
This is a summary of the issues that were discussed at the Firefox RAM
board design review.

.DL
.LI
Hai Vo-Ba (x3247) was recommended as a person to talk to about testing the
memory controller.  He worked on testing of the Focus MC.
.LI
I need to figure out how to tell how many banks are on the board and make
sure that the boards can be configured continuously.
.LI
Rick Greer questioned if we need an extensive power-up self-test.
.LI
Tom Myer suggested trying inductors in the spice model to replace the RC
filter.
.LI
Tom also said the the trace impedance could be higher than 60 ohms.
.LI
Craig suggested that we may need to turn off the error correction to
determine what chips are bad when there are two hard errors.  This was
also suggested as a needed feature to QA the correction circuit.

Turning off the ECC is not necessary to determine what RAM chips are bad
but it would make it much easier.  The following procedure could be used
to determine the bad bits.
.AL
.LI
Use the syndrome word to determine what bits were changed by the controller.
This requires about a 40 byte table.
.LI
Change all the bits back to what they were when they were read from the array.
.LI
Compare this to the word that was read.  All differences are errors.
.LI
Use the errors to determine what syndrome bits should
have been set.  (use the same 40 byte table).
.LI
Compare this to the syndrome word that the controller returned.  All differences
are errors.
.LE
.LE

