Summary of meeting on AFE8 status
                                                  Feb. 27, 2001
                                                  scribe: P.Grannis

Fred Borcherding summarized the status of 10 prototype boards (RevA);
one is fully operational, three nearly ready, one needing MCMs,
one set aside wtih some damage to fix and four more being stuffed.
Parts are available to stuff all 10 boards. 

The Gerber plots for the production of 146 bare boards went to
Sanmina Feb. 23.  Sanmina has received and is checking compatibility.
DZero has an agreement with Sanmina that they will notify us before
proceeding to production (we would have to be prepared to stop them if
there is cause; otherwise they continue.)  Sanmina will do the boards
in their R&D facility.

Fred outlined an order for 5 bare boards to be done on fast turnaround
by a special vendor.  Cost is high ($7500 for two boards).  We don't
see how to stuff them rapidly, so the tests would be restricted to
power/ground tests.   We did not see the utility of pursuing this.

An order of 28 boards (RevC group 3) is imagined to allow operation of
the full CFT stereo and axial before AFE12 arrival.  Sanmina has
agreed that they can add the bare board order, probably tacking them
onto the standard 146 board run.  Parts procurement for the extra 28
boards in underway but may not be complete when stuffing of the 146 begins.
D0 may have to stuff the cards or another vendor employed.

We discussed the two four week cycles for stuffing RHB and LHBs after
the bare boards are available Mar. 23.  This 8 week cycle is the
most painful for DZero and any way that we can find to speed this
process up should be explored.  We have not had serious dialog with
the technical personnel at Sanmina:
      +---------------------------------------------------+
      | We (John Anderson + Marvin/Fred?) should visit    |
      | Sanmina to investigate ways to speed up stuffing. |
      | Getting some LHB and RHB together early would     |
      | benefit DZero.  Earlier spots in their queue      |
      | would obviously help.  Can incentives help?       |
      +---------------------------------------------------+

Fred and Marvin summarized the state of prototype AFE8 tests
in Lab 3.   The SVX readout errors were fixed by tweaking the
timing of control signals.  Despite not understanding why the
changes were needed, operation is much more reliable.  Further tests
are needed to quantify error rates.

Single photopeaks are now observed after some timing mods.  The
fitting and modelling of the spectra is not complete, but qualitative
analysis suggests that we are within 25% of the predicted best
performance, based on stereo board results and bandwidth calculations
for the AFE.  Noise performance thus seems 'good'.
      +----------------------------------------------------+
      | We would like a more definitive statement of the   |
      | noise performance; clear distinction of where we   |
      | are quoting the charges (input to MCM, input to    |
      | SIFT, input to SVX) and statistic (sigma, 0%-100%, |
      | ... is needed.  We also need written statements    |
      | of the fC/count  and fC/pe for stated VLPC gain.   |
      +----------------------------------------------------+

Discussion outlined the tests remaining to be done; this is
not yet firm.  We need the test with cryo controls operating during
data collection;  operation of LHB and RHB together;  cross talk
studies for the digital signals;   some measurement of longer term
stabilities;  studies of the full complement of channels on an AFE
board;  and tests with lower gain cassettes.  The cryogenic control,
photopeak, and digital cross talk tests will proceed while the
second board is qualified for the LHB/RBH test.

Fred outlined the plans for test stands for production AFEs.
He expects several simple stations for visual tests;  two stands for
power, 1553 operation (with PC control), one 'AFE test stand'
where charge injection, tests of the bias and cryo voltages can
be done, and the Lab 3 cassette operation with up to 8 boards at a
time can be done.   Times/board at each stand are not well understood,
so the flow of the test cycle is uncertain.  We expect that first
boards will need to go through the full chain of tests.  If Lab 3
tests of these do not show large failure rates, we may be able to
shorten the production board tests and install in the platform after
the AFE test procedure.
      +--------------------------------------------------+
      | The procedures, times, resources needed for test |
      | of production AFEs needs more work.              |
      +--------------------------------------------------+

Mike Utes is now working mostly on AFE12 with John Anderson focussing
on AFE8.  We hope to have an AFE12 prototype by mid April.  Likely,
the virtual SVX discriminator readouts will not be present.  (John
Anderson notes the possibility of reading this information out on 1553.)
Projections would have production AFE12's by end of summer.

We will face a decision point (mid May) on whether to continue
with the extra AFE8 boards to allow operation of the full CFT.
The problem is MCM availability, and the failure rate if we try to
recycle them from one board to another.

      +--------------------------------------------------+
      | We need better counts of MCMs and estimates      |
      | of rejects and failure recoveries.               |
      +--------------------------------------------------+

The decision will be informed by the prognosis for AFE12.