Re: Restoring a NorthStar Horizon, problems with SRAM board

On May 18, 12:08 pm, bigb...@xxxxxxxxxxx wrote:
On 17-May-2007, lync...@xxxxxxxxx wrote:
I am continuing the debug process from last night. I verified the 2
MHz CLOCK signal is still almost zero volts on the NorthStar
motherboard which is quite odd. I should be reading a higher average
voltage than 0 but will have to double check what is going on with it.

It sounds like your trying to measure this with a DMM and that won't work.
Not only will digital meters not average, but a high frequency signal will
screw
them up even if there is some fixed DC component present. They're really
not
much use around computer circuitry except for reading power supply voltages.

As others have said, you really need an oscilloscope.

OK, I hear you on the oscilloscope. I have bought one on Ebay and it
will be a few days for it to arrive. It is a dual trace 40MHz model
which should be sufficient for this task.

In the meantime, however, I have been following the procedures in the
manual to verify operation of the NorthStar CPU board. I have made
some progress but I think a key point to note is the NorthStar CPU
test procedures are very friendly and do allow a lot of verification
without special tools.

For instance one of the procedures add a jumper PRDY to ground. This
task basically forces the CPU and circuitry to perform an auto-jump
routine to the specified starting address and stall out into a
permanent wait state mode. With the exception of the circuitry
related to the clock, the rest of the board should completely static.
This has been most handy in identifying a few bad circuits. I have
found a bad 74LS03 (open collector NAND gate) which I replaced and now
can get the auto-jump address to place on the bus. In addition, I
have found a 74367 which was clearly bad and replaced.

The Horizon is showing occasional signs of life as well. After
replacing the 74367, I placed the CPU board back into the Horizon
chassis, I placed it in with the disk controller according to the
"system integration" checkout procedures and it actually booted or at
least the drive motors turned on and the drive #1 light turned on like
it was seeking a floppy. It did it one time only though. So I
started over with the CPU checkout procedures and found another bad
74367 and a bad address 74LS241. After replacing those chips, I tried
the system integration procedure again and it booted one more time in
a similar fashion.

Now, the disk drive motors are now reliably coming on everytime I turn
on the computer or press reset which before, they did absolutely
nothing. However, the drive lights are not coming on anymore so there
is still something wrong. In addition, when I put in the SRAM board,
the LED indicator light is now turning on which it did not before.

Still, the system is not fully booting yet and I suspect there are yet
more bad chips on the CPU board and/or floppy controller. By placing
it into a wait state mode, it reduces the variability a lot and allows
a mostly static analysis of the board. By comparing voltages to known
good HIGH and LOW values, I can sort out which chips are receiving and
sending valid signals using the boolean logic of the gate. That is
how I found the bad 74LS03 and once I replaced it, now I get proper
HIGH values and correct SM1 signal operation.

So far, I have identified 3 bad bus transceiver/buffer chips which I
believe is a trend. The 74367's are really old and just barely are
capable of driving the bus. Some of the74LS241's aren't much better
and when I swap them around, they give suspicious values which move
with the chip. When they do, they put out barely HIGH and LOW signals
which are borderline into the noise margin. I have been checking the
board once in the test S-100 chassis and again in the Horizon to
verify operation before replacing any chips.

On the downside, I have pushed the static analysis pretty far. I may
get some more diagnosis out of it yet but the time is coming where I
will have to use the oscilloscope to make any more progress.

Thanks!

Andrew Lynch

.