Brazilian Microdigital 64K RAM Pack

[bola_dor]

I recently had this 64K rampack From Microdigital, intended for the TS1500 Clone TK85.

(Ad From K64 #1 an Argentinian magazine from that time)

The actual Pack

Well its not working, I couldn't find any info on-line beside it's "uncommon"
It has 8 NEC 4164 RAM IC on sockets that were tested and OK,

I've already checked for miscontacts or shorts..

With the oscilloscope D in/out /W /RAS /CAS and A0 are all fixed down. All other A lines are pulsing. At ls157 some lines are fixed down and the wors is that all PAL outputs are fixed down.
I don't have an IC tester but I guess PAL is dead. I ts not likely I could find anywhere the firmware or logical arrangement in it.
Unless someone tells me this is clone of another known 64Kb RAM pack.
What I wonder is how difficult could be to make a new multiplexor circuit with 74ls logic IC or to inverse ingeniering that PAL. That's clearly beyond my capabilities right now..

[1024MAK]

Going on your results, it does appear likely that it’s the PAL.

I myself have never seen a 64K byte ‘RAM pack’ in what looks very similar to a Sinclair RAM pack case…

Some other manufacturers did use PAL chips in their products (Memotech being one). But the Memotech RAM packs were more complex.

Reverse engineering something that uses a PAL or any other custom logic chips when it’s faulty will always be tricky. As you can’t confirm the signals during it’s operation.

Somewhere long ago, in the hardware section, I posted the DRAM control schematics for some RAM packs…

Mark

[bola_dor]

Hi Mark, That's what I thought..
I'll search for your post.

A guy at the brasillian TK group showed up wit an identical board but labeled 48k. Too many hypothesis but is it possible to read a PAL somehow?
Thanks

[1024MAK]

I don’t know. However, one method is to use a logic analyser to monitor and record all inputs and outputs over a period of time while all combinations of input signals are tried. Then work it out from there.

Or remove it from the existing circuit and then feed known test signals in while monitoring the outputs. Again, you can then work it out.

It does get more complicated to work out what is going on if it uses sequential logic (flip-flops, registers or counters) rather than combinational logic.

Mark

[bola_dor]

This person from Brazil is doing that job. I hope he succeeds with it !!!!!
The retro community never ceases to amaze me !!!!!
https://youtu.be/O7537rqQgIk

[mrtinb]

That's a great video, when you use YouTube's translation to English.

[mrtinb]

I don't understand Portuguese, but have just used YouTube subtitle AI translation with English subtitles.

But looking at the datasheet, there is only 4096 combinations. And there is no internal state in the chip, so wouldn't it be simple to feed all the 4096 combinations. This will give a truth table, and simple software gives the Sum of Products or Product of Sums.