate 3, with no noticable change. Could be
the cable I guess.
3. Had strange problem during access after replacing
adc, where adcs never got lams. This was due to a
bent pin on one of the adcs. Presumably I did this
while replugging the gate cable without the benefit
of a light near our crates. Anyway, this led me to
break and the kludge the connector on the adc gate
cable, so it should be replaced regardless.
4. No word from prep on the 5408 fix. Found the 5408
Doug had left in crate 4 to be at ~0V. Power
cycling the crate returned the value to -2V, has
he had found. Presumably this means we don't really know
the threshold voltage right now.
o Heat. Its too darn hot around our crate. Its noticably
hotter to the touch than others. As I suggested a few trips
back, a fan is now in place next to JGG. Should help, but
its not clear that its enough. AC is also now on in the
hall - though it appears to be in the very upstream end
of the hall.
o Gas. Looked for leaks again. Didn't find any. There is a
small pipe fitting on the beam-left side that does not so
great, but didn't seem to leak. Peter suggested checking the
back window itself for leaks (as opposed to edges), this is
hard to get to now. Maybe some length of polyflo would
help reach of the window. We should check the hex-bolts on
the mirror access panels on the side of the ckov. The leak
checker seemed to act up around them, but not in a convincing
way. You will not find the needed hex wrench in MC7, so if
in a controled access, be prepared (as I was not...)
o Data: Took test runs with a 100ns and a 300ns window
with a 50ns delay. Also have some short runs with a
100ns (check, maybe 30...) window, with various delays.
Not much data (700 events..). need to redo with ~1 shift
of data to make much sense of them.
o During last access, switched crate 3 2323 to channel 1. The
ECL for this channel is NOT delayed. Want to check that 50ns
delay (the minimum possible) is ok. If not, we can just go
with the undelayed 2323 signals (will need to change back
jumpers on the 2323 in crate 4).
o Need to find out what a bit on the tdc corresponds to in time,
that would help establish our timing. Also, do we need to delay
the gates for the tdc to allow enough dealy between the signals
and the common stop gate. Couldn't find the recommended minimum
value for the difference between these, though it is aparanetly
200ns for a differnent tdc (for the DCs).
o Discussion with Peter indicates that we may never get much
data in the outer mirrors of the ckov. LEDs are likely
important to convince ourselves that these are working.
Notes from April 20-23, 2004 trip (Wright)
Visited prep, BiRa 5408 repair was not in the priority que as it should,
talked with Tim, will be fixed tomorrow
Confirmed that threshold lemos are plugged into all Miniboxes
Burndy 30P NE, need female connector for LED box
Got Lange's raw2root to work for me, but have problem with database running anamip
Got 5408 working in slot 21 of Crate 4. Slots 15-17 do not work with the two 5408's recently
returned from prep. This appears to be a problem with the crate. Can definitley feel a difference
when plugging in the 5408 modules. Much more solid fit in slot 21.
5408 CAMAC commands
Set voltage: F16 Channel(0-7) Value(0-4095) 0=-10V, 2048=0V, 4095=10V
Read voltage: F0 Channel(0-7)
LED bias voltages (measured with DVM in diode mode)
note long lead is +, short lead is gnd
green 1.770
yellow 1.705
red 1.529
ckov led 1.75 - 1.76
Burndy connector lables pins A-Za-z, skips I,O,Q
+ -(gnd) +-(gnd)
AB NP
CD RS
EF TU
GH VW
JK XY
LM Za
Installed linear fanout in NIM crate immediately downstream of RR13
(labelled module CKOV LED)
Set threshold on 5408 to -2V (1,2,3,4) 10am 4-23-04
/ppcboot/common/usr/local/mipp/online/config/ckov_config.xml
changed DAC from slot 16 to 21
Notes from April 30-May 4, 2004 trip (Lange)
Visited Bob Jones: led box is in line behind tpc and rich
work. Also, Bob had a question on the design, which
I answered incorrectly. Doug followed up to fix
the misinformation and answer other questions.
Exchanged non-functioning 4300 with prep.
Talked to Terry about the c4f10 status. Aparently company
send c3f8 and is figuring out how to remedy the situation.
No estimate for time yet. Not yet urgent from our
standpoint.
Checked ckov log files. See that no longer getting any lams
from crate 3. run 5440 seems ok, 5481 is not (problem on apr 21 or 22)
The daq is back as of 4pm Saturday. Problems with missing lams
gone - at least with pulser trigger.
Found that the SP2 kludge cable was a) plugged into the tdc upside
down, and b) not flipped. Result was that the channels connected
to sp2 had no tdc hits (with one exception). As of run 5619, this
is fixed.
Raised ckov hv to 2100 volts for tonights running. Will need
to email Andrew about increasing the overvoltage point. Andrew
since reminded me that this is controllable in setup
files in ~daq/hv_control/voltage_limits.hv. Ckov limits now
30 (over) and 50 (critical).
TDC default resolution is 0.5ns (set in SetCommonStop method),
so peak at 400 counts = digital signals arrive 200 ns before
trigger (+2323). Hum - this might indicate that even 50ns
of delay is too much for adcs. Need to look at adc data (crate
3 has no delay in adc gate, crate 4 has 50 ns delay).
PH changed our trigger to come from a different source (next
to Rosie) - looking at data with this trigger, we had no
signals in our detector. Access today, I checked the relative
timing of these signals. Found that the 'new' trigger comes
500ns (!!!) later than the 'old' one. That would explain it.
Notes from June 1-3, 2004 trip (Wright)
Picked up LED fanout box, tested ok
but discovered polarity is wrong for -2V supply!
Linear fanouts only work negative.
Using positive output of 1340 pulse generator to
drive all bottom LEDS (put test LED in connector for top
to see light output) can see increased output on analog
output of PMT 81 at 1800V.
Pulse settings delay ~ 1.5microsec, width = 1microsec
amplitude is about 2V,
NOTE: amplitude has two knobs (red and white lined)
red controls the positive output, white controls the negative output!
Performed surgery on pulser box, swapped purple and green wires
going to/from resistors, now can use -2V, but now switch won't work
since the Vin lines are now all tied together.
1700V PMT 81 (bottom, way in the corner).
NIM Linear Fanout will now drive LEDs, in order to trigger on LED,
must use output of second fanout of quad. If you try to use same
fanout, can not trigger on the scope (everything is set to 50ohm input
impedence).
If you are driving only bottom, then get light with input voltage
starting at -1.8V, the maximum output seems to be at -5V (remember
fanout will limit the voltage). If you are driving both top and
bottom, no light output below -2.5V.
Note Pulse width must be greater than 1 microseconds or you get no
analog output. Can run frequency as low as you want (even 1Hz or less).
First light arrival is 1.5 microseconds after start of pulse.
Victory! bottom 1800 V
3000 ns gate 0 delay
pulser 100hz 2.5 micro width, amplitude -2.25 v on scope (3 on dial) inside
-2.0V on scope in counting ouse (-2.25V in MC7) (3.5 on dial) counting house
Found DAQ problem in crate 3. ADCs would not produce LAM. Crate 4 ok.
Turns out problem is caused by too many trigger modules, crate can't
put out enough power.
LED box design (pdf)
Notes from July 13-16, 2004 trip (Wright)
Crate 4 in strange state. Crate power was on but crate controller
and branch highway terminator (with the lights) were completely
out (no lights on). Cycling the power restored to normal.
DCH Trigger modules were still stuffed into crate 3 of our system.
Andrew Norman and Ken Nelson are working on installing new crate
immediately upstream (small black minirack). After brief discussion
with Andrew, I left all trigger stuff cabled, but pulled modules
so that not drawing power.
DCH using our NIM to ECL Camac module. We use it for timing TDCs. They
replaced the cable so that it drives both our TDCs and their
MALUs. Could not find original, cut DCH tail (there is enough room for
them to crimp on new connector).
Tested ADC modules (4300) in Crate 3 OK
used: i; w 3 3 255
w 3 4 255
w 3 5 255
Plugged in Disc. modules (4415A) in Crate3
(modules were in place, but not plugged in.)
Tested ADC modules (4300) in Crate 3 OK
Tested ADCs using Gate/Delay OK
used: pulse # generates 10hz using 2323 module
4300 1 3 3 10
4300 1 3 4 10
4300 1 3 5 10
4300 1 4 3 10
4300 1 4 4 10
4300 1 4 5 10
Someone removed the power strip next to crates 3 and 4. Cable tied new
power strip to rack, replaced long extension cord with small one,
cable tied in place.
Analog 350MHz scope was bad, weird artifacts on traces. Labelled scope
as bad, replaced with 300MHz analog scope.
Rewired NIM linear fan-out that drives the LEDs to make more obvious and
provide more output lines. Output of pulsers (one in worm and one in
counting house) now go to the input of the bottom fan-out. The output of
the bottom fan-out now drives the top two fan-outs. The fanouts need
terminators on three of the four outputs (see manual) and at least
two of the outputs must be terminated. Put two terminators
on all channels.
Andrew's hv_control and hv_monitor now run on e907hv.fnal.gov in
counting house. Each program runs in a separate xterm, can't tell them
apart from the icons, so I copied my 'name' script to /homehv/daq/bin
and named each xterm (monitor and control).
Found bug in hv_monitor, can't go back to CKOV display because there
is an empty card between last card and next system. Asked Andrew to
fix this. Also requested he implement the HV channel to PMT map.
Serial ports for HV controller are now on e907hv and not e907daq. Must
run hv_control on e907hv.
Andrew et al removed trigger cards from crate 3, now all hell has
broken loose. Neither crate will work. Crate 3 bottom branch highway
cable was a little loose, now crate 3 works.
Crate 4 has a variety of bizarre problems. Access light (lights up when
card is being addressed) is latched on for the 5408 and one of the
TDCs. This is an error, should only happen when DAQ addresses the card.
The ADCs no longer respond correctly. If you write a value, the read
back is incorrect. For example, w 4 3 255, reads back 169. The 2323A also
not working properly. Pushing "CHAN" toggle causes display to go "crazy".
Trying to pulse the 2323A with script 'pulse' causes the output to latch
on at high rate. Problem goes away if you plug module into another crate.
If I unplug all modules on crate 4 except crate controller, terminator
and one ADC, the ADC works. If add another ADC or any other card,
the ADC no longer work. Can read them, but get wrong value.
If 5408 is only card in crate (not counting controllers) the access light
is out, add any additional card and the access light on the 5408
latches on.
Note that crate 3 nominally works, but reading back ADCs gives the
wrong answer about 1 in 5 tries. Crate 3 card 3 seems to be the worst.
So, replaced crate 4 with new CAMAC crate. Talked to Mike Benke while
at prep, Mike suggested that high temp and high humidity could be
cause of problems.
With new 2323A (from prep) and original three ADC modules and single
5408. Crate 4 is working so far.
But... crate 3 has problems now! ADC modules give incorrect
readbacks about once in 5 tries. 2323A goes crazy as described in
crate 4 story above. Spare 2323A works fine in crate 4, but
does not pulse correctly (via pulse script) in crate 3.
So, removed first ADC module from crate 3 and put in crate 4. This
module works fine. Unplugged all modules from crate 3. Single
ADC in slot 4 showed periodic readback problems. Wrote script "darn"
to repeatedly write to a single card. Trying script now shows
that card 4 is fine! Try by hand again and it is still fine.
Now added card in slot 5, ok. Now put back card into slot 3,
still fine. Hurray!
Crate 3, put 2323A back in. Now it is fine, not crazy and pulses
at correct rate. It is noticable colder now in the worm (it is 6pm)
than it was earlier today, perhaps this is why things are working
again.
Wrote script 'testadc' to run pulser and readout all ADCs on both
crates. Both crates are ok. Note that only 2323A, ADCs and 5408
are plugged in. Asked David to change test4300 so that min, max, avg
outputs are in columns. Found bug in test4300, always got first
trigger even without real trigger, probably due to test command
in program, David fixed this.
4300all [trig] [file] command will take triggers from all adcs
and put results in file, each column is output of one channel.
Reinstalled LED pulser unit in counting house. Somebody removed it
to install a bunch of scalers.
Cabled both (crate 3 and 4) 2323A so that can get triggers from
LED pulser and from official trigger, inputs will drive both 2323s.
Tested both LED pulsers (in worm and in counting house) to drive
2323, ok.
Crate 3, 2323 trigger threshold set (using trim pot on front panel)
Chan A threshold -0.300V
B -0.303V
Trigger toggles should be set to negative(-) slope
How to program the 2323A:
2323 1 crate slot chan delay gate 0 0
chan=0,1 = A,B
NIM output triggers at same time as START and
has width is equal to progammed delay
DLY output is delayed by programmed amount with width
10, 30, 100, 300ns
ECL output is either NIM or DLY, depending on internal jumpers
Our modules should be set to NIM. Verified this for crate 3.
The reason for this (according to David) is that the minimum DLY is
about 50ns and the experiment trigger comes so late, that we can't
afford this much delay!
Found bug in test2323, always resets chanA and chanB, so can't program
them separately. Fixed by David.
Recables gates so that LED goes to chanA on crate4, DLY output triggers
chanB which is daisyed to chanB crate 3. The experiment trigger is also
in crate 3 chanB. So can now program crate 4 to generate pulse
with arbitrary delay (chanA) with arbitrary width (chanB). Wrote
scripte gate to set crate 4 chanA and chanB. Verified on scope that
this works.
Wrote test_dac script to test 5408 (slot 20 in crate 4). Cables threshold
lines. Test ok.
Installed NIM to ECL (to drive TDCs) and TDCs is crates 3 and 4:
NIMtoECL slot 7
TDCs slots 9,11,13
Wrote test_tdc to read them out. All ok (both crates).
Note that Discrimantors and scalars are not plugged in.
Cabled ADCs and TDCs is both crates. Wrote test_all script to test
ADCs, TDCs, and DAC. All ok!
Updated ckov_config.xml (/ppcboot/common/usr/local/mipp/online/config)
changed TDC slots to 9,11,13 and DAC to slot 20.
11:40am Although it is still cool on an absolute scale, it is starting
to get noticeable warmer by our racks.
1:00pm A bit warmer. Getting blasted with external trigger, this interferes
with test_all pulser. Unplug external trigger, everything ok. Can tell
if getting blasted because ADCs readback very quickly before hanging up.
Andrew is running a pulser on the experimental trigger. This is automatically
fanned out to everyone! I asked for trigger redesign, so that can
program (or disable) each detector trigger via computer.
Trigger guys insist on getting some data today. Ended access around
2pm? and waiting around MCC to generate beam. They have lots of
problems, so finally found someone to do controlled access with me.
Timing tests. Put digital scope by LED pulser. With three equal length
cables, connect scope (chan1) to common input to 2323 Crate 4 ChanA,
(chan2) to NIM output, (chan3) and to analog output of PMT 81 (must
unplug from delay board).
Set pulser to 30 Hz. Using chan1, set pulser output voltage to 3V
(beware, analog scope was not calibrated properly and read only 2V, I
believe the digital scope because it correctly shows -1V on NIM
output). Also using chan1 set pulser width to 1.5 microseconds. Set
PMT to 2000V. Will now see lots of PMT pulses (chan3) begining about
1.25 microseconds after beginning of chan1. The main pulse train is about
1-1.5 microsec. long. There are large single peaks at around 1microsec.
Can now use 'gate' command to move ADG gate. Note that the delay
boards introduce an additional delay of 300ns.
Can look at signal before and after delay. Put tee in PMT 81 on
input to delay board. Note that there is an additional lemo socket
on the outside of the delay board, this is tied to the delay output.
Connect (chan3) of scope to the tee and (chan4) to the back of the
delay board. Both should be 50 Ohm DC coupled. Chan3 set to 10mv and
Chan 4 to 5mv. With this setup, can trigger on big spikes on chan3 and
can measure delay on chan4, verified that delay is 300 nsec.
Pedestal scans
--------------
With HV off, check pedestals for various gate widths. Some pedestals
increase and some decrease! as width goes from 100 to 3000 nsec.
HV ADC
slot crate slot
1 4 5
2 4 4
3 4 3
4 3 5
5 3 4
6 3 3
Pulser in counting house. Amplitude set to 6 (on the dial).
Duration set to 1.5 microsec (on the dial).
Frequency set to 400 hz (on the dial).
Trigger for 100 events.
Did delay scan for various widths. Good setting is gate 900 1200 (delay width).
Friday, July 16, 11am. Very warm in worm. Closed all night and had
beam. Crate 3 and 4 now have severe problems: test_all fails because of crate3.
2323 showing pulser failure mode. The display at top is flashing
(wrong behavior) and the output frequently gets stuck on at high rate
(should only be 10hz). Note that 2323 and ADCs are extremely hot
(feel at top of rack). Other modules do not feel very warm.
ADC in slot 3 on both crate 3 & 4 shows read-back problem (using
'darn' do not get full counts, i.e. 255). The other ADCs are ok, a ha,
perhaps this is the cause of the failure! Moved ADCs down one slot, to
leave gap between ADC and 2323. Note I moved 5 to 6 and 3 to 5,
leaving 4 as it is. Problem with ADC readout moved with the card,
i.e. now readout problem is with slot 5. Leave this for now and move on.
2323 is not bad, must be crate 3 too hot, because moving the module
to crate 4, it works fine, moving it back to crate 3, it has problems.
This must be the bad connection to crate bus that Mike Benke talked
about. So, replaced crate 3 with new crate BiRa 6700. 2323 now
works fine.
ADC modules in new crate have read-back problem, i.e. writing and
reading register does not result in same number. It looks all but the
lowest two or three bits not working. Tried four different ADC modules
in many different slots. They work in crate 4, but not crate 3. It
must be the crate controller! Replaced Jorway 70A crontoller with
spare, and presto the ADCs work! It looks like the old crate controller
had problems with the high bits on the dataway.
Put in all ADCs and TDCs in crate 3. All modules read back correctly.
All modules getting proper triggers from 2323. Did pedestal measurements
on ADCs. Note that two modules giving abnormally low pedestals (0-2
instead of 20-40). Problem appears in middle ADC of crate 4 and
last ADC of crate 3.
Turn on HV and do LED pulser runs. Can see signal on all ADC modules,
except the two above that had pedestal problems! Always reading 0-2
counts. Replaced modules with spares (note that I had to move the terminating
resistors from the existing module to the spare). Now getting reasonable
pedestals and seeing signal on all ADCs.
It is still hot in the worm, but everything is working!
Edited ckov_config.xml to match new slot scheme.
Set thresholds on both 2323 to -0.2V. Note that ChanA of Crate 4 must
have terminator on unused input, primary input comes from linear fanout
of led pulser signal. Both 2323 ChanB ECL output should be jumpered to DLY.
Verified that pulsers inside worm and in counting house work (getting
triggers).
Time to catch my flight. Let's hope nobody touches this after I go.
Notes from Sep, 2004 (Wright from LLNL)
9-14-04
DAQ people unnecessarily changed CKOV crate addresses from 3,4 to 5,6.
This causes CKOV DAQ and all test scripts to break (total of about 20
scripts and files between Lange, Wright, and official DAQ).
Prevention: DAQ people should have used 5,6 for new non-CKOV crates,
thus CKOV would not need any changes.
DAQ people unnecessarily removed execCAMAC command, replaced it with
two different commands (exec16, exec24). This breaks subset of CKOV
test scripts. Also functionality of commands is now different. Could
not use trial-and-error to figure out which command to use since
entire racks were left off after power failure (no crates responding).
According to Andre (via phone) exec16 is equivalent to old execCAMAC.
Prevention: Should have replaced old execCAMAC version with link or
some other backward compatible fix (to use exec16) and added
execCAMAC24, thus CKOV would not need any changes.
Power to racks 12 and 13 was not restored after shutdown. Consequently
can not read out CKOV crates. Had to debug system (since software
changed), once discoverrf that system really isn't there, called
control room and ask someone to turn on power.
Prevention: DAQ and/or run manager should have had plan to restore
system racks to state prior to power shutdown. Would have saved CKOV
debugging time and random onsite person to fix problem.
'test_dac' now works as advertised. Note that this is in crate 6
(formerly known as 4)
> test_dac
setting to -0.5 volts
Read back -0.50061
Read back -0.50061
Read back -0.50061
Read back -0.50061
setting to 0 volts
Read back -0.002442
Read back -0.002442
Read back -0.002442
Read back -0.002442
Wrote test_there script to write and read back data ADC and TDC registers.
ADC and TDCs respond:
> test_there
writing 255
ADCs crate 5, slots 4,6,8
Data = 255 = 0xff, Q = 1, X = 1, timeout = 0
Data = 255 = 0xff, Q = 1, X = 1, timeout = 0
Data = 255 = 0xff, Q = 1, X = 1, timeout = 0
ADCs crate 6, slots 4,6,8
Data = 255 = 0xff, Q = 1, X = 1, timeout = 0
Data = 255 = 0xff, Q = 1, X = 1, timeout = 0
Data = 255 = 0xff, Q = 1, X = 1, timeout = 0
TDCs crate 5, slots 12,14,16
Data = 255 = 0xff, Q = 0, X = 1, timeout = 0
Data = 255 = 0xff, Q = 0, X = 1, timeout = 0
Data = 255 = 0xff, Q = 0, X = 1, timeout = 0
TDCs crate 6, slots 12,14,16
Data = 255 = 0xff, Q = 0, X = 1, timeout = 0
Data = 255 = 0xff, Q = 0, X = 1, timeout = 0
Data = 255 = 0xff, Q = 0, X = 1, timeout = 0
Notes from September 21-22, 2004 (Wright)
Debug crate 6 because not working
Found bad 70A, replaced with good from Prep
(two spares from hall were bad, returned these to prep.)
Test sequence to see that modules are there
test_dac
test_there 255
test_there 111
test_there 0
OK all modules present and accounted for.
But, ADC slot 6, crate 6, is flakey, read back test doesn't match data 111
do get triggers and reasonable pedestals
Replaced with spare marked possibly bad, read back test ok, but pedestal
values now nearly all zero. Put original ADC module back in.
Keep an eye on this.
Edited pulse, pulse_camac to work with new crate IDs.
test_adc 6 ok, get triggers and see pedestals
test_adc 5 fails, get triggers without pulser!
test_tdc 6 ok, get triggers and see data
test_tdc 5 fails, get triggers without pulers, all null data!
crate 5 controller was bad too! Probably when our colleagues
changed our crate id's from 3,4 to 5,6 they just took our good
controllers and gave us crap ones!
Trigger input to 2323 crate 5 needed 50ohm terminator.
Exp. trigger was on, disabled output register from DAQ
(required run start to actually download).
Now all triggers (pulser, LED, experiment) are working again!
Rewrote scripts test_dac, test_adc, test_tdc, test_there, test_all
to add features.
Wrote test_led script. Use with LED pulser to see signals on ADCs.
Results, using LED pulser in worm:
delay= 900 width= 1200 ns
set HV to 1900V (N.B. wait at least minute for bases to warm up)
set pulser:
frequency = 400 Hz
duration = 2 microseconds
amplitude = 6V (white, 0V red) if in counting house
5V (white, 0V red) if in worm
offset = 0v (white and red)
negative pulse output to LEDs
Check that ADCs get triggers: You should see six lines that end with 'Average'
110 59 357 109 58 32 1210 96 28 52 605 54 219 52 737 52 Average
67 35 217 82 1019 50 2047 69 273 69 1135 70 480 88 370 24 Average
2047 49 293 135 92 62 2047 180 148 110 150 106 68 85 0 89 Average
109 56 4 31 26 45 180 46 111 51 173 43 202 26 562 5 Average
1623 28 649 67 786 83 74 851 72 59 334 75 355 71 559 26 Average
32 42 136 41 84 88 116 0 68 115 180 26 23 602 37 27 Average
Tested spare 4300 ADC boards (3 of them) with LED pulser, none of them
worked, will return to prep.
Notes from October, 2004 (Wright)
.login script on beamppc disappeared:
source /usr/local/mipp/setup-mipp-online.csh
Found replacement by looking at ~alebedev/.login:
source /usr/local/mipp/bin/setup_ppc.csh
Switch to srt means that scripts are broken again.
Found exec16 and exec24 in
/usr/local/mip/releases/OLR-04.09.27/bin/Linux2.2-GCC/exec16
For now just copy them to ~/bin. I never did like srt. BaBar
was cursed with this and now MIPP.
initCAMAC routine is gone completely, found it in /usr/local/mipp/attic/
copied it and necessary libraries to ~/lib. Modified LD_LIBRARY_PATH
(in .login) so this will work.
To use new srt system (to get development executables in your path)
the user must type 'srt_setup', so I put this in my .login on beamppc.
beamppc:test/test_all still works, but shows that two adc modules on crate 6
are now questionable:
crate=6 slot=4 0 = 111 255
crate=6 slot=6 0 = 47 63
crate=6 slot=8 0 = 109 253
should be:
crate=6 slot=x 0 = 111 255
before, only slot 6 was bad. Note that pedestals look ok.
Andre replied that 'crateInhibit 0' should be used instead of initCAMAC.
Have to fix all scripts.
Got beamppc login to work without passwords (needed to use ssh -2 protocal:
cat id_rsa.pub | ssh remote 'cat - >> .ssh/authorized_keys')
David report on gas
3 months of gas arrived at Fermilab (500 ft^3)
cost ~$16k
Terry reduced overpressure on Ckov, which resulted in reduced gas usage
From Andrew Norman:
Just to repeat the info, since I am assuming you didn't get it, the
Jorway41 is in Crate 4, Slot 21. To dump the register use an F0 A0.
The register is 12 bits wide.
Be aware that while I did run a separate trigger line for you that can
be controlled from this register, I didn't remove or disable your old
trigger line. The CKOV should still be controlled of that original
trigger line unless you modified it. I can go out and see if that is
still the case if you want, but I can not imagine anyone would have
changed the cabling without talking to you.
The current bit assignments are:
Bit 1 -- TPC 1 (dsp trig)
Bit 2 -- TPC 2 (gating grid)
Bit 3 -- General to Rack 5 aux
Bit 4 -- CKoV
In addition bits 11 and 12 are currently open, the rest have other
detectors being fed off them (Let me know if you need these other
assignments)
The cables I ran for CKoV are labeled as "Exp Trig -> out 5" and "Exp Trig
Spare 1". The drops for both are right next to the lines you had
pulled. Both lines lead to the patch panel on 11b. If you have
problems with either line let me know.
I have no idea how the old cables are labeled, or even if they are. When
I was redoing the trigger down at 11b, I just figured it was better run
new (tested) good cables than to worry about tracing the old ones.
You are right the information PREP has on that Jorway 41 is pretty slim.
Basically they only listed the F16+ user commands. I don't have an
official manual from Jorway, just some old notes that mention "F0 A0 read
prog reg", your milage may vary.
scan_ped script modified to take pedestals from all ADCs (takes pedestals
for various widths using internal pulser). Note that some channels go
down (to zero), some go up, and some more or less remain the same. 4300
manual says that for gates above 500ns need to adjust potentiometer to get
non-zero pedestal.
e907hv: Can not talk to hv (minicom, hv_control, hvc all fail).
Device missing: /dev/ttyQ01e6
Tried reboot, did not resurrect device file.
/etc/rc.d/init.d/espx restart
Run this as root and you see it fails to find /usr/local/espx, which
is exported from e907ana1:.
Unmounted /usr/local and found espx exists in native directory. Commented
out the mount in /etc/fstab and rebooted. HV now doesn't complain about
/dev but still no control. Equinox box was reset, still didn't work.
Ran hv_control and everything ok, probably the old feature of changing
the port settings on the /dev file (see tty below). Turns out
that Andre is the one who made the error in the /etc/fstab file.
Reset all voltages to zero (z command in hvc), turned hv on.
10-12-04
Did pedestal scan with HV at 1000 and 1700v. No difference in pedestals
compared to HV off (all data in data.oct04 subdirectory). Wrote
perl script scan_ped.diff to compare the two files and look for differences.
10-13-04
Modified scan_ped to use pulser or external LED trigger. Did pedestal
scan with LEDs. Find difference between LED and pulser pedestals
for one channel in each ADC in crate 6. No such difference for crate 5 ADC.
Also checked that pulser pedestals did not change from previous day and
that pulser frequency does not matter. Checked that repeating LED pedestals
results in same values, only difference is LED compared to pulser in
crate 6!
Modified scan_ped to scan delay. No changes observed for pedestals vs delay.
(width=1000ns).
10-14-04
Wrote led_adc and ped_sub scripts to take ADC and do pedestal
subtraction with LED pulser. Did HV scan and wrote led.split perl
script to put output together by ADC module. Noticed strange
behavior for crate 5 adc 4, which is HV slot 6. At 1500V if
HV set to zero, then ADC output doesn't seem to go down. If set to
1000V, adc goes to zero. Perhaps the HV module is not really shutting
off (read back says it is). Also, sometimes when set to 1500V it
appears that there are no signals. This only happens for crate 5 adc 4.
See problem with ADC channels. Seems like every other channel is
severly attenuated. Need to check each channel with same input signal
(will do at next trip).
10-17-04
Tested new adc features from David. Wrote ped_get script for generating
pedestals and creating ped_set script that can download them to adcs.
Wrote read_adc_ped for reading pedestal subtracted adcs (i.e. 'noinit').
10-17-04
Note that can not enable and set pedestals subtractin in 4300 when there
are gets coming in! Must stop software and led pulser first.
Wrote 'ped_dump' script for communicating with pedestal features of all 6 adcs.
HV slot 6 @ 1500 V with LED pulser puts signals on analog line. PMT#7
has nice large signal ~500 counts above pedestal through delay board
(gate 900 1200), and 1050 straight into ADC (using LLNL
pigtail).
Tested 3 spare 4300 boards all bad. Two read out pedestals <10 on all channels
and do not register signal. Third module will not even trigger. Tested in
crate 5 slot 7. Checked that the slot and cable work ok by moving ADC from
crate 6 slot 4 to this slot. Worked ok, so put adc back into crate 6.
Checked all channels on crate 5 slot 4 ADC with Analog #7 get 1025-1070 counts.
All channels functioning ok.
Crate 5 slot 4,6,8 all channels ok and pedestals all above zero for
gate 900 1200.
exec16 1 4 21 20 0 2048 # turn on channel 12 (2^n-1, 2^12-1 = 2048)
22 0 2048 # turn off channel 2
Implemented LED pulser remote control. Use DAC, connected inside pulser
to line 7. Set to 0v for off, 6V for on. Wrote script 'led' to turn
this on/off.
Installed and labelled permanent cables for pulser gate for LED pulser
1 (control room) and pulser 2 (worm). Pulser gate is channel 7 of DAC
in crate 6, slot 20. This gate controls both pulser outputs (Note:
should only power on one pulser at a time).
Exchanged modules at prep. Two Jorway 70 camac crate controllers
and three lecroy 4300B ADC modules.
Complained to (Mike Behnke, behnke@fnal.gov) prep about "fermilab
only" web pages for 4300 adc. Turns out that the addresses on the page
were broken.
Swapped out ADC crate6,slot6 with new one from prep. Bad read/write
problem gone, pedestals look good. Marked old module as bad.
Checked all channels of ADCs crate6, slot4,6,8. Put signal from PMT#7
directly into ADC. Everything ok.
Tested other two 4300 modules from prep. Both are ok. One of the
units already has terminators on it. Only checked input to a few
channels (used crate6, slot 7).
Some channels have pedestals at zero for gate width of 1200ns. Decided
not to adjust trim pot (potentiometer "O", must remove module and
module cover to change) because we will probably not have such a long
gate for the real experiment anyway.
Checked that pedestals >10 for 500ns width (gate 900 500):
crate=5
38 43 48 61 36 33 39 49 29 44 58 42 37 37 27 42 Average
42 35 46 49 38 44 41 60 45 45 41 49 43 56 50 34 Average
76 51 83 62 77 58 75 60 75 72 81 81 62 74 21 56 Average
crate=6
45 41 36 31 29 35 28 30 48 37 28 34 33 28 39 17 Average
46 32 43 35 39 43 33 33 31 29 36 34 39 36 41 26 Average
37 35 60 30 50 29 40 38 20 37 18 26 19 39 29 27 Average
Checked delay board channels 49,51,53,55 and 52,54,56,58. Removed
analog cable from back side of board, fed in good channel PMT#7. See
proper ADC counts on all channels, so delay board and ADCs should work
as advertised.
Checked analog cables from detector patch panel 49-64 (bottom). From
underneath detector fed good channel (PMT#63 @ 1900V) to each analog
cable, proper ADC counts observed on all channels.
See odd/even pattern. Odd channels all have high counts (>100) with
the exception of #53. Found HV not locked in place on HV module, now
all odd channels look good.
Even channels barely above pedestal counts ~10. Start debugging #50.
o Put analog signal onto subminex #50, get observed proper ADC
counts, so base to ADC cabling ok.
o Swapped HV with #51, HV tests ok for both #50 and #51.
o Checked low voltage on subminex, get -5.2V ok.
o Wiggled base a bit find that can change output from 2 (essentially dead)
to 10 ADC counts (barely on).
o Base not level with shield because someone used long and short screw to
attach to detector. Replaced short screw with long screw.
Replaced cable anchor that fell off during debugging.
Debugging channels 65-80. Analog connections swapped at base between
71 and 72. Swapped back, now ok.
Verified that HV 73 cable is bad, but it also appears that the base 73
is bad too (subminix analog and low voltage cable are good).
Bottom Half
View from lying underneath detector
Left Beam Right
|
|
V
-----------+--------------+--------------+-----------+
81 84 | hvslot=3 | hvslot=2 | 91 94 |
| 49 53 57 61 | 65 69 73 77 | |
82 85 | 50 54 58 62 | 66 70 74 78 | 92 95 |
| 51 55 59 63 | 67 71 75 79 | |
83 86 | 52 56 60 64 | 68 72 76 80 | 93 96 |
+--------------+--------------+ |
87 88 hvslot=1 89 90 |
-----------------------------------------------------+
10-20-04 beamppc answers pings, but login hangs. Pushed reset
button on module (abort button didn't do anything). It took at
least a minute for it to reboot. Everything ok. Ran test_all.
Replaced bad HV73 cable with spare. Labelled Bad and spare cables.
Channel 73 appears to work, but has lower than expected ADC counts.
Perhaps this is ok because channel 65 (similar one via symmetry)
has unusually high ADC counts, i.e., the LED location may be
responsible.
Did HV scan for entire detector, 1700, 1800, 1900, 2000V.
Now try TDCs. Note that this is with spider cables (which do not correspond
to ADC channel order) with flip and shift, directly into 3377.
Remember to run set_dac which sets thresholds to -1.8V. Note that
light output from LED doesn't start until about 1.5 microseconds after
beginning of pulse. gate 1400 100 seems to work the best.
Don't see nice correlation with HV channel and TDC signal. Discover
that turning on just a single HV channel (PMT=7 HVslot=6 1900V) causes
most of TDC 1 (crate 5, slot 12) to come on:
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
0 144 180 11 510 117 111 112 106 45 113
With all HV off get occasional hits on channel 9, must be noisey.
Note that TDCs are currently reading out in units of 1/2 nanosecond, so divide
the above by 1/2.
Put known test signal (generated from NIM discriminator in crate with
LED linear fanout) into TDC via CAMAC ECL-NIM-ECL (PS Model
7126). Discriminator output is 20 nsec NIM (I wanted wider signal, but
that is max module could do). Ribbon cable fro NIM-ECL to TDC. Crate
5 slot 12 all channels ok. Note that can not measure delay greater
than 500 nsec (which appears as 1000 counts). Set gate to 300 nsec:
'gate 300 100'. TDC reads 725 counts.
Now put digital output cable into camac ECL-NIM-ECL. Looking at NIM output see
signal at 2000 nsec after LED pulser (using discriminator output, see above).
See ratty signal on channel 16, see big single pulse on channel 1. Don't see
any hits in TDC. Scanned delay fro 100 to 2500 nsec. Tried with and without
cable flipper.
Went back to direct input to TDC (with flipper). Found signals with
gate set to 2300 nsec. gate 2300 100. Signals are very ratty, so get
lots of multiple hits. Just take first hit. Channel map is a bit off,
HV channel 1 comes out as channel 14 (in 0-15), channel 2 as 13.
Switched experimental trigger line. Was 'ckov adc trig',
now 'Exp Trig Out 5' (from Andrew's addressable module).
10-21-04
Fixed carriage return problem in hvc program for fast_command_send
(compile on e907daq to avoid weird complaint about makedepend on e907hv).
10-22-04
Converted default scripts to use new 4300 routines (with ped subtraction
capability). Tested and works ok.
Created tar file backup of test directory on beamppc. Put file in
my home directory on e907daq.
Switched all test programs to new versions from David (2323, 3377,
4300, 5408).
David modified 3377 to report in ns (before was reporting in units of
0.5 nsec) also removed reinitializing every time (takes about a
second! per module!). So now '3377 1 crate slot 0' will initialize in
commond stop mode, while regular read does not redo the
initialization.
Created init_tdc script to do the initialize all six cards.
Fixed ped_set script so that 4300 are initialized properly for
pedestal subtraction mode. This avoids annoying hangups.
Can use gate 2300 100 for adc and tdc. The LED pulser is set
long enough that there are adc counts as well, but note that
these counts are not in time with TDC. Should redo relative
timing study and set up double gate when at fermilab (so we
can correlate TDC with ADC).
Took adc and tdc data at 1900V (all.1900) and 2000V (all.2000).
11-1-04 (DMW @ FNL)
beamppc is not responding again. Can ping it but not log in. Pushed
restart on the front panel. This is the third time I have noticed this.
First time was 10-20-04, second time was noticed by Andre last week.
11-2-04 (DMW @ FNL)
CKov LED pulser 2 is laying on the floor (not in rack 17 were last
seen on 10-20-04). The input gate cable is broken (conudctor was pulled out).
Reinstalled pulser, replaced cable, recabelled-tied cables to rack. It looks
like someone tried to remove the pulser but didn't notice the cable ties
and yanked the cable out of the connector. I sent email to Holger et al
asking that people let us know when stuff gets broken.
test_all and read_adc (with led on) works with pulser 2 again.
11-3-04 (DMW @ FNL)
Tested TDC channels 1-16 (we don't use 17-32) on all 6 modules, one at a
time. All ok. Test using LED pulsers (with HV off). Put 32nsec delay
(via lemo cable) in trigger (between 2323 and NIM-ECL). Use NIM output
from opposite crate into NIM-ECL-NIM module in crate 5. Use ribbon
cable from N-E-N to TDC to test. Move lemo cable down N-E-N and look
for signal on TDC. Should see either 32ns or 16 nsec delay (depending
on whether you are testing crate 5 or 6 respectively)
Tested two spare 3377, labeled ok.
Added feature to hvc control program (compile in e907daq). Can now
set ODD or EVEN channels separately.
11-4-04 (DMW @ FNL)
See if LEDs illuminate the tubes on the same side (top/bottom) or opposite
side. Answer is they illuminate same side (bottom LED illuminte bottom PMTs).
Notice that light output of LED goes up if you remove top/bottom, because
total resistive load is reduced.
1900V, gate 1300 900
read_adc_ped
LED top and bottom flashing
212 11 539 14 22 1 2047 48 2 13 1299 17 326 14 1288 2 Average
49 0 317 32 1280 2 2047 1 276 14 1508 18 619 13 427 0 Average
2047 7 239 119 8 582 2047 190 100 45 91 11 12 2 11 33 Average
135 3 588 12 580 17 379 26 149 12 461 15 413 15 1137 6 Average
2047 33 996 20 1531 30 1607 20 35 3 781 2 624 21 1063 5 Average
133 3 78 114 11 122 142 36 156 215 482 6 27 825 5 13 Average
LED top flashing only (unplug bottom)
435 12 1089 21 39 1 2047 81 3 10 2047 28 670 15 2047 4 Average
80 0 569 53 2047 1 2047 6 654 22 2047 35 1354 16 924 0 Average
2047 7 357 249 14 1089 2047 394 174 103 156 18 17 2 18 47 Average
6 1 2 2 15 10 0 4 16 8 1 2 14 9 6 2 Average
25 14 3 0 12 15 3 0 1 2 0 1 10 1 1 2 Average
4 2 1 4 1 2 1 7 1 5 8 0 2 2 1 0 Average
LED top flashing only (unplug top)
20 7 1 1 1 0 4 2 0 9 12 1 17 9 4 0 Average
16 0 2 1 11 8 4 0 9 7 4 0 12 9 1 0 Average
1 0 1 11 1 0 2 2 1 0 8 1 1 0 1 0 Average
291 2 1409 23 1146 18 764 50 348 18 1181 26 825 18 2047 12 Average
2047 28 1907 9 2047 40 2047 33 63 6 1902 6 1340 23 2045 11 Average
196 6 120 253 15 211 277 91 303 482 1118 11 60 1236 5 17 Average
LED top and bottom flashing
200 11 561 13 22 1 2047 49 2 11 1310 17 329 14 1279 2 Average
46 1 293 29 1275 0 2047 0 271 17 1520 18 617 13 416 0 Average
2047 7 240 114 8 364 2047 193 101 45 108 10 10 1 11 33 Average
130 4 589 11 575 17 380 26 145 12 457 14 414 12 1148 6 Average
2047 6 1010 20 1534 29 1580 19 36 3 797 2 623 19 1084 5 Average
134 3 77 116 10 123 137 35 155 211 492 5 30 831 5 12 Average
OK, Now try to increase the amount of light seen by the odd (relative) numbered
PMTs. Turn on a single card and turn on just the odd channel HV. Tried to raise
the output voltage, but the linear fanout maxes out at pulser voltage setting of
about 5.5, was already running at 5V (pulser 2 which is in the worm). So can't
get any more amplitude.
hvslot=3 crate=6 ADC=4 PMTS=49-64 ODD=1900V Even=0V
> 4300 1 6 4 100 noinit | grep Average
0 2 1 9 1 15 1 22 1 9 1 10 1 10 1 5 Average
Increase the LED pulse width by factor of 10 to 20 microseconds width (normally is 2)
and increase the ADC gate: ped_get 1300 3000. Note that ped_get will set
the width delay and then execute ped_set.
Raise HV to 2000V for ODD, still 0 for EVEN.
> led on ; 4300 1 6 4 100 noinit | grep Average
turning LED pulser on
0 1 0 152 0 177 1 315 1 148 1 119 0 195 1 41 Average
Lots of light now, so turn on ODD channels of hvslot=2 crate=6 ADC=6:
0 0 0 152 1 176 1 330 2 150 2 121 0 195 1 39 Average
7 137 5 222 5 319 5 148 7 4 5 4 5 126 2 36 Average
Wrote hv script on e907hv:hvc to control hv from command line
Wrote plateau script to plateau a single hv slot.
Wrote plateau.all script to plateau all adc slots.
Wrote plateau.each script to plateau each channel.
From adc*.plateau files:
adc3 2047 45 871 2047 116 12 2047 2047 689 1177 1306 85 46 8 0 116
adc2 1500 25 2047 349 2047 243 2047 139 2047 233 2047 87 2047 170 2047 42
adc1 2047 183 2047 136 148 22 2047 437 2 234 2047 187 2047 187 2047 2
adc4 2047 32 2047 180 2047 193 2047 323 2047 198 2047 142 2047 228 2047 81
adc5 2047 240 2047 270 2047 332 2047 132 536 6 2047 8 2047 137 2047 41
adc6 647 19 387 2047 88 1623 2047 1711 2047 2047 2047 64 338 2047 19 74
adc3 dead? dead?dead
adc2
adc1 dead dead
adc4
adc5 dead dead
adc6 dead
It turns out there is significant cross talk between channels, especially when
the a channel saturates the adc. Refer to the individual channel plateaus (on
e907hv) to see which channels have no signal.
no signal channels (0:15)
adc1: 8, 15
adc2: 1, 15
adc3: 5, 13, 14
adc4: 1
adc5: 9, 11
adc6: 1, 15 channel 14 pmt is connected to #15 readout!
ADC to PMT map
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Top (Crate 5)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 adc1
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 adc2
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 adc3
Bottom (Crate 6)
49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 adc4
65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 adc5
81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 adc6
1st Check all top PMT that do not see light
Check adc6 pmt 14/15 swap?
Check weak channel output reduced by lowering LED voltage
adc1: 15 fixed, delay line chip was loose
adc1: 8 signal reaches delay board, fixed loose wire from delay chip
but also lemo connector is broken on user side! must resolder
adc2: 1 & 15 patch panel to adc is ok, no signal at patch panel
11-9-04
beamppc has problem with ssh again, cannot get plateau scripts on e907hv
to work anymore, beamppc hangs on ssh. Stopped DAQ (was running) and
rebooted beamppc. Worked for about 3 ssh sessions, then hung up again!
adc2: 1 analog line from base is ok
hv is ok
Low voltage power is bad, it reads -5.2V, but base doesn't work
base does work if use another low voltage cable
adc2: 11 ? redo plateau, check at constant HV because ADC counts seem to drop with time
adc3:14 hv ok, analog from patch panel ok
adc3: 10 for a while the output was much lower than last week, now ok
adc=1187 @ 1900, earlier this evening it was only 12!
adc6: 14 HV cables are labelled correctly
analog lines from patch panel to delay board are labelled corretly
subminex to patch panel cables are mislablled
change labels and connect correctly. now 14 is ok.
11-10-04
adc6: 15 HV, analog, lv ok
adc6: 1 HV, analog, lv ok
4 was swapping lv with 1 and now after putting lv back the output is
fluctuating, was consistently about 40 but now is often zero!
adc6: 7 pedestal has changed dramatically now 130 counts higher. redid
ped_get 1300 3000 and now seems ok, see light on it.
adc4: 1(50) analog, HV, lv ok
adc5: b9(74) analog ok, HV BAD
adc5: 11(76) base very wobbly, analog, hv, lv ok
ssh to beamppc is horribly slow, it effectively kills plateau scripts.
ssh -v says it is the exec after authentication
that is the problem -x doesn't help, but switching to -1 does!
modified plateau scripts and we are back in business.
adc6: 4(85) redo plateau, is about factor of two lower now, but still
see signal
adc2: 11(28) redid plateau and it looks fine (e907hv:adc_data3)
but ADC drops with time to zero!
swapped hv with another channel, hv is fine, it is the base/tube
adc3: 14(47) replaced analog line between base and patch panel, now ok
Turned off low voltage to check subminex cable, power back on and
pedestal for adc3:14(45) is now lower by 25 counts. Or maybe just
because cable was bad (see above) and I didn't notice after
changing it. Redo pedestals: ped_get 1300 3000.
adc2: 1(18) need to replace low voltage subminex
Made low voltage splitter box. Used low voltage from mirror 41, split to
41 and 18.
11-11-04
Crash during afs editing of this file. Lost all notes from today!
Reentering some of it. Will work locally now.
adc1: 8(9) replaced lemo on delay board, ok
adc5: 9(74) replaced bad HV cable, ok
Check TDCs. With HV and LV off get few noisey channels.
With LV on get lots fo noisey channels
LV off.
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
12 0 414
14
16
12 0 141 22
14 0 252
16
set_dac -1.8, LV on
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
12 0 448 465
14 0 307
16 0 0 0 0 110 46 189 98 279
12 0 333 147 221
14 0 81 251 9 32 476 312 428 89
16
set_dac -1
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
12 0 331 5 219 88 118 160
14 0 33 193 20 76 15
16 0 39 189
12 0 131 217 68 259 36 416 330 219
14 0 68 373 64 463 389 105 208 361 198
16 0 44 18 19 18 498 26 62 69
set dac -3
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
12 0 229 419
14
16 0 181 293 166
12 0 156
14 0 120 288 105 120 419 321 394 129
16
How to restart ifix server (from Holger) PC in Mtest is pixel06.
Click Login/Logout
Logout Public
Login Holger, killi
Ctrl-W to get ifix window such that you can minimize it,
so can get to Windows Start Menu.
Start:Run, cmd
cd hmeyer
iFix2XmlRpc 80
Minimize
Ctrl-W
Logout Holger
Login Public,
Back to TDCs... Tested 4434 scaler channels 1-16 using NIM-ECL and
pulser (pull off signal from start of 2323 using led pulser
input). Put TDC cable directly into scaler -> get counts. Put daisy
chain with TDC get TDC hits and scaler counts.
Decided to re-examine digital output directly. Tube 53 has excellent
output. Ran blue and white digital output on coax to rack 17. Trigger
on analog output (pull off delay board, lemo connector is tied to
output of delay). At 1800 V can trigger at 200 mV. See nice (negative)
signal. Set threshold to -1.8V
Put blue (+) directly into scope and set to AC coupled, 20 mV. See
fuzzy bipolar signal +/- 10 mV. Note that digital output comes 300 ns
before analog signal rise because of delay board. Can do same with
white (-) digital output. Signal is about twice as big (also bipolar
and fuzzy, i.e. burst of high frequency bipolar pulses). Now change
threshold (should be between -.8 and -1.8 which is the ECL logic
swing). As you go down from -1.8 to -1.5 start to get much more
noisey. By -1.3 the output is constantly ringing. Going abve -1.8
doesn't seem to have any effect.
OK now go back to the old trick of using ECL-NIM module to get -5V
pull down. Switch scope to DC50 Ohm input (was AC 1Meg). Using NIM
Module PS Model 726 ECL-NIM.
If you put blue (+) signal into ECL+ and connect GND to ECL- you get nothing.
If you let GND float and only connect signal to ECL+ you get a string
of short negative pulses about -300 mV. These pulses are 500 ns after the rise
of the delayed analog pulse. About 300 ns before the analog pulse there are
even smaller, about 50 mV pulses. Note that we are looking at the NIM output.
Now if you put blue (+) to ECL+ and white(-) to ECL- into ECL-NIM
module you get much larger pluses. The 300 mV pluses are now about 1V
and the formerly 50 mV pluses are about 250mV. Swapping polarity does
not seem to matter.
Now switch to Lecroy 4616 ECL-NIM module. If grounded, the lecroy does
not produce any pulses from the single ECL+ or ECL-. Letting ground
float, no signal from ECL+ into + or -, but find inverted signal if ECL-
into +.
Now differential input (ECL+ and ECL-) works with correct polarity
blue(+) to ECL+ and white(-) to ECL-. This produces large (-1.5V)
short pulses like the PS, but all pulses have the same
amplitude. There is still the bunching of early and late
pulses. Swapping polarity causes logic to flip, output is inverse of
normal polarity. Strangely I found that I had to invert the polarity
for this to work and that normal polarity produced nothing. Perhaps
one of the two lines on the jumper was not making contact? It is
very stable today (11-12-04).
Note that I am using a special widget for getting the two coax signal
lines onto a single twised pair jumper.
Now try Camac version of ECL-NIM PS 7126. This works just like the NIM
version. Swapping polarity doesn't seem to matter. Now see what
happens with the CAMAC TDC and Scaler.
Scaler sees about 50 counts over 10 milliseconds: 4434 1 5 21 10000 1
If you flip polarity the scaler still gets counts and slightly more
of them (~25%).
TDC gets hits. gate 2300 100; 3377 1 5 12 1
Flipping polarity still get counts. Daisy chain TDC and scaler, get
hits on both. Flipping polarity doesn't matter.
11-12-04
ADC value for mirror 53 is affected severely when extra delay board output
is connected to scope. With HV off get 1100 counts (should be 0). With
50 ohm terminator get zero counts, even if HV is on.
Pedestal for mirror 88 has changed. Pedestal is now 400! Was 174 yesterday.
Pedestal subtraction doesn't seem to work with such a large value.
Shortening the pulser width does not reduce the length of the base
digital output, it would appear that the long pulse train from the base
is a feature of the electronics.
Try digital output directly into 4415a discriminator. Does not work.
If go through ECL-NIM-ECL (only works for Lecroy, PS ECL to ECL
doesn't work neither NIM nor CAMAC version), can use discriminator.
Note that LeCroy ECL-NIM produces many more subpulses than the PS
ECL-NIM.
Now count pulses (set_dac -1.8)
4434 1 5 21 1000000 1
average is 930 ECL-NIM-ECL (LeCroy) + discr.
7600 Directly (no ECL-NIM or discriminator)
Discriminator cleans up multiple pulses.
Put digital output directly into TDC, blue->+, white->-. Get TDC hits
in proper time. See nothing until about 1200 nsec delay. As increase
delay see that the maximum TDC time is ~(delay-1200). If you swap
polarity, get too many pulses. If remove + or - digital input get too
many pulses.
Also see odd/even effect. TDC inputs 0,2,4,... are ok, but 1,3,5,...
get too many hits. This is probably due to the internal termination
of the board! If you send digital output through ECL-NIM-ECL to
discriminator then to TDC all channels readout ok.
Aha. If someone else resets the crates it will appear as if the
digital signal switched polarity! The threshold is automatically
set to 0 and this causes the base to put out the inverse signal.
Now daisy chain TDC and scaler. This works and seems to solve
termination problem for TDC odd channels! To see how well this
works can compare this scheme with ECL-NIM-ECL+Discrimantor to
check TDC efficiency.
Now see how the digital signal looks over the ribbon cable
instead of the pair of coax. Mirroor 53 is SB4 channel 5(of 16).
Must adjust scope to 100mV. See broad and jagged negative pulse.
Total width is about 1.5 microseconds, 100 mV peak at about 1micros
from beginning of the pulse. Leading edge is only about 25 mV.
Getting TDC hits and scaler counts. Scaler counts are much higher
than with coax. Factor of about 5. Timing seems the same, note that
pulser is set to 400 hertz, so getting significant multipulsing
from digitial.
coax
> 4434 1 5 21 1000000 1
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
4604 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
ribbon
> 4434 1 5 21 1000000 1
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
19809 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> gate 1300 200 ; led on ; 3377 1 5 12 1
Sending triggers :: Here is the data
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
0 27
0 66
4434 scaler dip settings (on side panel):
All off, input_level(veto,...)=NIM, Overflow=24
Two modules per crate, slots 18 & 19.
TDCs not in proper order (first two had terminators, last did not. can
tell because there is led between the bottom two connectors). Moved
14 to 16, 16 to 14 and replaced 12 with new module. Now only last
TDC in chain has terminators. Ran 'test_all'.
set_dac -1.8
using jumper to put 53 onto channel 0.
4434 1 6 18 1000000 1
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
20114 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
>
gate 1300 200 ; led on ; 3377 1 6 12 1
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
0 41
using full ribbon
> 4434 1 6 18 1000000 1
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
0 0 0 0 5232 0 0 0 0 252 0 0 0 0 0 0 0 0
>
gate 1300 200 ; led on ; 3377 1 6 12 1
delay= 1300 width= 200 ns
turning LED pulser on
Sending triggers :: Here is the data
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
0 28
And now tdc and scaler are practically noise free. Need to make daisy
cables for all channels. Note, do not need swaps!
Moved pulser2 inside rack 13, had to run new cables for output and
gate. Works ok and tested pulser 1 as well. It is very hot inside
the rack, let's hope it doesn't kill the pulser.
11-14-04
Tested ECL output of TDCs does not work with daisy chain to scalers
(i.e., is not tied to TDC inputs in a simple way).
Decided to use 32 channels on TDC, this simplifies cabling and reduces
module count (since scalers are also 32 channel). So removed TDC from
slot 14 (two remaining in 12 and 16). Put scaler in slot 14 and
18. Need to make daisy-chain cables.
Asked Lange to modify TDC test program to read-out 32 channels.
11-15-04
Picked up parts from stockroom for cables. Note stockroom doesn't
stock male connectors for ribbon cable! Scrounged a few from Bob Jones,
cleaned him out.
Made daisy cable. Wings on male connector need to trim otherwise prevents
cable from plugging in.
11-16-04
Bought utility knife at Ace hardware $5. Trimmed connector, now it
fits. Test failed because TDC periodically hangs. Seems to be software,
called David. David fixed bug in TDC test program. Now works ok.
Made five more cables (two female and one male connectors). Had to
trim "ears" so that they would fit next to each other on TDC in slot 12
(in both crates).
Cabled crate 6. Test channel (mirror 53) works as before. Now all channels
are nice and quiet with HV off and threshold set to -1.8V, but with HV
on see TDC hits only in channel 53 with gate 1300 200. Need to delay
TDC trigger to 2000 to 2300 ns to get hits on many channels. Also see
funny multi-hits on TDC. In some cases, when there are many
hits on a single channel, the time increases and then jumps to a lower
value and increases again. For example:
hv all set to 1800
gate 2000 200
set_dac -1.8
0 1 2 3 4 5 6 7 8 9 10 11 12 13 ...
0 94 121 93 15 7 22 19 54
0 196 134 70 20 55 38 96
0 217 194 91 84 131 50 117
0 236 234 123 136 69 76 168
0 161 12 179 17 99 190
0 220 55 30 48 54 207
0 109 49 70 272
0 132 74 313
0 192 331
0 343
0 404
0 18
0 134
0 209
0 247
0 293
0 307
0 323
0 384
This bizarre problem appears to be due to grep! This error happens
only if output is piped to grep -v, e.g.:
3377 1 6 16 1 | grep -v Here
This was done in read_tdc script. This is also reproducable on the
command line. About 30% of the time the above command will produce
the strange output. Ommitting the grep eliminates the problem.
Installed remaining TDC daisy cables on crate 5. Scalers and TDC
output are almost perfectly quiet with HV off and threshold set to -1.8.
See lots of counts on scalers and TDC hits with HV on. Don't see
hits on all channels however.
Discovered delay board chip fell out of board below crate 6, was
channels 81-88. Found chip inside rack and reinstalled.
11-17-04
Check ADCs using plateau.each. Noticed that pedestals on adc5 went down by
about 8 counts. ped_get 1300 3000. Redo plateau.each 5. Adc5 pedestals
fluctuate more than other channels (this has always been true). If you
average over 100 events they are fine, but with single event see about
1-4 counts. Note that read_adc_ped averages over 100, but plateau
uses just 1 event.
Old problems:
adc2: 15
adc3: 5 light at 1500 then nothing
adc3: 13
adc4: 1
adc5: 11
adc6: 1
adc6:15
New problems:
adc1: 8 wirewrap came out again, fixed
adc4: 2(51) hv, lv, analog ok.
adc6: 4(85) hv ok, disconnecting and reconnecting lv seemed to solve problem
Debug mirrors 28 and 32 on top. Replaced tube in 32 with group 2 from
spares. No change. Swapped bases with 28, new tube sees light. Base 32
must be bad, note that there seems to be a loose socket on base. When
you pull it out of tube, one socket sticks up a bit (can push it back down).
base from 11 put on tube in 15 (32)
adc=2 crate=5 slot=6 hvslot=5 width=3000 chan=11 (actually is 15)
1500 0 0 1 0 0 0 0 0 0 1 0 9 1 1 0 0
1600 0 1 2 1 1 0 1 2 1 1 2 16 0 1 1 0
1700 0 1 1 1 1 0 0 0 0 1 1 24 0 0 0 0
1800 1 0 1 1 1 1 0 1 1 1 2 47 2 2 1 0
1900 1 1 1 1 0 0 1 0 1 1 1 70 3 2 1 0
2000 1 1 2 1 1 1 1 2 1 1 2 96 1 2 1 0
0000 1 0 1 1 1 0 1 1 2 0 2 0 0 1 2 0
Tube in 28 must be bad (doesn't hold HV), so put tube from 32 (26463)
into 28. Using 28 base. Works great now!
Replaced tube in mirror 50 worked great until put screws back in. This
tilts the base and it is extremely sensitive. After a long while I got
the bolts in and the tube still works.
Tube 51 probably has the same problem.
26499 group 2 spare
? group 2 spare
? group 1 spare
11-30-04
No exec16 command on beamppc, because srt is broken (points to devonline which
is empty of binaries). Andre is in Russia, Holger said call Mike. Mike also
has this problem and has ugly workaround. After a bit of investigation I
discover the workaround is:
srt_setup SRT_BASE_RELEASE=pro
Lange and Heffner fixed the default so that it is now pro.
Now my remote plateau jobs (from e907hv) fail because ssh environment must
explicitly point to PATH, fixed .ssh/environment path to pro, but still
fails. Finds exec16, but complains about missing shared library.
exec16: error while loading shared libraries: libVMEInterface.so:
cannot open shared objey
Copied PATH and LD_LIBRARY_PATH from interactive session into environment.
Now ok.
Now ssh failing on beamppc, gets stuck at sending public key. Reboot beamppc.
Now ok.
Base on mirror 32 is #94.
Base on mirror 76 is #57.
Wrote script adc on e907hv. Reads back adc values and prints both
relative and absolute channel number.
Modified plateau script to record absolute channel number too.
Repair of channel 32, bad base (#94). This base was not originally
tested (nor where 95, 96) because they were missing at the time of the
original test. By the time we rounded them up, the PMT test stand was
ripped apart.
Peter swapped pcboard (from hodoscope base) on base #94, still doesn't
work. Peter tried again (he noted that the pcboard he put in had
some visible defects). The second time was the charm, it works.
Debugged mirror 76. Checked HV, LV, and analog lines are ok. Removed
the base (#57). This is the base with the white PMT connector that
gets shoved in. It is bad, tube worked with newly repaired base #94.
Peter attempted to replace white base connector with blue one from
hodoscope spare. First attempt did not work.
Mirror 51 had bad base (#6). Tested with base #94 and it is reading out
ok again (is saturated, i.e., 2047). The worrisome feature here is that
this was a good channel for many weeks and then the base failed. This is
the first instance of base failure during use. Peter replaced the
circuit board on base #6, reinstalled.
In the process; however, mirror 50 is bad again. This is the base that
doesn't work when tilted. Squeezed the bracket on the downstream side.
This fixed it again (120 ADC counts at 1900V).
Checked mirror 32 with repaired base #94. Tube works. 61 adc counts at 1900V.
Mirror 38. Bad base 2047 adc counts at 1900V with base #94. Tube is
very loose, slides out easily. Noticed that gas pressure is much less
than last time I removed a tube. Can still feel initial wind that dies
down to a very light breeze in a few seconds.
Swapping HV on 43 caused 43 counts to reduce from 1000 at 1900V to 40!
Pushing down on HV connector reduced adc counts to 20, pulling up
restored to 40. Swapping LV connector with 46 caused this to come back
to adc counts 1100. Put original LV back on 43, still 1000 adc counts!
Gain was low again (must have bumped while working on 46). Push on LV
connector. The whole base moves (on springs) and that seems to solve
the problem.
Mirror 46 is in the beam-right cut out section. Found that the analog
subminex connector on the base is out of round and center pin was not
going into analog out cable. Also, tube was not lined up properly
with holes in shield, so base was probably not attached correctly.
Fixed this and now it works.
Mirror 45 is also touchy. Went dead while working in vicinity. Push on
base next to HV cable brought it back to life.
Mirror 96. Swapped base with #94, see 50 adc counts when HV first comes
on but then quickly goes to zero. Try swapping tube.
Pedestal on mirror 88 drifted up by 100 counts. Redid ped_get 1300 3000. ok.
Peter tried to repair many bases by swapping circuit board, but did
not work. Now he noticed that the resistor network on the PMT
connector are bad.
Peter suceeded at making a test base out of hodoscope spares and a
a HV connector scavenged from a third style of base.
All mirrors debugged and fixed. Run plateau.each. adc_data8
adc6 85 chan=4, low light output and is half of what it
used to be (is now 30 counts at 2000V, was 60)
Redid plateau.each 6 and copied to adc_data8
adc3 mirror 45 chan=12 plateau bad, this is touchy base#19
This is a whitey PMT connector. Should replace with blue base so
it is stable.
adc4 52 chan=3 flakey plateau. Spare base did not work at all.
swapped in tube 26499 group 2 (was 27035), still flakey. Bad HV
cable. Rerouted spare 2 from top to bottom.
Redid plateau.each. adc_data9
1 5(6) dead, touchy base syndrome
adce907hv$ plateau 1 5
adc=1 crate=5 slot=4 hvslot=6 width=3000 chan=5 tube=6
1600 1 0 0 1 0 25 1 0 1 0 0 2 2 1 0 3
1700 1 0 0 0 0 32 0 1 1 0 2 2 2 1 0 0
1800 1 0 0 0 0 43 0 0 1 1 1 1 1 0 0 0
1900 1 0 1 1 2 77 1 1 0 2 0 2 1 1 0 2
2000 0 0 0 0 1 110 1 0 2 0 1 1 1 1 0 0
0000 1 0 0 0 1 0 0 0 1 2 1 2 1 1 0 0
adc4 1(50) dead, touchy base, squeeze downstream side
e907hv$ plateau 4 1
adc=4 crate=6 slot=4 hvslot=3 width=3000 chan=1 tube=50
1600 2 41 2 2 1 2 1 1 0 2 0 0 2 0 2 0
1700 1 68 4 1 2 2 1 2 0 1 0 0 2 0 1 0
1800 3 97 2 0 2 1 0 0 0 1 4 1 2 0 0 0
1900 0 137 3 1 2 2 2 2 0 0 0 0 2 1 1 0
2000 0 189 2 1 1 1 2 1 1 1 0 0 3 1 0 0
0000 1 2 3 2 0 0 4 1 0 0 0 0 1 0 0 0
adc6 4(85) nearly dead, touchy base, push on LV connector, ok
adc=6 crate=6 slot=8 hvslot=1 width=3000 chan=4 tube=85
1600 3 0 4 2 6 3 0 17 0 9 0 2 0 0 3 2
1700 0 1 2 3 16 1 4 23 0 5 0 1 0 0 4 0
1800 1 2 6 3 30 1 0 25 0 7 0 0 0 0 7 2
1900 1 1 3 5 48 2 4 23 0 6 0 1 0 0 5 1
2000 0 2 5 4 63 3 1 35 0 6 0 1 0 0 4 2
0000 2 1 1 2 0 0 3 28 0 6 0 2 0 2 5 2
e9
Redid plateau.each 1 (copied to adc_data9).
Redid plateau.each 4 (copied to adc_data9).
Redid plateau.each 6 (copied to adc_data9).
Redid ped_get 1300 3000.
Turn all HV to 2000:
2047 181 2047 101 108 125 2047 455 2047 254 2047 190 2047 165 2047 212
1467 212 2047 354 2047 210 2047 100 2047 200 2047 684 2047 165 2047 108
2047 9 970 2047 112 2047 2047 2047 906 1292 1308 76 8 2047 139 132
2047 198 2047 204 2047 208 2047 472 2047 187 2047 151 2047 231 2047 75
2047 68 2047 177 2047 269 2047 107 442 264 2047 270 2047 87 2047 0
662 90 427 2047 59 1587 2047 1668 2047 2047 2047 59 358 2047 66 80
STATUS
------
adc5: pedestals on all channels fluctuate upward, note that average
reduces this effect
ToDo
o check TDC signal is due to light by reducing led amplitude
o look at digital output on scope for other mirrors to check timing
o Update Ed's document 10ohm resistor is 50ohm
Hodoscope base is six stage, ours is seven, hodobase backboard is different
o get HV connectors and have Andre repair CKov end of bad cables (3 on bottom)
o get NIM crate or make extension cable, try digital into lecroy
ECL-NIM to see what you get
Repairs
-------
adc11 5(6) dead, touchy base syndrome 12-4-04
adc1: 8(9) resolder lemo connector on delay board
reinsert wirewrap on back of delay board
reinsert wirewrap again (11-17-04)
adc1: 15(16) delay chip was loose
adc2: 1(18) bad low voltage, made splitter
adc2: 11(28) tube was bad 26742, replaced (could not hold hv)
adc2: 15(32) bad base#94, Peter replaced circuit board
used this as a spare for while
adc3: 5(38) bad base #85 (very loose fitting tube),
Peter replaced circuit board
adc3: 10(43) if gain is low (40counts at 1900) push on base at LV
connector, shold be 1100 counts at 1900
adc3 12(45) plateau bad, touchy base#19, replace white PMT connector
David replaced white connector with rebuilt blue, saved the
small backing board, but replaced the circuit board
adc3: 14(47) replaced analog line between base and patch panel
adc4: 1(50) Touchy base, replaced tube, but problem was screws tilted base
11-30-04: must have been nudged while debugging 51
fixed it by squeezing downstream side screw
12-3-04: dead again, squeeze downstream side screw
adc4: 2(51) bad base#5, this base worked fine originally
Peter replaced circuit board, ok.
adc4 3(52) Flakey plateau. Swapped in tube 26499, still flakey.
Bad HV cable. Rerouted spare 2 from top to bottom.
adc4: 4(53) HV connector loose on HV module
adc5: 6(71) analog line was swapped with 72
adc5: 7(72) analog line was swapped with 71
adc5: 8(73) replaced bad HV cable
adc5: 9(74) replaced HV cable with spare
adc5: 11(76) bad base#57, white PMT connector was pushed in, Peter rebuilt
PMT connector from two hodoscope spares and swapped board
didn't work, Peter replaced circuit board again,
still doesn't work. Bad HV resistor chain?
adc6: 1(82) bad base#27, this is the damaged mirror
Peter swapped board, PMT reistors checked out ok, but still bad
Peter swapped board again, ok.
adc6: 4(85) touchy base, reconnecting lv seemed to solve problem
12-3-04 nearly dead, push/twist LV connector
adc6: 14(95) analog to patch panel mislabeled, fixed ok
adc6: 15(96) analog to patch panel mislabeled, fixed but still dead
bad base#50, Peter swapped board but still bad.
Peter found bad 1Mohm resistor to ground in HV chain. Now ok.
touchy base syndrome adc1_6, adc3_43, adc4_50.
[] = waiting for repair
<> = touchy base syndrome
Beam left Beam view Beam Right
----------------------adc3----------------------------+
35 38 39 40 41 42 45 48 |
+-----adc1-----------adc2------+ |
| 4 8 12 16 | 20 24 28 32 | |
34 37 | 3 7 11 15 | 19 23 27 31 | 44 47 |
| 2 <6> 10 14 | 18 22 26 30 | |
33 36 | 1 5 9 13 | 17 21 25 29 | <43> 46 |
-----------+--------------+---------------+-----------+
81 84 | 49 53 57 61 | 65 69 73 77 | 91 94 |
| <50>54 58 62 | 66 70 74 78 | |
82 <85> | 51 55 59 63 | 67 71 75 79 | 92 95 |
| 52 56 60 64 | 68 72 76 80 | |
+-----adc4-----+-----adc5------+ |
83 85 87 88 89 90 93 96 |
-----------------------adc6---------------------------+
Spider Box Grouping
-----------------------------------------------------+
39 40 | 41 42 |
SB1 +-------+-------+ SB3 |
35 38 | SB2 | 45 48 |
4 8 | 12 16 20 24 | 28 32 |
34 37 3 7 | 11 15 19 23 | 27 31 44 47 |
2 6 | 10 14 18 22 | 26 30 |
33 36 1 5 | 9 13 17 21 | 25 29 43 46 |
--------------------------+--------------------------+
------------------+---------------+------------------+
81 84 | SB5 | 91 94 |
SB4 49 53 | 57 61 65 69 | 73 77 SB6 |
82 85 50 54 | 58 62 66 70 | 74 78 92 95 |
51 55 | 59 63 67 71 | 75 79 |
83 86 52 56 | 60 64 68 72 | 76 80 93 96 |
+-------+-------+ |
87 88 | 89 90 |
--------------------------+--------------------------+
12-14-04 DMW & DL @ FNL
Ran plateau.each, beware hv_monitor on e907hv. Will interfere with
my direct control and confuses shifters.
Mirabile Dictu, see signal on all channels! A few channels do not reach 100 counts
(above pedestals) at 2000V:
adc4 15(64) adc5 15(80)
1600 0 8
1700 0 13
1800 1 13
1900 25 17
2000 47 29
0000 0 2
adc=2 crate=5 slot=6 hvslot=5 width=3000 chan=15 tube=32
1600 13
1700 22
1800 34
1900 58
2000 80
0000 2
adc=3 crate=5 slot=8 hvslot=4 width=3000 chan=1 tube=34
1600 8
1700 5
1800 10
1900 14
2000 22
0000 2
adc=3 crate=5 slot=8 hvslot=4 width=3000 chan=11 tube=44
1600 0 1 0 0 1 1 2 1 1 0 1 14 0 1 1 1
1700 1 1 1 0 0 1 1 1 1 0 1 18 0 1 2 0
1800 1 1 1 2 0 1 1 1 1 1 1 39 0 2 3 2
1900 1 1 1 0 1 1 2 1 1 2 1 47 0 0 2 1
2000 1 0 0 1 0 1 2 0 1 0 0 77 0 1 0 0
0000 2 2 2 2 1 2 2 1 1 0 1 1 0 1 1 1
adc=5 crate=6 slot=6 hvslot=2 width=3000 chan=1 tube=66
1600 0 10 1 0 0 0 2 0 0 0 0 5 1 0 0 2
1700 0 17 2 0 0 0 0 0 0 0 0 4 1 2 0 2
1800 0 24 0 1 0 0 1 0 0 0 0 5 1 1 0 3
1900 0 52 1 0 0 0 1 1 0 0 0 1 0 0 0 0
2000 1 53 3 0 0 0 0 1 0 1 0 3 0 1 0 0
0000 1 0 2 0 0 0 0 0 0 0 0 1 0 2 0 2
adc=6 crate=6 slot=8 hvslot=1 width=3000 chan=1 tube=82
1600 0 9 4 3 0 3 0 0 0 19 0 1 0 1 0 1
1700 0 17 1 0 0 0 0 0 0 18 0 4 0 1 0 0
1800 0 33 2 0 0 1 2 0 0 18 0 2 0 1 0 0
1900 1 42 1 2 0 1 0 0 0 19 0 1 0 2 0 0
2000 0 78 2 0 0 1 0 0 0 18 0 3 0 1 0 0
0000 0 1 3 0 0 2 3 2 0 17 0 1 0 1 0 0
adc=6 crate=6 slot=8 hvslot=1 width=3000 chan=4 tube=85
1600 2 0 4 0 7 1 0 0 0 18 0 1 0 0 0 0
1700 0 0 3 3 13 2 1 0 0 19 0 3 0 1 1 0
1800 0 1 0 1 24 1 3 0 0 15 0 1 0 2 0 0
1900 0 1 2 1 32 1 0 1 0 18 0 3 0 1 0 2
2000 0 0 5 1 70 1 2 0 0 16 0 1 0 0 0 0
0000 0 0 2 1 0 0 4 0 0 18 0 3 0 2 0 0
adc=6 crate=6 slot=8 hvslot=1 width=3000 chan=11 tube=92
1600 0 1 6 1 0 2 0 0 0 18 0 7 0 0 0 0
1700 1 0 2 0 0 0 6 0 0 21 0 18 0 1 0 0
1800 0 0 2 1 0 0 4 0 0 18 0 29 0 1 0 0
1900 2 1 5 1 0 1 6 0 0 18 0 34 0 0 1 0
2000 0 1 6 1 0 2 4 0 0 17 0 45 0 0 0 0
0000 0 1 6 1 0 1 0 0 0 18 0 1 0 0 2 0
adc=6 crate=6 slot=8 hvslot=1 width=3000 chan=14 tube=95
1600 0 1 6 2 0 2 0 0 0 18 0 0 0 0 8 3
1700 0 1 3 2 0 2 0 3 0 20 0 3 0 2 12 0
1800 0 1 5 2 0 2 1 8 0 16 0 1 0 0 24 1
1900 0 0 1 0 0 1 2 0 0 18 0 2 0 1 31 0
2000 2 3 7 1 0 2 0 0 0 18 0 1 0 1 38 0
0000 1 1 1 0 0 0 5 0 0 18 0 4 0 3 0 0
adc=6 crate=6 slot=8 hvslot=1 width=3000 chan=15 tube=96
1600 0 0 6 2 0 2 0 0 0 15 0 1 0 0 1 19
1700 0 0 6 2 0 1 0 0 0 17 0 0 0 0 2 21
1800 0 0 2 1 0 2 0 6 0 17 0 3 0 2 0 34
1900 0 2 2 0 0 0 2 0 0 18 0 3 0 0 0 46
2000 0 1 6 2 0 3 0 0 0 17 0 1 0 0 1 64
0000 0 1 2 0 0 1 6 0 0 18 0 2 0 1 0 0
Test spare bases
adc3 5 (38)
original base#85
adc=3 crate=5 slot=8 hvslot=4 width=1300 chan=5 tube=38
1600 2 2 0 1 1 359 1 1 0 1 0 1 0 1 2 1
1700 1 2 0 1 2 633 0 2 0 1 1 1 2 0 0 1
1800 0 1 1 1 1 883 0 1 0 1 2 1 1 0 1 1
1900 2 0 0 1 1 1264 2 0 0 1 2 1 0 1 0 0
2000 2 1 0 2 1 1650 1 0 0 0 0 1 0 1 1 0
0000 3 0 1 1 1 1 0 0 1 1 1 0 0 0 1 0
spare base#S1
e907hv$ plateau 3 5 1300
adc=3 crate=5 slot=8 hvslot=4 width=1300 chan=5 tube=38
1600 2 1 2 0 1 383 1 1 1 2 1 1 2 0 2 1
1700 2 1 0 0 1 519 1 1 1 2 2 1 1 0 0 0
1800 1 2 1 2 1 795 2 1 0 1 1 1 1 0 0 0
1900 1 1 1 1 1 895 2 1 2 1 1 2 1 0 1 2
2000 2 1 1 2 2 1170 2 1 2 2 1 2 1 0 1 2
0000 1 2 1 1 1 1 0 1 2 2 0 0 2 0 2 2
e907hv$ plateau 3 5 1300
adc=3 crate=5 slot=8 hvslot=4 width=1300 chan=5 tube=38
1600 1 1 0 1 1 336 1 0 0 0 1 1 0 2 0 0
1700 2 1 1 0 1 471 2 0 1 1 1 1 1 2 0 1
1800 0 1 0 0 1 702 1 0 1 1 1 1 0 0 0 2
1900 1 2 1 2 2 810 1 2 2 1 1 2 0 1 2 2
2000 2 3 2 1 2 1225 3 2 1 3 3 2 2 1 0 0
0000 0 2 1 1 0 2 2 0 0 0 1 0 1 1 0 0
spare#2
e907hv$ plateau 3 5 1300
adc=3 crate=5 slot=8 hvslot=4 width=1300 chan=5 tube=38
1600 1 0 0 0 0 33 2 0 1 2 1 1 1 1 0 0
1700 1 0 1 0 1 42 1 1 1 0 0 1 2 0 0 1
1800 0 0 1 0 1 69 0 1 1 1 1 0 1 0 1 1
1900 2 1 0 1 1 89 0 0 0 1 0 1 0 1 1 1
2000 0 1 0 0 0 102 1 0 1 1 0 0 0 0 1 0
0000 1 2 2 1 1 1 0 0 2 0 2 1 1 2 0 0
spare#2 works, but has 1/10th the gain! Don't use for now.
Proceeding with gas. Used Matheson leak hunter 8065 to sniff for CO2
around PMTs on top. Was working with tube=38 anyway, found had to
practically remove tube before leak monitor detector gas, wiggling or
sliding tube back out somewhat (there are two orings inside that
provide seal) did not register. Nominally sensor should detect .02
cc/min (this is reading of 10 when set to scale=1). Looked for leaks
in about 20 tubes on top, beam left. Did not detect anything.
Nominal leak rate of entire vessel is 60cc/min, which would be about
0.5cc/min, which should be easy for the sniffer to detect.
4:40pm Put first CKOV C4F10 bottle online. Scale says installed bottle plus
regulator is 202 lbs. Label on tank says Gross weight 90kg, tar weight 27kg,
net weight 63kg (i.e., weight of gas). Bottle #1 installed on left side
flex line, connected to regulator #1. Bottle pressue is 13psi. Regulator
set to 12psi
28.32 l/cuft
2.20.. lbs/kg
10.12 g/l density of C4F10
remote scope
1 tube 13
2 NIMbar crate 6
3 tube 17
4 tube 65
Check timing of ADC gate. Separated led gate for inside and outside
the hall. DAC Channel 7 is now inside, channel 6 is outside (counting
house) pulser enable.
Noticed that turn on of LED is slow, see signals building up, but
shut off is steeper. Shortened (inside) pulser width to indicated
0.25x10microseconds on dial. The peak of the analog output is
about 600 ns after the initial (slow) rise of the output.
Tube 65 is really strong and clean signal at 1900V. The analog line is
also on the top of the delay board, so easy to access. Can split the
input and look at both input and output of delay board. Verified, once
again, that the lemo output on the outside of the rack (no chips) is
the delayed output. Can see that it is about 250ns as advertised.
Verified that NIMbar and NIM output of 2323 are correct. Measured
time difference between 2323 start and NIM output is ~22ns.
Scope now set with
1 tube 13
2 tube 17
3 tube 65
4 NIMbar
Captured traces at .25x10micro, 1x10micro led pulse width.
Put output of linear fanout into scope (chan2) set trigger to
shortest possible delay 50nsec. Did not account for difference
in cable lengths. Captured trace with LED signal on tube 65 and 13.
Ran signal directly from phototube 77 @1900V to scope using 48ns+
coax. Compare to input to chan A of 2323 from led pulser. First
signal is 1200 nsec after start of led pulse. The bulk of the
signal is at 1600 nsec. Capture data from scope.
Back to filling Ckov with C4F10. Noticed that consumption (via weight loss)
continued at about 3lbs/hour at fast (manual) fill, but pressure change
per hour became reduced when near indicated 0.55 inches of water at
the bottom. Switched back to automatic, slow fill.
Can not maintain tank pressure with intake turned off, consuming
1lb/hour to keep up with leak and increase pressure. Note that at
this point the differential pressure between the gauges is not changing.
Found (with Matheson leak hunter) big leak on back (downstream) window
on beam left, bottom. Tightened about 8 bolts in the vicinity (to move
the whole corner and not just pinch a single screw). Can now hold tank
pressure with no intake. Unfortunately noticed that bottle seemed to
sltill lose weight without flowing, which means there could still be a
leak between bottle and solenoid valve.
Did more timing studies with Andrew. Measured trigger delay and our
cable+chip delay. Regular trigger should arrive about 40ns before our
signal, but noted that output of chip is very ugly (wide signal) for
nice clean 300mV pulse input. The leading, soft rise of the signal
comes before the nominal leading edge of the original signal with the
nominal delay. Our cables introduce a delay of <130ns.
1-4-05
C4F10 gas bottle ran out at 3am 1-4-05. New bottle put online by
Andrew and Andre. New bottle reading 175lbs. Should be 90
kilos=200lbs. Andrew thinks he broke the scale (lifted the bottle by
himself).
Multiple power failures over the weekend and today. Crate 6 was
powered off. Turned on and DAQ seems ok.
Proposed scheme to use pretrigger ORed with regular trigger so can do
timing studies without interferring with the TPC fake-trigger at start
of run. Andrew says that should work. The only caveat is that we can't
keep this for the real experiment because the trigger system has a
feature to prescale the trigger depending on the beam cerenkov output,
i.e. if the user requests to prescale K or pi or whatever, then we can
get pre-triggers sent to our crates without the event being
readout. Since there is no event clear sent for every pretrigger this
can cause incorrect data (from the wrong trigger) to be read out of
our system. So to use the pretrigger we must make sure that all
pre-scale values are set to zero, i.e. no prescale.
I found manufacturer of passive (analog) delay chips that can be used
to increase the delay on our boards. Can get delays of 300, 375, 500,
1000ns. Circuit C has the same pinout as our current chips.
http://www.rcd-comp.com/rcd/passive.htm
http://www.rcd-comp.com/rcd/rcdpdf/SMP01S-P01S-P01-S01.PDF
1-5-04
2pm bottle weight 170lbs. Used 4lbs in 20 hours.
Trigger timing study.
Now have pretrigger to crate 5 (and original trigger). Found that
pretrigger needed upstream terminator to work properly in 2323.
Connected pretrig and trig to 2323 Start channel B.
Now with gate=100 delay=10 (minamal settings) see two pulses 144ns apart,
each has width of 88ns.
Increase gate width from 100 to 200. Pulses merge at gate>150ns, i.e.,
see single pulse of 276ns -> getting both triggers.
But set gate>170ns and get single pulse of 158ns (nominal-12ns).
2323 in crate 6 started acting up. Chan B trigger light stayed on with
no inputs! Replaced with spare (had to change spare to jumper BC so
ECL is GATE). Also set trigger threshold to -300mV. Spare worked ok
until tried to use Chan A for led trigger. It then was pegged on and
stayed on without any inputs. Reprogrammed channel 0 (A) gate delay to
be 100 10 and it works fine now.
Trigger setup that allows LED, trigger, and pre-trigger on both crates:
Crate6 ChanA
Start1 LED
Start2 50ohm
DLY T with 50ohm and 1ns cable to ChanB Start
Crate6 ChanB
Start1 1ns cable from ChanA DLY
Start2 10ns cable to Crate5 Start2
Crate5 ChanA
nothing
Crate5 ChanB
Start1: T with Trigger and Pretrigger
Start2: 10ns cable from Crate6 Start2
Note that trigger and pretrigger are both 50ohm. Pretrigger required
T and terminator at the origin. Can removed either trigger or pretrigger
and still get signal as expected.
So Crate 6 should get triggers 10ns after Crate 5.
Verified ECL output of crate5 and crate6 by sending to ECL-NIM.
Jumper cable plus module adds 12ns of delay. Verified with scope that
width of ECL is set by GATE, Delay is ignored. Also verified that
crate6 triggers come 10ns later than crate5.
Verified trigger+pretrigger will trigger both crates. If GATE set
>170ns get one pulse with width=GATE, if 150-170 get one pulse with
width=GATE+144ns (time beteen trig and pretrig), if <150nsec get two
pulses separated by 144ns.
NIM and ECL output of both crates is about 12ns shorter than GATE setting.
Better solution: Put Logic Fan-In/Fan-out into crate5 (need CAMAC to NIM
converter from Prep). Then run all trigger inputs to Fan-in and need
only one output to each crate. Requires 10ns cables to reach from
one crate to the other. I set this up with a temporary NIM crate in
between the racks. No need for upstream terminator on pretrigger line.
Time between trigger and pretrigger is now 160ns (used shorter cable
between racks).
Setup trigger so that pretrigger goes into Chan A of each 2323. Also
have scope set up to show difference between actual trigger and fixed
experimental trigger (which we believe is about 20ns too late). Can
now delay pretrigger and can do timing scan to see how much we lose
with fixed experimental trigger. Unfortunately, MCR just took away
beam so can't take data.
1-6-05
Checked out two CAMAC logic units 4564s (to OR TDC output) and
replaced possibly flakey 2323 with fresh one.
Prep did not have CAMAC to NIM adapters. Did have mini-NIM crate with
four slots. This turned out to be negative work because crate power
supply was weak so that some modules worked (fan-out) but the
coincidence unit did not work. It took a while to prove that the crate
was bad and not the module. Was able to squeeze in full-size NIM
crate standing up behind rack 12.
Now have world's-greatest fast trigger for CKOV. So-called pre-trigger
goes to Coincidence NIM module in our stand-up NIM crate. Use inhibit
line from trigger crate to veto coincidence module (so not really
coincidence since only one input=A). Output of coincidence goes to
logic fan-in/fan-out. Fan-in also has LED pulser input and direct CKOV
trigger input (this is needed to avoid problem when TPC starts up and
DAQ sends software trigger for the first event). Put direct CKOV
trigger on the scope, with the same cable length as would be required
to run to 2323, so on the scope can see the delayed pretrigger and the
direct CKOV trigger in time.
Took a bunch of short runs with various delay settings, captured scope
traces to .png files on my laptop.
Look here for .png files with scope traces.
It looks like delay of 142ns puts pre-trigger in time with direct
trigger. So it looks like the best setting now would be delay of
100ns which is 42ns earlier than the current direct trigger.
Let's try taking data at 50, 100, 120, 140, 160, 180.
Note that Ckov readout time is affected by digital output threshold:
-1.8V 2.75 microsec
-1.2V 3.5
Now take data with 50ns delay and 300ns gate. ADC 5 is not reading out.
DAC set -1.2V HV 1900V
deltaT based on TDC peak
10947 50 300 12336 ---
10948 100 300 10318 102
10949 120 300 10222 122
10950 140 300 10242 142
10951 160 300 11651 162
10952 180 300 11257 181
ADC5 was not reading out. Raw readout reports all pedestals below 5
counts at 300ns gate. Typically is about 30. Replaced ADC module, but
now not getting LAM. Touched ECL cables, started from terminated
module with single jumper. Did crateInihibit 0 by hand, this is done
at beginning of test_all, but may be useful when testing adcs
individually. Eventually got all modules working again. Did test run
with beam, online monitor concurs.
1-7-05
Night crew took data for us with 3% interaction target. 20k events
per run at various magnet settings.
10978 20.0 GeV
10979 18.5
10980 17.5
10981 16.5
10982 15.0
Analyzed timinqg data from yesterday with offline code
CkovPlots. Implemented TDC cut (plot ADCs only if event contained TDC
hit within peak). Does decent job removing pedestals.
Look here for ADC, TDC plots for runs 10947-10952
and *.gifs with normalized distributions with cut on TDC values.
TODO
----
Pressure-drop test on line between gas shack and 3-way valve in MC7.
Set DAC value by taking beam data and looking at amount in peak over
background base level.
Measure pedestal versus gate width distribution, if no dependence, then
scan left and right edge of timing window
To test my theory that 140ns gate chops off negative part of signal (and thereby artificially enhances high charge part of plot) make very narrow
gate (comparable to size of feature) and scan it across this region, should see negative pedestals just before this region
Wish list
o Repair HV cables on bottom, vessel side
o Spare bases x 2
o Spare LV supply cables on top and bottom
o Fix LED fan out box
o Permanent ladder to top
o LEDs that illuminate all mirrors
===================
1-11-15 Lange @ fnal
0. Fixed lam problem by adjusting cable. Will try to find more of the
appropriate connectors and cable to make a new one tonight.
1. Running c4f10 leak check. Have valved off bottle, closed 3-way valve
in mc7 rack as well as valve at the top of the rack. At 11:20pst, the
gauge in mc7 reads 15.5 and the one in the gas shack reads 20. Ckov in
manual mode.
12:34 - stopped test. Opened 3-way followed by valve at top of mc7
rack. Saw pressure at gauge in mc7 rack increase after each opening, so
there are small leaks in the rack (and up to the ckov) still. Time to
resnoop.
Resnooping did uncover a leak - however my attempt to fix it created
other (bigger) leaks. Fortunately I realized this quickly and have more
or less recovered now. Am redoing the leak check. Restarted as of
2:20pst after topping off the ckov.
At 324pst Andre opened the values for me in CA. Found that the gauge did
not move when opening the 3-way, but moved up 1 psi when opening the
2-way at the top of the rack. So thats good news. Not sure how to do a
longer test - guess need to find a time when a storm rolls through and
the atm pressure is dropping to make up for the gas fill rate.
Will watch the weight more closely. At 160 now (however opening the
bottle will change the reading, at least for a moment)
Found that I do have a key to some of Miller's stuff. Found the
appropriate socket for the torque wrench indicated by Doug.
Andrew showed me the ape-shit gas sealer. I've added it to
some of the gas panel connections. Got kicked out for S&S, but will want
to do more. Note that the ape-shit stuff is $100 can from the stockroom.
Will need more if we consider doing it on the ckov window itself.
Note that "ape shit" is Apiezon Q -- purchased initially from Kurt J
Lesker company 800 245 1656. is 1kg can Fisher Scientific has it for
164.14 for a ca.
Beam - what beam?
Used the rest of the ape shit - stil have some connections to go. WIll
consider buying more tomorrow, pending torquing the window. Mike knew
about this stuff as well.
Nick/Andre used what crappy beam we got tonight.
Tim called this afternoon - will be bringing the shorted delay chips
with him from usc on Friday afternoon. Will help test that they work or
not.
===============================================================
Wednesday
Bought new fancy scale from Target. Is digital, unfortunately it has an
auto-shutoff feature that requires the weight on the scale to be very
close to 0 before it comes back on. Stepping on scale does not work if
there is already something on it. Will take back to target
Had a few hours of beam overnight. Looks like about 10k events
in 1 hour. Possibly useful runs from the overnight:
11224
11225
11226
Updated to Doug's version of CkovReco. Decided to increase the label
sizes on x and y axis
Added trigger connections per Doug's instructions. - well not
completely, I did not yet connect the 16ns cables into chan-A
for either crate. Need to be reminded of the required termination.
Thursday
Beam overnight - eventually. 130am ish.
Shifters took data for me.
500ns gate 11288 15k
400ns gate 11287 15k
375ns gate 11286 15k
350ns gate 11285 15k
325ns gate 11284 15k
300ns gate 11283 17k
time to look at the data. Looks like 300 ns is enough. Should have also
gone the other way. Sigh. See /home/lange6/dec1.off/endtime_scan.gif
Doug points out that it would be easier to understand with mirror 13.
See try2.gif in the same dir. Clearly there is little or nothing to be
gained extending the gate beyond 300 ns. Will look the other direction.
Beam plans - no target - Andre has priority. Guess there will not be
much to get done. Can do the studies I want, I think, especially if we
can tune the beam energy.. MCR calls to say they have problems anyway.
Searched the db for Ed's data. Found runs
10802 16.5 5K
10804 17 5k
10805 17.5 5k
10806 18 5k
10807 18.5 5k
10979 18.5 22k
10980 17.5 21k
10981 16.5 21k
calculated the number of photons over the 'max number of photons' per D
Christians documentation. At 20 GeV, its only 20% for protons. (1-
pth^2/p^2)
Wrote root script to compute N/Nmax for pi/K/p in e907/050112/nnmax.cc
on my laptop.
Replaced scale - old reading appears to have been reliable. 155 pounds.
MI broke, no beam tonight
=================
Friday
Ringing channels
1, 6,7,24,25,27,31,36,37,40,44,62(barely),71,74
(44 also is on Doug's list of low signal level tubes (with led data))
(74 - we replaced the hv cable with spare one)
(6 - dead touchy base syndrome)
Per flow meter, we burned up a lot of gas last night, thanks to the
pressure change. The plots say 4ft^3 since ~8pm yesterday, when Heffner
did the check list. We lost only 1lb on the scale. This seems
inconsistent, but maybe it means we had a leak in the gas rack that is
fixed. Will see as time goes on.
Pictures of the fire on the mcr log for evening shift of 01/13. Flying
debries caused damage to a truck. Likely down or limited beam for 1
'fermilab week'. Hum, then there are the safety issues...
Ed and I have tightened the bolts on the back window of the ckov to 31
foot-pounds using the gas-techs wrench and Miller's socket. Will see
how things go overnight or so.
'Ringing' bases --- plotting them on map shows that there is a pattern -
central circle of mirrors do not have this problem (assuming beam is
higher than the center of the ckov, which the data supports). Maybe its
beam.
Tried fitting the tdc distribution of tdc1. See that for
sin((x+[2])/[3]) that [3] =6 bins so that the period is
2*pi*6/2 seconds = 18.8 seconds - 19 is claimed to be the batch
structure of the beam....
Lots of software, no beam. Maybe Ed will have better luck.
Andrew says:
- they are retuning MI with 11 of the 12 whatevers..
- will be 3 weeks before things are back to normal.
Saturday
Raised the hv maximum high voltage limit to be 2250 so that we can raise
the HV to 2200, which is the e910 (and 690) nominal run point.
Monday
Ed raised hv to 2200. no trips yet.
Things to do
- never did replace the cable going to adcs i crate 6.
- Another round of tightening back window? Wait to see if we see
a differece in gas usage.
- summarize software studies.
2-16-05
C4F10 bottle almost empty. Installed new bottle on pigtail#2, old bottle
still on pigtail#1. Note that new bottles have different fitting. Pigtail #2
was already modified to accept this. No washer used, appears that there are
no leaks.
Pigtail#1 is CGA and contains teflon washer. Pigtail#2 is a hack (epoxy job)
by FNL techs and has no teflon washer. Terry Tope confirmed that this is
the correct situation.
The labels for the three-way valve on pigtail#2 were wrong. Moved labels
to the correct position. Pigtail #2 is online. Bottle weight is 195 lbs.
Empty is 60lbs.
2-17-05
Wrote new script for reading scalers: read_scaler_rawv2. Prints channel
map and for large count rate converts to kilo or mega hz so that
columns always line up.
Dave Northacker gave me hardcopy of the following paper that compares
light yield for various gases including C3F8 and SF6. It appears that
light yiedl scales with Flourine content.
"Luminosity Monitor Based on Cherenkov Counters for P-Pbar Colliders"
Nucl. Instr. & Meth. A441 (366-373), 2000 - PS file (cdf-5003)
Returned torque wrench to Proton Assembly Building.