From novaes@fnal.gov Wed Aug 21 16:06:19 2002 Date: Wed, 21 Aug 2002 16:03:01 -0500 From: Sergio F. Novaes To: novaes@d0mino.fnal.gov Subject: FW: Abracos, Sergio __________________________________________________________________________ Sergio F. Novaes Fermilab E-mail: novaes@fnal.gov P.O. Box 500 Phone: (630) 840.2454 Mail Station 352 Cell: (630) 272.1819 Batavia, IL 60510-0500 Fax: (630) 840.8886 USA __________________________________________________________________________ -----Original Message----- From: Sergio F. Novaes [mailto:novaes@fnal.gov] Sent: Wednesday, August 21, 2002 3:48 PM To: Sergio F. Novaes Subject: Michael Strang Wednesday, July 03, 2002 Cal/FPD shifter instructions Here is the link to the instruction as they now exist. http://d0server1.fnal.gov/users/strang/web/fpd/documents/nonfpd_shifter.html Please let me know about any changes so they can be incorporated. __________________________________________________________________________ Victor Bodyagin Friday, July 05, 2002 pictures for oklahoma I've prepared three ps-files for your talk in Oklahoma. They are in /home/bodyagin/d0/fpd/offline/oklahoma directory of online cluster. Their names are: adc3.ps, planes.ps, and two-dim.ps Let me know if you face some problems. __________________________________________________________________________ Jorge Luis Barreto Tuesday, July 09, 2002 Data Analyse Summary We (M. Strang, J. Molina, V. Bodyagin, J. Barreto) met today at SDH to have updates in the data analyse issues. M. Strang: presented ADC distribution for D1 and D2 diffractive data and others plots that can be found at: All of my plots are located on d0mino in /prj_root/810/fp_1/strang/analyze/ps where it is broken down by runnumber or runlist name. adc* are the tdc and adc plots plot* are the other diagnostic plots Today I showed information from Run278 (adc plots) and Run281 (diagnostic plots). Plots for X-X' distributions for both detectors show a pattern that is similar to MC diffractive data, although shifted to lower X and X' values (a shift in the vertical direction for D1 and D2 ?). A final data sample for tracking purpose should be released soon. J. Molina - updated the list of bad channels for D1 and D2: The list of bad channels after taking out the extra set of cable in D1 is: D1 ADC Bad channel number (0 to 15) # of bad channels 0 1-2-4-7-13-15 6 1 - - 2 0-11-13 3 3 3 1 4 15 1 5 13 1 6 1-2-3 3 D2 7 12 1 8 16 1 9 0 1 10 6-9-15 3 11 1-2-4-6-9-10-14 7 12 15 1 13 0-15 2 --- total 31 So, we have in total 30 dead channels due not only to bad cable but to A/S and and MAPMTs also. He also showed a 47m flat cable where the weldings were ver poorly done, both at the ground and signal connections. Some time was spent discussing how to solve this issue of bad channels, with no good solution. J. Barreto: updated on MC issues related to the Proton ID document. __________________________________________________________________________ Victor Bodyagin Wednesday, July 10, 2002 AFE signals the matter is that 100mv is an abnormally huge signal for any PMT. Moreover, many things depend upon the light source. Was your light source calibrated for MIP? If not you could get any PMT current you wanted. The signals I observed were nearly an order of magintude smaller and that was after using A/S. Pierrick tried to get ADC distributions having bypassed A/S and did not get something in ADC spectra which could remind somebody Landau distribution. Let me deliver the numbers I have: 1. Gate width = 150nsec 2. 1 FERA ADC count is equal to 0.25 pC. 5. The load I used at the input of the scope was 46 Ohm (~50 Ohm). I did the measurements by comparison of pedestal shift in ADC distribution with the shift in the scope. Having used formula: Q = V*G/R, Where Q - charge measured by ADC, G - gate width, R - 46 Ohm resistor I got that 1 ADC count corresponds to 0.066 mv of DC coming from A/S, or 1mv of DC to 13 ADC counts. Typically pulses coming from MAPMT+A/S have ~50 nsec width. So the voltage of DC current averaged over 150 nsec interval for 100mv amplitude can be estimated by 30mv. It will give us 390 ADC counts. This value is well above most of our fiber signals, since Landau distribution is rather steep in the left part and has a very long and flat due to delta electrons (practically infinite!) tail in its right. Therefore, I'm really cencerned about this numbers. Anyway, I'm going to repeat and cricically examine my measurements. __________________________________________________________________________ Jorge Molina Thursday, July 11, 2002 DI bad channel list Hi, I made this afternoon the measurements of bad cables channels using the device made by Newton to look at the reflection of the wave created by a pulse generator. The list is: D1 Cable Bad chann # (0-15) Problem RC32-1 1 shorted at the connector in SCR 4 shorted at the connector in SCR 6 shorted at the connector in SCR 9-10 shorted between them at the conn in SCR 15 shorted at the connector in SCR RC32-2 - - RC32-3 11 shorted at PW08 13 shorted at PW08 RC32-4 6 RC32-5 15 shorted under the floor in SCR RC32-6 0 shorted under the floor in SCR 1 shorted under the floor in SCR 13 shorted at PW08 15 shorted at the junction in the tunnel RC32-7 1 shorted at the connector in SCR 3 shorted at the connector in SCR 8 shorted at the connector in SCR D2 FERA 8 0-1 shorted between them at PW08 14-15 shorted between them at PW08 FERA 9 6-7 shorted between them under the floor in SCR 15 shorted under the floor in SCR FERA - FERA 11 6 no reflexion -> broken wire 9 shorted under the floor in SCR 11 shorted under the floor in SCR 15 shorted at PW08 FERA 12 14 shorted under the floor in SCR FERA 13 15 shorted under the floor in SCR FERA 14 0 shorted under the floor in SCR So in total we have 31 bad cable channels. __________________________________________________________________________ Michael Strang Thursday, July 11, 2002 detector mapping, is it worthwhile As I mentioned in the video conference on Tuesday, the mapping of the detectors I performed shows that there should be some gaps in the coverage of the frames due to an increased offset between the primed and unprimed planes. At best the accuracy of the measurements performed were 10 microns but some of the gaps are as large as 100 microns. So, to see if these gaps actually exist in the data, I created segments plots for the large elastic sample. These plots can be found in d0mino area at: /prj_root/810/fp_1/strang/analyze/ps/el-pd3 in the file plotsel-pd_p1_d0_50.ps on the last page. To first convince myself that the segments plots were accurate, I determined what segments should be dead based on the dead channels. I found correspondence between the segments I would expect to be off and what is plotted. However, I do not see the holes that my mapping says should be there. I do, however, see evidence that the offset between primed and unprimed planes is greater or less than 0.54 as can be seen by the effect of 3 large segment peaks followed by a weak peak. This can be explained by a different offset ( 0.54 was chosen since it was roughly 2/3 of a fiber width, if we vary from this it forces one of the segments to be much smaller than the other three). So, now we need to decide if detector mapping is worthwhile. I still think I should measure the gross features of how the U and V frames are aligned with the X frame. This takes all of 20 minutes. I think we can get a reasonable mapping (better than the current 0.27, 0.26 one) if we use the dimensions of the grooves and fins from the drawing used to produce the original metal frames and hence the molds and put 0.78 (or 0.77) fibers in the center of each groove to reflect the size of the actual fibers used. Then, perhaps, play around with the offset a little to see what is needed to look like data? This is just a suggestion but we should decide if we want to continue with full mapping or not. __________________________________________________________________________ Harry Melanson Sunday, July 14, 2002 Updated Algorithms web page The Algorithms web page http://www-d0.fnal.gov/computing/algorithms/ has been updated. The following new items are available: * Update on p11 status * p11.09.00 CPU / event size statistics * p11.09.00 standard plots * http://www-d0.fnal.gov/computing/algorithms/status/p11.html * New general documentation * Thumbnail * Thumbnail ROOT tree * http://www-d0.fnal.gov/computing/algorithms/#doc * New ^ÓHow to^Ô documents * ^ÓHow to run RECO^Ô * ^ÓHow to use the software releases^Ô * http://www-d0.fnal.gov/computing/algorithms/howto/index.html __________________________________________________________________________ Victor Bodyagin Tuesday, July 16, 2002 a new DAQ version I've introduced some changes in DAQ in order to make life of our calorimeter guys easier. Now after the finish of first run one can start data collection by one LINUX command-line: ./daq_v03 To start the first run one has to fulfill all the steps as before but instead of: -> VME DaQ should be started as: ->tDaQ Having done all of the above once one can continue data taking LINUX command-line. Seemingly, it works OK. But I would rather prefer to test it more thoroughly... __________________________________________________________________________ Jorge Luis Barreto Tuesday, July 16, 2002 Status for workshop Ichep Hello, You can find the FPD slide at http://www-clued0.fnal.gov/~barreto/fpd/fpd_ichep02.ppt __________________________________________________________________________ Wagner de Paula Carvalho Wednesday, July 17, 2002 Equations I've just created a page containing links and some explanation about the C++ code that I used to create the trigger logic equations, following VHDL standards: http://alpha1.lafex.cbpf.br/~wagner/L1Code.html This is all standing alone code and has nothing to do with the D0 software structure. The several programs perform tasks like define segments, validate them, convert hit/segments maps into logic equations, etc. Just to remind you, there's another page which is complementary to the previous one: http://alpha1.lafex.cbpf.br/~wagner/l1equations.html Take a look on them and feel free to ask me any questions you may have! __________________________________________________________________________ Jorge Molina Wednesday, July 17, 2002 transparencies I put the transparencies (in .ppt) from today's talk at the halo meeting in my home page area at: http://www-d0.fnal.gov/~molina/fpd/separator.ppt __________________________________________________________________________ Victor Bodyagin Friday, July 19, 2002 some DAQ updates I've introduced a couple of new options into DAQ. Option No.1: Now one can run DAQ as follows: 1. In the beginning of data taking one can issue in host Linux machine: ./daq_v04 after that in VME window one has to issue a command which will load a version of VME daq which can operate with daq_v04: ->tDaQ And finally all DAQ will start. After data taking run in host machine stopped tDaQ will be in a waiting mode. To start next run one has to issue in host Linux window: ./daq_v04 and all things will go automatically. Note that to stop tDaQ (for instance for pedestal taking) one has to issue . I hope this will make calorimeter shifter's life a bit easier. Option No.2 Now the picture "det_planes" will pop up automatically. I've tested it thoroughly enough and it works. __________________________________________________________________________ Andrew Brandt Monday, July 22, 2002 notes from ops meeting 1) Too many mistakes recently (control lines, HV, etc.). People WILL start using the checklists again (even/especially experts) in order to remind them not to leave out any crucial tasks. Also Carlos will put up a note, like the one on the back of SCR door, as well as mount a white board to keep track of ongoing problems and comments. 2) Rates at home position should be recorded every store, including halo rates for all pots, p and pbar intensities, lum, and p+pbar halo. Halo task force will discuss how to use this rate information in separate meeting later this week. Should discuss with Vladimir and Victor how to automate collection of this information 3) Pot Rates during run not necessary to record due to title file, but recording unavailable information at start of run, like p+pbar intensities, lum+halos would be useful if pots are in. 4) We discussed turning tubes not being used to standby, but for this week let's keep them on to collect more information for halo studies. 5) Mike S. needs to immediately make P2 database current change so that it is done before access to allow for testing of P2 PMT's. Note if this is not done before access, the tubes should be checked one by one (following careful safety procedures on HV!!!!!), and then all of them should be left plugged in. 6) Carlos is preparing access teams and task lists. Everyone should send Carlos blackout times if they have any. I am available ~24x7 at 630 579-3086 (appartment), except tomorrow 10-8. HV should be turned off before access from standby. accessers should discuss controlled access procedures beforehand, and have a clear scheme prepared to avoid dropping interlocks. 7) We need to swap an interface board with A1I, and A1I should not be moved now since it is currently unreliabl. If we can test one first we should do so, otherwise, should swap a currently working but unused one (A2U or A2D for example?). Andrew sent note to Newton with question about problems and also A2I LVDT flakiness (is there a spare conditioner at Lab6?). 8) 4 dipole triggers were discussed a) diffractive b) elastic c) dipole+A2I (high t) --operating position determined at 24.9 for now Mike set limit for this and 22.9 for A1I. d) Junk dipole +VC+LM 2 triggers/store is probably a good amount, starting with 1-2 diffractive runs to understand conditions 9) We discussed trigger efficiencies. Need to get a pattern unit for this and consider modifying setup of camac crate. Jorge+Victor should map out a plan for this (perhaps Jorge can write a short note?) Jorge would like to do this before he leaves, and it might be possible, if we can get Pside functionality back. 10) Shifts were filled out and Carlos will provide email version which Mike will enter in calendar, along with last weeks shifts! Note no accel studies this week. __________________________________________________________________________ Newton Barros de Oliveira Wednesday, July 24, 2002 interface board or mux problem? On June shut down last days I saw this problem with A1 (I don´t remember exactly if was IN or OUT pot now). First, I checked the encoder, no problem found. Then I changed the IIB (and checked it, it was ok) but we still have the problem with a new IIB. I made a complete test of the system sending signals, locally, with the switches rule board to the MUX (disconnected from the Rack Monitor) and all pots worked perfectly. At that point the conclusion was the local movement system (MUX, IIBs, Drivers, Motors and encoders) was ok. I suspected the software was sending wrong signals so I tested this possibility. With Gilvan help seeing the strobe activity on the LEDs test box I asked Vladimir to move the pot to a desired position. What we saw : when the pot reached the desired position and stoped the software sent three more commands showed by the strobe light activity. It was not possible to verify what commands the software issued because it was in quick sucession. We expected not to see any command after the pot reaches the desired position ( one Stop command should be ok but not necessary) but why three more commands ?. I though it was a software bug and informed the situation to Vladimir. Now, you and Mike are talking about problems with the LVDT readings. As I was informed, the software sends the pot to Home if LVDT readings doesn´t agree with the desired programmed position. Maybe the extra commands issued by the software are related with this fact, I´m not sure. Important: Fluctuations on LVDT readings may be dued to bad contact on signal conditioner connectors. I found this problem several times in past. Some times the connector press the plastic wire insulator instead of the wire itself and we can´t see by eye, causing a bad contact. Instructions to repair this situation: Attention- There is live 127V inside signal conditioner box at wires connected to the fuse terminals. Unplug from the power outlet before working. 1- Open the signal conditioner box ( at right side of IIBs) removing four small screws from the perforated aluminum cover. 2- Localize the signal conditioner (light blue unit) corresponding to the pot with problems. From left to right (front view) the sequence is UP, DOWN, IN and OUT. 3- Identifie the green coaxial cable (there are four of them, one for each pot at the lower portion of the box) connected to the signal conditioner. 4- The cable runs to terminals 4 and 5 on signal conditioner. Center wire to terminal 4 and the mesh to terminal 5. You will see another wire connecting all terminals 5 on all signal conditioner too. Check to see if the wires are firmily connected. It is more probably the center wire is not well connected. This wire is very thin and if it is not in the right position the connector will press the plastic cover (clear plastic insulator) instead the wire. 5- Unscrew the terminal 5 screw several turns with a small screw driver to reposition the wire in the correct place. You can use a mirror and a flash light to see if the wire is well located inside the press connector. This operation is critical. The wire must be at the center of the connector to be correctly pressed. Tight the screw to hold the wire and check if it is well connected by pulling it on a delicate way. 6- Check the mesh wire too. 7- Plug the power cable on the outlet and ask some one to move the pot and check the LVDT readings. 8- If everything is ok, cover the box. If correcting LVDT problem doesn´t solve the movement problem, check the software logic to understand why it is sending the extra commands after the pot reaches the desired position. By now, I don´t believe on any problems with IIB or MUX since I tested in place and it was fine. Maintain me informed about the situation. __________________________________________________________________________ Helio da Motta Filho Wednesday, July 24, 2002 LVDT fluctuation I and Molina have looked at the LVDT's readout. A2IN shows very large fluctuations that must still be investigated. The interesting observation is that, for each castle, the fluctuations are highly correlated. So, for P2, UP, DOWN, BOTTOM LVDT all fluctuate together. The IN LVDT shows no fluctuation. P1's LVDTs show similar behavior. The dipole castles show no fluctuation. Fluctuations in P2 are correlated to fluctuations in A1. Fluctuations in A2 seem correlated to fluctuations in P1 (apart the fact that A2IN shows bigger fluctuations). Whatever is causing the fluctuation it should be in any commom part of the system. You can see the plots of the LVDT fluctuations at http://cns40.fnal.gov/userb/pic/capture/dzero/lvdt_p2_24jul_***.gif or lvdt_p1_24jul*** or lvdt_a1_24jul***.gif etc Do you have any idea about the cause of this? The fact that the fluctuations are time correlated can not be a coincidence. __________________________________________________________________________ Michael Strang Thursday, July 25, 2002 HV changed I was finally able to get hold of David Meder with LiAr and he ramped down their HV. I had Jorge Barreto ramp down our HV then I reset d0olctl45. So, the current setup is: DIL0 set for 2 tubes (1.5mA) A2L0 set for 1 tube (1mA) A1L0 set for 3 tubes (2mA) P1L0 set for 3 tubes (2mA) P2L0 set for 3 tubes (2mA) These were the values recommended by Carlos, Gilvan, Jorge M when they were studying the current draws in the last tunnel access. __________________________________________________________________________ FPD Group account Friday, July 26, 2002 changes in docs Several changes to documentation in Run Checklist Details: 1) If starting FPDGui.py there is no message "Initialization ok" and you see "pots are not in home, move them home first" message, press ok and after that move pots home by: swithing ON hardware control line and pressing "INIT" button. Then software will turn on automatically all amplifiers and drivers, will move pots which are not at home to home position and after that turn off amps/driver lines. 2) in DAQ section: tdaQ should be replaced by tDaQ 3) Rate Watcher: it's a cron job on d0ol19 and there is no need to start it because it may be confusing. So first two paragraphs should be removed. If you have to check that cron job is set up, just do > crontab -l and check that line starting with 00,05,.. in not commented out by # 4) All that business with setupdaq.sh,setupdaq3.sh is confusing, I'll talk to Victor how to make in simpler Vladimir __________________________________________________________________________ Carlos Avila Tuesday, July 30, 2002 access report ACCESS REPORT - July 30 2002 ============================ 1) P1OUT SIGNAL ============== It was found that the pmt base had a problem. We swapped the pmt base in the tunnel and Mike turned HV on and cheked in the scope for dark current signal coming from P1out signal and P1out was ok. 2) P2OUT SIGNAL ============= Found that HV connector for P2out was loose. We connected HV for P2out and then for P2U, Mike in the control room turned HV on for P2 and checked that the gui was able to drive current for the three pmts. Also dark current signnal was observed at SCR. 3) A1IN POSTION VERIFICATION PROBLEM ================================== Initially the mux box was replaced but the problem persisted. Then we swapped interface boards of A1D and A1in, found that the problem was in the interface board because problem moved to A1D. We replaced the interface board at the end and now A1in is moving withouth problems. 4) A2OUT FLAKY LVDT ================ Newton's instructins were followed to check the center wire to terminal 4 in the signal conditioner box. The center wire was well connected. The A2in has some jumping in the LVDT readings during the first mm of insertion, it seems to get stable when it keeps moving further in. 5) LMB POWER CABLE ================ Lmb power cable was connected. At this moment it still needs to be tested. 6) TLD BADGES ============= TLD badges were placed in the outside of the cartridge tops in the following order: Detector TLD-REFERENCE # =============== ================= A1U 0575492T A1O 0574514T A2in 0307438T A2U 0438131T (on top, no cartridge top there) D1 0279732T P1out 0586766T P1U 0307411T P2U 0569857T P2D 0573268T P2out 0728138T 7) A/S low voltage power supplies for the A side were taken out of the tunnel for testing. They will be connected back next acces. 8) A/S in D1 crate was replaced. Thanks to all three teams of people who helped in the access: team 1 - in the tunnel A side Victor, Jorge Barreto team2 - in the tunnel P side Jorge Molina , Carlos team3 - in the control room Mike Strang, Helio __________________________________________________________________________ Jorge Luis Barreto Wednesday, July 31, 2002 data analysis meeting summary Yesterday, at 3pm we had our weekly meeting at SDH. 1) Short rapport on previous week data taken: J. Molina presented 2D plots from last week elastic and diffractive data taken with dipoles spectrometers. Compared with previous plots on same type of data, D1 presents an improvement wrt spurious hits, while D2 shows same pattern as before. TDC studies need to be done to clarify this issue. V. Bodyagin presented plots from D1 ADC distribution of L0 signals before (not biased) and after (biased by discriminator), that show a good signal efficiency, after triggering, above 100 counts. 2) Plans for next data taken: C. Avila presented the trigger list to be used in next shifts this week: 1) En of store study for P1out and P2out 2) Runs to take: a) D1.D2.DCLK.OLVNbar.LMSbar.VCS ---> low mass diffractive b) D1.D2.DCLK.OLVNbar.OLVSbar ---> dipole elastic c) D1.D2.DCLK.OLVN.OLVS ---> junk d) D1.D2.DCLK.A1I.A2I.OLVNbar.OLVS ---> High t diffractive e) D1.D2.DCLK.A1I.A2I.OLVNbar.OLVsbar.P1O.P2O ---> High t elastic f) D1.D2.DCLK.OLVNbar.OLVSbar.P1O.P2O ---> High t elastic_P 3) We want to take runs of about 100K events. 2 runs per trigger. 3) Data Analysis update: M. Strang sent a short update on diffractive Dipole data analysis: The TDC cut needs to be tighter up, to remove halo contamination in data sample. With the contamination, and from what I could see, for diffractive data, the % of total events in each sample breakdown was: cut before trigger change after trigger change tdc 44.8 39.3 mult1 12.7 12.1 mult2 3.6 3.6 all 3.1 3.1 So, in the end, the same fraction of events pass the cut before the change in trigger timing as after. 4) MC update: J. Barreto presented new MC acceptance values for quadrupole and dipole spectrometers (8Sig), using as an input a t exponential f(t)=exp(-b*t) distribution, with b = 4.0 (xi=0) b = 4.5 - 0.5ln(xi^^-1) (xi#0) Wrt the flat t distribution, the new results show that most of the cross-section lies on in the intervals: 0.025<=xi<=0.07 and 0= 20 out of 24 (83%) FFs --> 3281 out of 6,912 (47%) LUTs --> 3821 out of 13,824 (27%) In our case, there is not a general number for this measurement. However, I would say that the utilization of the XCV600 devices for these components is around of 35 %. It excluding RAMs use. __________________________________________________________________________ Helio da Motta Filho Tuesday, August 20, 2002 parts for Gentil I have been at NWA and have seen the parts. I have called LNLS (Campinas) and have gotten the data for the cam follower The piece was made by SKF, model number KR 16 (the thread is metric). The sleeve is custom made. We do not have an extra unit in Campinas. As long as it works (it works now) it should be OK. They have sugested cleaning it well with acetone. My understanding is that acetone is not an option here but James has told me that he can probably do it with some other solution. He was trying to have safety aproval, or something like that, to have the piece immersed in the solution. The cam follower is a very unexpensive piece but you may have dificulty getting it here because it is metric. Let me know if you have any problem. __________________________________________________________________________ Mario Vaz Tuesday, August 20, 2002 LUT - Sender and Receiver requirements Good news for he FPD PU L1 trigger implemented with wide segments for high t and fine segments for low t : it fits into Xilinx Virtex 600 : Number of errors: 0 Number of warnings: 0 Number of Slices: 3,917 out of 6,912 56% Number of Slices containing unrelated logic: 0 out of 3,917 0% Number of Slice Flip Flops: 725 out of 13,824 5% Number of 4 input LUTs: 7,773 out of 13,824 56% Number of bonded IOBs: 228 out of 404 56% Number of GCLKs: 1 out of 4 25% Number of GCLKIOBs: 1 out of 4 25% Total equivalent gate count for design: 52,438 Additional JTAG gate count for IOBs: 10,992 Adding the numbers from Ricardo Ramirez for the Sender and Receiver firmware : > RAMs --> 20 out of 24 (83%) > FFs --> 3281 out of 6,912 (47%) > LUTs --> 3821 out of 13,824 (27%) > However, I would say that the utilization of the XCV600 devices for > these components is around of 35 %. It excluding RAMs use. gives a total utilization of 11594 LUTs (13824 available) - 84 % and 4006 FlipFlops (13824 available) - 29 % of a XCV600 FPGA. It is tight for LUTs but it seems feasible. Ricardo, please try to put together the Sender and Receiver code and our code, that is availalbe at http://d0br1.lafex.cbpf.br/~mario/L1_PU/BlkP.vhd (top level) http://d0br1.lafex.cbpf.br/~mario/L1_PU/segf.vhd (fine segments) http://d0br1.lafex.cbpf.br/~mario/L1_PU/segw.vhd (wide segments) http://d0br1.lafex.cbpf.br/~mario/L1_PU/PU_t-low_xsi-low_trigger.vhd http://d0br1.lafex.cbpf.br/~mario/L1_PU/PU_t-low_xsi-high_trigger.vhd http://d0br1.lafex.cbpf.br/~mario/L1_PU/PU_t-high_trigger.vhd Working now on PI and DI again. __________________________________________________________________________ Michael Strang Wednesday, August 21, 2002 pedestals From: "Jorge Molina" August 21, 2002 ]We were looking with Gilvan the pedestal distributions for a lot of ]runs and we found funny things about the numbers that appear in the ]file. ] ]Lets take for example the pedestal file y02m08d03t1628.txt ]corresponding to runs 317-319. ]Looking there we have the two initial numbers ] 0 0 0 442.07 1.58 444 ] 1 0 1 523.60 1.89 525 ] ]After the unpacking we observe inside the ped file run317.ped the ]numbers: ]46 ]31 The run*.ped file is created with my kumac pedsfrwrd or pedsbkwrd depending on how the channels need to be unpacked. The ordering in the run*.ped file is different than the ordering in the *.txt file. The run*.ped file is only applicable if you are using the analysis code that I came up with. If you are using your own code, you will need to transform the *.txt file to whatever form you need. For reference, the raw ADC ped file (the *.txt file) has things in the order (from top to bottom): UP1(1:16), M1(U1(17:20),UP1(17:20),V1(17:20),VP1(17:20)), VP1(1:16), V1(1:16), XP1(1:16), X1(1:16), U1(1:16) and same for detector 2 then l0(1:16) for forward mapping and: UP1(16:1), M1(VP1(20:17),V1(20:17),UP1(20:17),U1(20:17)), VP1(16:1), V1(16:1), XP1(16:1), X1(16:1), U1(16:1) and same for detector 2 then l0(1:16) for reverse mapping. In my remapped ADC ped file (the run*.ped file) first, it contains as the pedestal the mean + 3 * sig + 2 and it is reordered (from top to bottom): U1(1:20), UP1(1:20), X1(1:20), XP1(1:20), V1(1:20), VP1(1:20) and same for detector 2 then l0(1:16) where the mapping has been taken care depending on which kumac was used to generate the file. You can look in the pedsfrwrd (the one being used) and pedsbkwrd for the mapping correspondence between the two files. ]In case we use Victor's program I can see the distribution (without ]pedestal subtraction) that the pedestals are around these numbers ]showed in the last column (444 and 525). Then when we opened the ntuple ](without ped aubtraction) we see in the distribution that the pedestals ]peaks are around 46 and 31 for the respectives channels. In Victor's programs, he is not remapping the ADC info so the *.txt file should be used. In the ntuple, the ADCs have been reordered by frame and channel so the run*.ped file should be used if you are using my analysis code. ]What does it means?, according to the mapping these two numbers ]corresponds to fibers up1(1) and up1(2) but what is the operation in ]the middle (during the unpacking we think - but we could not see any ]operation in the unpacking program) to make them change in the scales? It's because Victor's ADCs haven't been remapped so you are comparing apples to oranges. ]Another issue is that while in Victor's program before the unpacking ]the value for fiber up1(2) was bigger than for up1(1) while after the ]unpacking now up1(1) is bigger than up1(2). Once again, it is because of the mapping. Victor's code is setup so we can easily debug the hardware. My code is setup to put things into the detector system. IOne other thing about the pedestals. In the database, I record what ped file to use based on the y*m*d*t*.txt format. When I transfer the ped files over to d0mino, I rename this *.txt file into run*.ped but if it is in the ped directory with the unpacked data, it has not been remapped. I then create new run*.ped files in my analysis directory that have been remapped. You can easily tell the difference because the unmapped ped file has 6 columns per row whereas the mapped ped file only has one column.