Does a vintage system have a limit as to the fastest RAM access time it can use?

[T-Squared]

Just in case the SRAM I bought doesn't work, I also have a PSRAM chip on the way as well.

 

[Eudimorphodon]

Just in case the SRAM I bought doesn't work, I also have a PSRAM chip on the way as well.

Well, if my theory about what's going on here is correct then the SRAM isn't going to work... probably. Do you have the part number/datasheet of the particular SRAM you tried to put in there? Maybe there are SRAMs that latch the address bus the same way the PSRAM appears to, but I don't think "normal" 8x32K parts do.

 

[T-Squared]

It was a CY62256N; a Cypress Electronics 32k X 8 SRAM memory chip. (Yes, It says Alliance memory, but Cypress was absorbed into Alliance)

https://www.mouser.com/ProductDetail/913-C62256NLL55SNXIT

 

[Eudimorphodon]

It was a CY62256N; a Cypress Electronics 32k X 8 SRAM memory chip. (Yes, It says Alliance memory, but Cypress was absorbed into Alliance)

Yeah, so, if the chip really is wired in like in the schematic I found it definitely will not work, full stop. On the timing chart there’s a value, tOHA, that flatly indicates the current data on the output pins is only valid for 5ns after an address change and there is no latching, data follow address *always*. If the control ASIC is expecting A0-A7 to latch on OE so it can change AD0-7 from an address to a data line for the read/write operation that is no dice.

 

[Eudimorphodon]

… by contrast, look at the timing charts in the HM65256 datasheet, pages 6-9 and up, and look for time interval tRAH. This chip needs the state of A0-7 to be stable for 20-30ns on asserting CE, and after that it *does not care* what they are, it’s fixed until CE is released. A8-14, on the other hand, follow address changes like a SRAM does. That’s how the static column mode works.

 

[T-Squared]

Ok, good. So I most likely did do the right thing in getting a similar PSRAM chip (Same HM62256, but 80ns), just in case. Too bad I don't have a use for the SRAM chips that I bought... yet.

BTW, this was the extra SRAM (Not PS) type I got after the Cypress Memory didn't work: https://html.alldatasheet.com/html-pdf/65379/HYNIX/HY62256ALJ-I/1219/9/HY62256ALJ-I.html It looks like it has the same timing that you described, so that might not work.

 

[Eudimorphodon]

Ok, good. So I most likely did do the right thing in getting a similar PSRAM chip (Same HM62256, but 80ns), just in case. Too bad I don't have a use for the SRAM chips that I bought... yet.

Just to be clear, it’s a HM65256 you need. There is a datasheet out there for an HM62256, and it’s just a SRAM with the same characteristics as any other 62256.

 

[T-Squared]

OH CRAP!

Thank you for reminding me! I really appreciate it!

So we do have a confirmation, at the very least, that normal SRAM cannot be used in a Game Gear? It needs to be Pseudo-Static RAM?

 

[Eudimorphodon]

If that schematic I found is accurate then, yeah, it looks like it was designed to be *very specific* with regard to what memory chip it needs.

If I really had to guess I think this VDU was originally designed to work with DRAM, but the designers shuffled the guts a little to work with this PSRAM because it’s about perfect for the job.