TRD electronics Flight Software

1
TRD electronicsFlight Software

• Wim de Boer, Chan Hoon Chung, Florian Hauler,
• Mike Schmanau, Andreas Sabellek, Georg Schwering
• IEKP - Universität Karlsruhe (TH)
• RWTH-Aachen I

2
Outline

• Data Reduction Algorithm
• Calibration
• Slow control
• Outlook

3
Data Reduction Algorithm (1)

• Functionality Rejects amplitudes close to zero.
• Algorithm 1. Read event amplitudes 2. Subtract
  pedestal from amplitudes 3. Calculate and
  subtract common mode from amplitudes. 4.
  Perform zero-suppression (amplitudelt3 s) 5.
  Write compressed event.
• Current Timing 14 448 ops 3 64 ops 6 14
  ops 6548ops0.02 us/ops 131
  us.
• However, algorithm not finalized, small changes
  expected.

4
Data Reduction Algorithm (2)

• Raw event size (physics) 5376 (real 5248)
  channels x 2 bytes 10.5 kB
• Zero-suppressed event size (physics) TRD has
  twenty layers Ideally 20 hits / track. For the
  3 sigma cut, we expect 0.3 more amplitudes
  16Delta electrons 2Interactions with detector
  material 5 (average from interaction hits (10))
• Format is channel number amplitude
  valueSize is (1 word 1 word)43 86 words
  172 bytes
• Will be verified from data of cosmics test of
  full TRD.
• Some more data has to be considered, which is
  needed for event administration, e.g. amount of
  hit channels, node status, 12 x 2 words 48
  bytes.

5
Pedestal / Noise Calibration (1)

• Functionality Calculates pedestals and noise.
  (with elements from Claude Zurbach)
• Algorithm 1. Calculate pedestals from a
  sample of 512 events. 2. Calculate raw channel
  noise from a sample of 512 events. 3. Flag dead
  or noisy channels. 4. Calculate final noise
  taking into account common mode correction,
  dead or noisy strips from a sample of 512
  events
• Time Needed approx 600 ms, depending on amount
  of idle cmds.
• Frequency of repetitions Temperature variations
  modify the value of the pedestals. Frequent
  calibration will be required depending on the
  current temperature change.

6
Pedestal / Noise Calibration (2)

• Pedestal shift is composed of1. pedestal shift
  due to temperature variations on the
  frontends. shift is approx. 1 ADC per 10C
  temperature variations between -25C and
  35C
• 2. pedestal shift due to temperature variations
  in U-Crate, resulting in voltage shifts in
  frontends. shift is approx. 1.25 ADC per 10C
  (preliminary) temperature variations between
  -20C and 50C

7
Slow Control HV
Currently Slow Control is beeing performed using
nice GUI programmes written by A. Lebedev. Each
click results in a 32 bit Lecroy command and a 32
bit Lecroy answer, sent and received via
AMSWire. To properly initialize and ramp 12 UHVG
boards with 7 channels each at least 12 x 7 x 4
336 Lecroy commands need to be issued.(168 per
crate) In order to read out the full UHVG
status about 12 x 7 x 20 1680 Lecroy commands
are required. (840 per crate)
8
Slow Control UPSFE
To initialize 6 UPSFE boards by switching off
redundant hardware at least 6 x 4 24 Lecroy
commands need to be issued. (12 per crate) In
order to read out the full UPSFE status about 6
x 11 66 Lecroy commands are required. (33 per
crate)
9
Slow Control S9011AU
To initialize 2 S9011 boards at least 2 x 2 4
Lecroy commands need to be issued. (2 per
crate) In order to read out the full S9011
status about 4 x 4 16 Lecroy commands are
required. (8 per crate)
10
Slow Control

• Standard procedures like HV ramping commands and
  initialization commands can be combined into
  scripts, which will be executed by JMDC on
  request, thus reducing communication with ground.
  (maybe similar to the scripts which we used for
  the ESS and TVT)
• A SC command is composed of the Lecroy command
  embedded in an AMSWire command 2E1D writes or
  reads Lecroy registers2E5D initializes the
  Lecroy link
• To read out or set a slow control register, a
  time of 150 us is needed. Reading out all the
  slow control status registers amounts to 876
  commands, which takes a time of 131ms.

11
QList

• Slow Control should not interfere with data
  taking. Therefore the QList has been introduced -
  a command table defining the commands which
  should be executed in defined time intervals. The
  execution is performed in a way which does not
  cut down the calculation time available for event
  building. The Lecroy answers are stored and can
  be read out on request.
• So far no experience with the QLIST in the U
  system has been gathered. But the definition of
  such a QList will be done as soon as possible.
• Commands in the QList will treat temperature
  readout, UHVG ADC readout of current and voltage,
  Status of UHVG, S9011 and UPSFE.

12
Outlook

• In the next time, efforts for the development of
  the data reduction and calibration will be
  reinforced.
• A QList will be written.
• Test the complete code in a cosmics test of the
  full or half octagon, maybe end of June/beginning
  of July.

13
Power up/Failure recovery
Power On (if conditions are safe)
All primary and secondary hardware is on
JMDC (or from ground control) Verify
UPD/U-crate hardware status
OK! Switch off redundant hardware
Not OK! gtManual recovery in control center.or
if problem is known and JMDC has explicit
instructions correct problem automatically
Gas ok?
HV ramping
data taking and slow control monitoring (JMDC)
Problem detection
Known problem Automatic treatment in JMDC
Unknown problem Manual failure recovery in
control center
14
TRD(U)-Electronics Overviewafter weight reduction
Ucrate TRD electronic crate UBP TRD
backplane UPD TRD power distribution box
V2
V2
UPSFE TRD power supply for front end UDR TRD
data reduction board JINF data concentrator and
link to higher DAQ for TRD UHVG TRD high
voltage generator UFE TRD front end UTE TRD
tube end UHVD TRD high voltage distributor
removed USCM USCM functionality covered partly by
JINFV2