Action items from WALTA wishlist and Sept. 18 phone conference. With commentary, of course. The original list is reproduced at the end of this note. Compiled by Rik Gran All items are firmware bugfixes or features for Terry and Sten except as noted. Very soon (almost all bugfixes) --------- RB reset sets scalars to "1" and not "0". Either a new command "RS" should do the same without changing CE and CD Or RB should be modified to not change CE or CD. Status flag shows '@' instead of '9' if bits 0 and 3 are set. Fix the TMC delay setting bug [Hansen, help from Gran] comment: Sten is looking at it. You can see this with a pulser. Locate and fix the DS scalar output bug comment: the DS command does not participate in the interrupt/polling scheme that is implemented for most other commands. Probably it should. Until this is changed users might be advised to stop the TMC before requesting the scalars, or live with garbled data. Final note, Rik thinks there might be an additional corruption of the actual data in the register, separate from garbling when the register is pushed out to the PC. Locate and fix the first-line-flag late or absent bug. comment: The fermilab folks have not seen this with their pulsers, but it appears in both the WALTA and CROP data under all conditions and all different DAQ programs (labview, linux, hyperterm). I think we [WALTA] may have triggered this with pulsers by putting pulses at the very edges of the gate width and TMC delay values, but I'm not sure yet. Locate and fix the burst of lines that contain no edge data. comment: both WALTA and CROP see this. it might be due to EM interference, but we wonder at the fact that such lines appear at all. I worry that it is a symptom of another problem. Get a hold of CASA thesis with details about the hardware we are using. [Wilkes] Complete the several hardware documents that are in progress. Wilkes is preparing a "users manual" which is about to see the light of day. Gran has additions to two fermilab documents. Berns has an updated document (done?) about the data format. A very serious warning about changing the baudrate should be included. Guidance about the firmware update plan, now that there are 50 cards out in the field. I see one important firmware bugfix release that does not change the data format and some significant firmware changes and feature additions that do affect the data format. How best to implement them for our users? [Tom Jordan] When we get a chance (firmware and data format changes) --------------------- WALTA/CROP? folks should check their PMT/HV systems for ground loops. [Berns] Add another 16 bit internal counter with the lower one bit overlapping the highest bit in the 32 bit counter. The DS command response should be changed to a single line: DS 0123ABCD 0123ABCD 0123ABCD 0123ABCD 0123ABCD 4444 33333333 where the 0123ABCD are the 32 big singles counters (4 channels + coincidence) and the 4444 and 33333333 are the high 16-bit and low 32-bit counters latched approximately when the DS command was requested. Some error flags should come through even with no GPS data. When there is no GPS data, the high order 16-bit timestamp should be added to the data string instead of the absent GPS information. All modes can be saved to flash so that the card starts up in the same configuration it was left. This will require a SAVE command and a RESTORE (to factory defaults) command. The modes we are particularly concerned about are the Baudrate, CE/CD on startup, GPS info on or not, but really anything that is a "mode", including default values of control registers in the CPLD, TMC, and MC, should be saved in this way. comment: already partly implemented. Change baudrate on the fly. comment: implemented in 9/9/03 testing version. General comment about the following items: Some of us use DAQ programs, some use hyperterm (Tom, who uses what?). And all of us analyze our data offline. Since we have several modes that can be toggled on and off, and since software designers probably want to accomodate all these modes, maybe it is not really important to offer such changes. This is the start of an argument that these wishes should be striken from the list. "TP" command to toggle the prompt on and off. Output of some registers should be unique, not @00 And they should be invalid hex values. R0 R1 R2 R3 and T0 T1 T2 T3 for cpld and tmc registers GPS 1PPS logged separately and automatically every second. comment: Danger! this opens a can of data format worms. In a high-rate environment this saves space because you would not need to attach GPS info to every data line. In a low-rate environment, it enables you to unambiguously catch and correct any GPS glitch because you have a complete record of every 1PPS pulse. It could be turned off for people who do not need the timing information at all. It is the most significant change in the data format proposed here, as significant as the move to multi-line events. [Berns, Hansen, Kiper] Eventually (Qnet DAQ model 3.0 with major hardware changes) ---------- Even more detailed help page that can be queried by hyperterm users. comment: give them the text and they can program it. Threshhold voltage monitor via analog inputs. HV monitor input. comment: discussion about HV control system for CW tubes. USB connection comment: we worry about cable lengths. ethernet was nixed. Roll your own TMC with a GHz FPGA which may soon be available. General ------- If we make one change in the data or error flag formats, we have the opportunity to consider other changes at the same time. Better to suggest improvements now. Nevermind --------- Last line in event flag. comment: this issue should be dealt with in offline analysis software. ------------------------------------------------------------------------- ------------------------------------------------------------------------- Wishlist for Quarknet DAQ card. This revision Summer 2003 Be careful what you wish for and watch out for feature-creep. Probably we don't _really_ want all of these things. NOT ORDERED by preference. Easy bug fixes -------------- Setting TMC delay to 50 sets channel 1 to 50, channels 2-4 to 44. RB reset has a bug that resets the scalars to "1" not "0" status flag shows '@' instead '9' if bits 0 and 3 are set If there is no gps data, the flags (including fifo-not-empty) are not read. Three different kinds of glitches seen in UW/WALTA data, detailed in www.phys.washington.edu/~gran/WALTA/DAQdocuments.pdf 1. scaler output is garbled or glitchy 2. first-line-flag is sometimes late or absent 3. occasionally see bursts of lines that contain no edge data Card Behaviour Open Questions ----------------------------- If the scintillator/PMT apparently produces double-pulses for a significant fraction of the events (true for some CASA setups), does this affect the number of pulses in the scalar counters as well? What is the dead time following each event readout? Is the gate generated by the CPLD like a retriggerable one-shot or not? Is the coincidence generated by the CPLD "retriggerable"? (yes...) Feature Requests ---------------- Default start up is with TMC disabled. The user needs to explicitly enable operation. Save settings into Flash ROM, control registers so that things recover to the same settings after poweroff or reset. baudrate selection, control registers--coincidence, gate width input select and all things that can be enabled or disabled. [comment: already a feature in firmware version 9/9/03 except the enable/disable features, i.e. CD/CE, NA, ST] Another help page that can be queried for information about the control registers from within hyperterm. Change baudrate on the fly, or with a reset (not with reprogramming) 19200 38400 57600 115000 [comment: also a feature of firmware version 9/9/03] GPS logged separately and automatically every second. Add filter to catch and correct 1PPS glitches (Hans). DS written out automatically at some programmable interval. DS currently reports register 05 which is the current 1PPS counter. But if no GPS information, it would be great if DS reports some kind of time information, like the value of the counter when the DS was requested/sent. **The following item we do not really want.** **NO** DS reports 2-fold, 3-fold, and 4-fold coincidences through internal **NO** counters instead of through analysis of data. Make sure we like the short vs. long output. only turn off GPS to make things short. Is this enough? Add a "last line in event" to complement "first line" for easier analysis of streaming data. **The following item we do not really want unless the above is possible** **NO** Change "fifo not empty" flag to "fifo full" in the error flags. **NO** This is instead of having to query for a fifo_full. Do we like our current set of error flags? Some error flags should come through even if GPS is turned off. Resets that affect only certain subsystems: example RS resets only the scalars Turn off the ">" prompt if not using hyperterm. Or turn it off anyway. The output of some of the register values all start with @00 regardless of which 00 register it is. Can this text be made more unique? Hardware Features ----------------- High Voltage monitor inputs. USB or ethernet connection Threhhold monitor readout via the analog inputs. PC-DAQ or Labview software -------------------------- Spedometer with information to tell user of too-high rate.