• Please review our updated Terms and Rules here

Sol-20 Processor Technology 8K RAM stuck bit

O

Old_hitech

Experienced Member
Joined
Dec 30, 2016
Messages
112
Location
Ventura CA
I was checking the memory in my Sol-20 and I have a generic 32K RAM card that works fine. I can write and read all bits just fine. I have an 8K RAM Processor Technology board that appears to have a stuck bit in MSB (Bit 7) position. I have the 8K RAM setup to start at 8000h. When I enter data with EN 8000 :00 00 00 00 00 00 00 (other bytes 00 also) and read back the data with DU 8000 800F I get 8000 80 80 80 80 80 80 80 (other bytes 80 also). I reseated all eight ICs in the BIT 7 row location, but the problem remains. What should I try next?
 
I did a little more trouble-shooting and found that one of the 2102 chips seems to be bad. I noticed that the bad chip is a 2102-2 while all the working chips are 2102-4. Not sure if the dash number is the reason my bit is stuck, but now I am looking to see if I have another memory board with a spare 2102-4 chip.
 
Old_hitech said:
I did a little more trouble-shooting and found that one of the 2102 chips seems to be bad. I noticed that the bad chip is a 2102-2 while all the working chips are 2102-4. Not sure if the dash number is the reason my bit is stuck, but now I am looking to see if I have another memory board with a spare 2102-4 chip.
Click to expand...

You can swap the IC's around is a good test to see if the stuck bit moves with the IC, or remains at the socket location corresponding to that bit is a good test.

I would think a 2102-2 would work, assuming it was not a damaged IC.

It is not uncommon to find these IC's occasionally failing, but on the PT cards I have, most of the problem is the TI sockets/IC pins. Look on the sides of the pins of the IC, to see if you can see a thin grey or a black line where the socket claws were contacting it. The corrosion due to dissimilar metals creates this oxide, which is an insulator, that needs to be scraped off. Gold plated pins are less troublesome, but they still need a gentle clean. Re-seating the IC's sometimes will not cure a connection issue if there is corrosion on the pins, or if it does the problem returns in short order.
 
I could not find another 2102 memory chip, but I did find a couple of Tarbell 32K RAM boards in my box of S-100 boards. I removed all S-100 boards from my Sol-20 and installed the first 32K RAM board and tested a few memory locations with read/writes throughout the 32K range of 0000h to 7FFFh. Everything checked okay. I removed the first board and installed the second 32K RAM board with the same settings as the first board. I performed the same memory tests and everything checked okay again. I referenced the Tarbell 32K RAM board manual at DeRamp.com and changed the settings of the second board to provide 16K from 8000h to BFFFh. I installed the second board by itself and checked the memory at various spots from 8000h to BFFFh. Everything checked out okay. I installed the first board providing 32K from 0000h to 7FFFh and starting double-checking various memory locations from 0000h to BFFFh. Initially, I found that I had some issues from 8000h to BFFFh, but was able to resolve these issues by removing the second board and reseating a few of the iCs and re-installing it into the Sol. I now have a 48K Sol. Funny thing is that I have had these 32K RAM boards and my box of S-100 boards for over 20 years. I bought them from a electronics surplus warehouse in Southern California. I did not have a S-100 computer at that time, but I thought it would be a good idea to buy the cards in case, in the future, I get an S-100 computer. Most of the boards were Tarbell boards. The box even contained several brand new bare boards. I think these boards were liquidated from the Tarbell company when they ceased operation. Their manuals show that their location was in Carson, California.
 
Sounds like you are fine for memory boards.

Initially in my SOL I was running three PT 16kRA boards (an interesting lockup problem happens with these that I mentioned on another thread, but its easy to cure). I wrote a brief description of the problem and how to fix it:

http://worldphaco.com/uploads/SOL_Lock_Up_problem_3_x_16KRA_Ram_Cards..pdf

Then I moved to an SCP 64k board configured to 48k, this freed up some card slots for other projects, one like this:

http://www.s100computers.com/Hardware Folder/Seattle Computer Products/64K RAM/64k RAM.htm

But the one I have has really interesting and beautiful looking Inmos gold & ceramic IC's. So far it has performed flawlessly. And I bought a few spare chips to be on the safe side, in case one dies.
 
I made a debug board to help diagnose KIM-1 boards. My main purpose was to determine if one of the 6530 had failed. One had so I made a really small 6532 replacement, similar to Ruud's boards but small enough to fit under the 6532. Since I'd invested in making the circuit boards, I wanted to see if others wanted them. I got 10 costumers. They'd build the diagnostic board and connected it. Almost all of the failures they had turned out to be from the 2102s. One was the address decoder. None needed the clever 6530 replacement I'd made.
Dwight
 
I'm curious...Did you use a program to help fault find the bad RAM?
My Sol 20 is playing up too, and I need to test the RAM, but not sure what memory test programs are available.
 
I believe there are some more extensive memory tests available, but my quick memory test on my Sol-20 is to store some various data in the first few bytes every 1000h and read them back. For example, I will type EN 0000 <return> then enter 11 22 33 44 55 66 77 88 99 AA BB CC DD EE FF/ and then read it back with DU 0000 0000F command. I then look and see if the data on the screen matches what I entered. Next, I test the next 1000h with EN 1000 <return> and then enter the same 11 22 pattern or change up the pattern to 5A A5 5A A5 5A A5. No real reason for the specific pattern in my case, but if I did notice a specific bit failing I would change the pattern to focus on the failing bit. I continue testing the first few locations at 2000h, 3000h, 4000h, etc. I used this approach as I really wanted to see if I had the RAM memory card addressing setup up correctly as I was using 2 different 32K RAM memory boards to yield a 48K RAM system.
 
You must log in or register to reply here.
Back
Top