Minutes of Interface Card Working Group 4/1/99
----------------------------------------------
Attendees: Utes, DeMarteau, Juste, Matulik, Johnson, Zverev,
Leflat, Fogleson, Sidwell, Rapidis, Sobering (via telephone).

Layout: Foglesong showed new drawings of the crates and power
supplies with the IF card crates such that backplane is down,
and cards face up. This has advantages for cable routing since
output cables exit up towards calorimeter, and input sequencer
cables come in from below. Space needs to be allowed for bends.
If fans are used for cooling, they would need to blow 
horizontally.

Marvin suggested that after drawings have been edited a bit,
that we mockup the system. This could be done with plywood, and
aluminum panels. We'd want to mount enough cable connectors do
a meaningful layout. 

Power: power supplies would be water cooled, and magnetically
shielded. 

SVX power requirements: at the SVX, the operating range for
AVDD is 4.95 to 5.20v. We need low impedance power paths.

Bias voltage control: longish discussion. Petros urged us to 
contact CAEN, to price a new system which would allow individual
control and monitoring. Sidwell suggested we try to reach an
SMT consensus in ~two weeks, since this item is holding up the
board definition.

Sereral possible relays exist which have off impedance of 
> 10^10 ohms, 200-400 max voltage, and which can switch either
+ or - voltage. 

Current limiting: Sasha Leflat showed a circuit which he has
tested. This resettable "fuse", has a comparator looking at
the voltage out from the e.g. AVDD current monitor. See figures. 
If this voltage exceeds a threshhold then AV_FALSE is set.
This adds one comparator per 
voltage to the board layout, and 3 lines to each PLD. He was
unable to melt wire bonds (size?) with currents up to 2A.
Since the current board layout has two comparators per channel
which can probably be eliminated, the impact on space is
probably acceptable.

Wire bonds for power: Petros volunteered to contact Ray Hall
and Promex about the feasibility of adding another 1 mil 
bond for power, or doubling the wire size to 2 mil. 2 mil is
already used for many SVX bonds along the top according to
Matulik. Bill Cooper raised the question of retrofitting 
existing HDIs, and expressed a preferency for two 1-mil
bonds, rather than one 2 mil, from a reliability viewpoint.

Impedance matching on Rev 2 card: Evgeny showed results from
studying crosstalk and signals. He is able to reduce crosstalk
to < 0.3v, and improve signal shape substantially. Proposal is:
  a) remove input series terminator to IF card (now 43 ohms)
      These are e.g. R67-74.
  b) add 82 ohm (Z of seq cable) on sequencer to ground
  c) remove 150 ohm resistor to ground on far side of transceiver
      (R47-54)
  d) replace c) with Ge diode (where to find more?)
  e) replace series resistor on "out" side of transceiver,
    now 43 ohms, with Zc=75 ohms (R59-66)
  Mike Utes was intrigued and wishes to pursue this. 

Jobs:

Foglesong, MJ 1) Update drawings, and investigate cable runs and space issues.
 RAS          2) Cooling and power ratings (SVX and IF)
 RAS,EDL      3) Study existing sequencer back plane schematics to
                 check if this can be recycled for the IF cards.
 MM           4) Test photovoltaic relays (several choices) 
 RAS          5) HV control and monitoring: writeup existing plans,
      and advertize, and seek SMT consensus. Two week time 
      scale. 
 ??           6) Specify and price HV system from CAEN. They have built
                 systems for BaBar and CDF recently.
 Berestov?    7) Find source of Ge diodes, esp. surface mount.
 EDL          8) Start stuffing Rev 2 IF cards by the week of 4/12/99
                Check with Zverev & leflat re proposed changes to see which
                can be implemented.
 MM           9) Start stuffing connector adapter modules- after impedance
               tests are through?