Status of HV power supply the Brick

1
Status of HV power supply- the Brick -

• Marco Incagli - INFN Pisa

2
Update since last TEM at CERN

• The HV power supply will be built in Italy
• Problems in HV controller as of CERN meeting of
  feb 7/8 2005
• the HV controller has a potential problem in the
  VHDL code
• This problem, observed in the code by a CAEN
  engineer, by Franco Spinella and by Vladimir
  Koutsenko, shows up in the simulation, but it was
  not verified experimentally
• More tests were asked in order to check if the
  problem is real or not

3
Tests in thermal chamber

• The brick was put in a thermal chamber and cycled
  from -40o to 60o no anomalous behaviour observed

4
The potential problem depends upon the relative
timing between the internal HV clock, the LeCroy
(slow control) clock and the LeCroy data
HV clock (20MHz) LeCroy clock (100-500
kHz) LeCroy data
100 msec
50 nsec
Expanded view
5
Forcing the error

• Simulation says that the brick should not work if
  the slow control clock arrives later than the
  data
• To force the error, we had to artificially delay
  the LeCroy clock signal with respect to the
  LeCroy data signal by adding 25 meters (!) of
  twisted cable on the clock wires.

6

• In this extreme configuration, the brick does not
  work, as expected from simulation!

Return word (RX) is missing
CLOCK
DATA
96 nsec
7
Safety margins

• What prevents LeCroy clock from coming after
  LeCroy data (condition for the error to appear)?
• The data signal from JINF arrives to HV 6-8 nsec
  after the LeCroy clock . Always true?
• The internal path in the LeCroy FPGA delays data
  wrt clock by 272psec . Dangerous!
• CAEN inserted (without notifying us!) a flipflop
  that delays data signal by one internal clock
  cicle (50nsec) in the new version we have asked
  to add another flipflop to increase the safety
  margin .

8
Conclusions - I

• The HV controller VHDL code, although not
  formally perfect, works correctly in the current
  configuration thanks to the extra FlipFlop
• The controller is now in CAEN to be modified
  adding a second safety FF and including some
  improvements in the slow control commands
  suggested by me and by A.L.
• Next week the controller will be back in Pisa for
  new tests. A test period at CERN in march will be
  planned.
• Our collegues from Annecy and Madrid (RICH) have
  asked to have the brick for additional tests.

9
Conclusions - II

• If no other problems show up, the brick currently
  in Pisa is considered to be the QM and will be
  tested for Space Qualification in Italy (Terni)
  or in Spain (Madrid) as soon as the mechanics is
  ready