Can 35 year old gate arrays be duplicated

Thread Starter

jimkk

Joined Mar 11, 2019
32
Does anybody out there have experience with FPGAs? I'm part of a club that restores old synthesizers (musical instruments) for our own collections. Replacing obsolete IC's is always our greatest challenge. You guys have been the BEST at helping me with earlier issues, so I'm reaching out again with what I think may be the most challenging project I've embarked on. The past couple years, we've seen an increasing number of these old synths die because one of the custom gate arrays has gone bad, but oddly enough, there's this guy in Poland who always seems to have a replacement in stock at 10 times what they're worth......followed by VAT, customs, shipping costs and months before the chip arrives. Thing is, where's he getting them? For the most part, when we need to replace one of these custom IC's, we scour the internet for somebody parting out a unit. As the years click by, this resource is becoming more and more difficult and we find ourselves going to him. So here's my question. Can old technology gate arrays be read and duplicated in a FPLA like EPROMS? Or....is there a solution to this and I'm asking the wrong question? For the sake of discussion, the custom gate array we're currently working with is a Hitachi HG61H25 (HG61H25B18F). I have attached the original data sheet. We have a good original chip that could be read. I'm not concerned with reverse engineering. Would just like to figure out a way read a good IC so we could replace dead ICs with a functional equivalent. I don't think I'm ready for a step-by-step procedure, but rather, maybe a short overview if/how this can be done, what equipment/software would be required and maybe identify a modern FPGA that we could use for discussion. Thanks, everyone. Jim
 

Attachments

ElectricSpidey

Joined Dec 2, 2017
739
I have programmed EPROMs to replace logic arrays, so depending on what you need to do, I would definitely look at that possibility.

And I would also look into the PsoC…
 

danadak

Joined Mar 10, 2018
3,626
If you had the Verilog or VHDL code you could possibly hack them into
current technology parts. Even a schematic of the logic solution.

But without that just almost impossible. Also getting the same package
with the same power domains just very difficult to find, if ever.

Its unusual for an ASIC, lower power types with low clock rates, to fail
unless I/O subject to ESD or power transients.

PSOC might work if gate count fairly low. But you would need either code in language or
schematic. Xilinx, Altera, Lattice more appropriate.



Regards, Dana.
 

Thread Starter

jimkk

Joined Mar 11, 2019
32
Thanks, Dana,

What I've seen (x3 in 2 years), its never just the gate array. First, I repair a seemingly unrelated issue and after completion, find that I'm only getting partial functions and the gate array ends up being the culprit. I talked to the factory and they tell me they don't have any issue with my intentions and would help if they could, but, as they recite their rather strict company policy, "the original schematics and/or programming data has long been deleted or recycled" and besides, top management frowns on even their old timers (who they "hint" could probably locate what I'm asking for in about 10 minutes) helping guys like me with products that are well past their EOL. For example, it was suggested that Hitachi most certainly still has all the original data and one guy mentioned that he could probably locate it with a single phone call, but they're not even allowed to ask. Meanwhile, before writing here, I looked at the Xilinx and Altera stuff but to be honest, is all waaay over my head, so then, I came to you guys to see if there were any options I'm not considering. I suppose the guy in Poland could just have access to synths that he's buying for parts (good business), but this particular manufacturer doesn't even sell their products in Poland. In my simplistic mind, I was kinda hoping there might be a way to set up an emulator made from a working product and record all the I/O and give the results to one of these FPGA companies that could convert it to a "burn" instruction set (like we do with EPROMS). but I guess that's just wishful thinking. If anyone has anything to add, please chime in. Thanks for your help.
 
Top