View Single Post
  #2  
Old 06-07-2007, 01:56 AM
Grant Stockly Grant Stockly is offline
Administrator
 
Join Date: Jan 2005
Posts: 447
Default

From: Grant Stockly
Date: Sat, 02 Jun 2007 03:43:51 -0800
Subject: [Altair Computer Club] 88-PPG Troubleshooting


NOTE:
I JUST programmed a EPROM to find out if letting the thing sit powered up
would do any thing. Well, I now have a perfectly programmed PROM. I
seriously doubt the problem is with my 88-PMC, because it is brand new and
the programming program is running off of it. If the source data were
corrupted from a bad PMC then the program would just crash.

I did replace a CS4410 in the programmer responsible for the programming
pulse, but I think it was working before I replaced it.

I'll find out in the morning if when cold it still does the same thing. It
had a while to warm up because I had to trace out all 18 signals, then
write all this junk, then I programmed a PROM. Probably a good 3
hours. Grrr.... I wish I knew if it was fixed for good. : ) At least
the logic analyzer is hooked up to it and ready to go. In the morning I'll
see how it behaves. : )

The last file I uploaded to the altaircomputerclub I set the "notify group"
button. I thought it might provide a link to the files section for those
who don't know how to get there. Make sure you find all of the stuff! : )

3AM and this programmer testing is testing ME. : ( Now for the message
that I didn't want to let go to waste...

===

I borrowed a MITS 88-PPG 1702A PROM programmer and can't seem to make it
work right. I've programmed about 8 EPROMs. My success rate is 30-60% PER
EPROM. I have never been able to get one good EPROM yet. I will be
uploading some software and data files to the user area for anyone
interested. I'd like any input I can get. : )

I have a DigiView USB logic analyzer. It is a GREAT tool. I located TTL
versions of A0-7, D0-7, Program pulse, and the VGG-VDD power pulse. I
attached those 18 points to the analyzer with little clip on probes and
took ground off of an '04. The address is normal, but the data byte is
inverted. Using the software I've uploaded (its free) you can search the
activity of the programmer on all 18 connections with 10ns resolution over
177 seconds. Did I mention I love this thing? : )

First I'm checking the timing tolerances.
tACW is the Address Complement Set-Up. These 1702s require the desired
EPROM address to be supplied in an inverted state before programming power
is applied. Once the power is supplied and tACH Address Complement Hold
expires the real address comes in. Finally there is a programming
pulse. Seems overkill, but I'm going to check every single possible point
for failure. : )

tACW Address Complement Setup: Minimum of 25us, is actually 286us (wow?!?)
tACH Address Complement Hold: Minimum 25us, is actually 292us
That is where things look weird. At the end of tACH the program pulse is
supposed to go negative, but it doesn't go negative for another
287us! Maybe this is the problem?
tVD VDD VGG Hold: Min 10us Max 100us, measured 65us - Looks Good
tVW VDD VGG Setup: Min 100us, measured 580us
tDH Data Hold Time: Min 10us, measured 172us
tATW Address True Setup: Min 10us, measured 287us
tATH Address True Hold: Min 10us, measured 201us
tDW Data Setup Time: Min 25us, measured 826us
tPW Program Pulse Width: Avg 3ms, measured 2.839ms

I don't know what to think now. I thought that the tACH could be where
things went wrong, but tATW looks like its within spec... Maybe I should
focus on programming voltages and figure out what timer controls the
programming pulse? My problem isn't enough time to program a 1, its that
extra 1s seem to be getting into the EPROM!

Some random info...
From the falling edge of the first power pulse to the rising edge of the
last power pulse for one 256 byte programming cycle I measure 7.17
seconds. With a programming pulse being 2.8-3.4ms on average we have a
good 10% duty cycle, well under the 20% duty cycle maximum.

Where is my problem! : ( What I need to do is hook the logic analyzer
directly to the EPROM socket. There are 128 resistors and 32 transistors
after the points where I was sampling. Still, I got extra bits programmed...
Reply With Quote