cascadian wrote: ↑
Fri Feb 21, 2020 5:36 pm
It's as if the sample RAM is reset, all sounds reduced at once to a faint click, eventually followed by total silence. I do have a PS Systems 4X Memory Expander
, unfortunately uninstalling it did not resolve the issue.
About part number:
Ensoniq service manuals do not have schematics or part numbers, so that's to be expected. In the case of the EPS, the schematics do happen to exist though. The RAM in question is 41256-type 256K dynamic RAM (DRAM), but I do not believe it is going to be worthwhile for you to acquire or replace any of the DRAM.
About EPS sample memory:
When using the 4X expander (such as you have) the EPS internal DRAM memory is allocated to the sequencer exclusively. The DRAMs within the 4X expander then become the only place where samples are stored into memory.
With the 4X expander removed, the EPS internal DRAMs are shared between sample and sequencer use.
Therefore, the fault should not persist with the expander installed and removed, if we're going with the bad DRAM hypothesis. Besides, if it was a bad DRAM or two then it would very likely manifest as pops and clicks within certain sample memory ranges rather than all sample memory being empty.
The normal EPS system RAM is static RAM, so it does not need to be updated constantly -- there is no refresh cycle. The sample/sequencer RAM is dynamic, so it has to be periodically refreshed or else the contents will be lost.
the dynamic RAM contents seem to be disappearing, that points toward something like column or address strobe timing being too slow to properly keep the RAM refreshed.
I disagree on the point that it cannot be PSU related. If the PSU is performing marginally, some sub-systems could be affected even though most or all of the logic chips appear to work.
EDIT: gkmaia makes a good point though. On one hand, it seems like the +5M power supply line to the DRAMs might be an obvious culprit, yet that line is also shared by the 27256 program ROMs, so how can the EPS still function with unreliable power for the critical EPROMs?
On the other hand, the EPROMs and 74LS logic which share the +5M line with the DRAM could be more fault tolerant than the DRAM, and that the DRAM is just slightly more sensitive. The +5M line could be close enough to spec that mostly everything works, but also out-of-spec just enough that the DRAM occasionally fails and loses all its contents.
Totally guesswork, but I think it's worth measuring PSU J4 pins 7+8 for ripple and VDC*
*Represented in the EPS service manual as terminals "13" and "16", if you're following along with that.