Thursday, August 21,  Notes

Present:	E. Barsotti, A. Byon, S. Chappa, R. DeMaat, G. Drake, 
		J. Elias, S. Hahn, S. Orr, K. Schuh, A. Tollestrup

Clock Digital Links
      Steve Chappa reported on studies of three different types of cable and two
different receivers.  The original chip, a 10H115, was not a true differential
receiver and was upgraded to a 10H116.  The best cable performance occurred for
a rigid, foam-dielectric, pair.  A slight annoyance with this cable is skew
between the two leads which produces a small common mode signal.  The
consequence is that proper ac termination of the shield is that much more
important.  Two suggestions were made:
1.) Reverse the leads at the driver to absolutely demonstrate that the common
mode signal is due to cable skew.
2.) Check for an ethernet version of the cable as the 100 MHz Ethernet Standard
has a very tight specification on cable skew.
Finally, Steve showed similar results for the new high common mode tolerant
receiver, the AD96685, which tolerates common mode offsets from -2.5 to +5
volts.  This appears to be the chip of choice.
      The working group then agreed to reverse recommendation #5 for the clock
which read:
Concern 5.  Noise and EMI problems arising from the cable braid grounding
configuration for the links between the clock fanout crates and the front end
crates.
Recommendation  Hard ground the cable shields for these links at the front end
crate and make provisions for a special termination (soft ground) at the fanout
crate.
The argument was that hard grounding the driver is best for returning any common
mode imbalance immediately to the driver chip itself.  The consequence is that a
proper, cable-specific, termination must be developed for the receiver at the
TRACER including an ac network for the cable shield.

SMD Data Cable Tests
      Gary Drake reported that the non-repeatability problems of the past few
weeks were traced to the pseudo-random bit generator not working properly.
After repairs, the original result that putting currents into the steel was the
worst source of noise in the CES was confirmed.  Gary has sent out a request to
each group to provide hardware for a test of their proposed digital link.  Ed
asked that Vince Pavlicek be added to the list for the FIB to Port Card link.

CMU Grounding
      John Elias reported on a conversation with Lee Holloway about the CMU
grounding.  At Urbana, the group had noise problems that were traced to rf
(radio stations) signals picked up by the negative high voltage connection and
coupled electrostatically from the high voltage board to the charge injection
board, and thus right onto the wires.  The shield of the high voltage cable is
interrupted by 300 ohms to break up the ground loop.  It was found that
bypassing this resistor with a capacitor such that RC was about 500 KHz fixed
the noise problem.
      The problem with this fix is clear; noise signals are injected into the
steel of the detector.  Detector grounding rule #2 states "Detector subsystems
shall not drive ac signal currents, analog or digital, through the steel of the
detector.".  Testing at B-Zero is clearly necessary using the actual high
voltage configuration planned for Run II, and the scope may have to be enlarged
from one wedge to a whole arch.  The "good" solution is to reduce the antenna
capabilities of the high voltage system to the point where the CMU operates
properly without shunting signals into the detector steel.

Power Supply Monitoring
      Stan Orr brought up the issue of links to the racks for power supply
monitoring, and links for DMACS as well.  Is copper going to be OK, or is fiber
the answer?  Fiber is not cheap, especially when configured as a fault tolerant,
self healing loop.  More information is needed, especially for DMACS.

New Business Items
1.)  What is the specification for noise in the CES, CPR, and CCR chambers?  An
analysis expert's help would be great.  Suggestions?
2.)  A meeting with the SVX group to update the situation is due as much has
transpired since the follow-up session back in March.

Next Meeting:  Thursday, September 11, 9:30