After a lot of headaches I have OPL4 playing sound ! I managed to fry one chip though, I had !OE and !WE swapped in CPLD pinout conf and I damaged the IO buffers of one OPL4...
The CPLDs are dead end and I have been working on a kludge :

Flex10K FPGA with its conf memory that will go in place of the original CPLD. Now there are no limits on what I can do. Parts are coming from China though so it will take a while... In the mean time I'll begin work with the STM32.

First code runs on the ARM 😁

Long and painful road ahead, lot of peripherals to learn and whatnot...

Do you have any test code implemented yet? Are you planning a purely bare-metal implementation, or with a supporting OS?

The first positive results are always encouraging. Do you know the source of the "Cannot read memory" error?

I have only done very preliminary code, setting up interrupt vectors and few other bits, just enough to get to a point where the CPU runs my code which currently is just putting some values into the registers. It is going to be full bare metal where every cycle and byte counts so that I can get maximum polyhpony and have most amount of space for any samples and LUTs.

"The cannot read memory" was a me stumbling in the cable 🤣. In the beginning I was unable to actually run anything but that was matter of changing few directives on the assembler end of things and a misunderstanding of memory map.
Once I finish one PCB I'll focus a bit more on the code side of things and will get a LED on the board light up or even blink. Some sawtooth coming out the DAC would also be quite nice to have but there's a lot of peripherals that need to be set up for things to start working nicely. The documentation is very good and I already have some experience with a different family STM32 so I mostly know what needs to be done to get the results I'm after, things should be fairly smooth once I really get into it.
Most of these chips are both pin compatible and also software compatible or at least only needing recompile (symbols etc. have same names and often same location/function). ST has done a very good job making parts that are upwards compatible.

I also got the FPGA configuration memory, FPGA itself shouldn't take long to follow and PCBs I'll get next week most probably. Then I'll work in MPU-401, SB and ARM comms interfaces. Getting SB going is probably gonna happen first, it is much less work compared to a synthesizer.

After half a day of chasing a problem that didn't exist (I had an assembler bug instead) I finally have the LEDs lit up :

😁

I still have to initialize the peripherals I'm using (UARTs, audio interfaces, DMA, FMC and some other stuff) but for the time being I'm pretty happy 😁

leds are important 😀
nice to see this progressing!

Sometimes just getting a new chip ready for useful work takes almost as much time as actually getting it to do what you need. I don't think you mentioned it yet: what development environment are you using? Did you start with an off-the-shelf package, or assemble your own using a free IDE and cross-compiler?

Also, do you have a good (and cheap) source for the 29F160? I have been using the Fujitsu equivalent, which also supports "fast mode" programming. At the time, they were available on eBay for somewhere between $2 and $2.50 each, and shipping was quite reasonable. If they don't have any more, I may need a new source.

Because i now have ISA slots machine, i hope that these leds could be disabled.. they are great for optimal diagnosis, but i hate blinking especially in case with transparent side..

The LEDs are there solely for diagnostic purposes, I can show internal state with them or whatever else I might need during development.

As far as development goes, I used Cube MX to generate pinout and config that I have referenced for correct values to use in PLLs and whatever else in my assembly blob. I am using FASMARM + ConTEXT + several monitors showing PDFs as my development environment. Getting started part is the most difficult, actual logic and synth will not be too much of a problem since those work how I want.

As far as flash goes, I'm using few parts from my stash that are left. I got a handful more and it'll be enough for these prototype needs. All future things are 3.3V only anyway so that I can use modern parts instead.

Next week I'll get the PCBs of the FPGA kludge and can expand the card to what it is supposed to get. I'm possibly able to do most of SBpro code entirely in the FPGA, I'm not quite sure how much the Flex 10K can hold but it'll certainly be much more than the CPLD currently. Once FPGA is in place I can finally work in ISA <> ARM comms.

We have FLEXkludge ! Things can continue on hardware front now ~

This looks amazing

Hahaha, thänk you !
Your card is pretty nice too ~

That has to be the neatest, cleanest "kludge" I've seen in quite a while; nice work. Are those standard 1.27mm-pitch pin headers?

Yeah, these are standard 1.27mm, bought from ebay. If the original chip was something else besides PLCC this would have been much more complicated, possibly to a point a new PCB is needed 🤣.

I now have solid MPU-401 data access interface, only thing needed is ARM actually doing something. FIFO would be nice but first test used up lot of LE's in the FPGA (in addition to memory bits) and I still need them for other stuff that are yet to be implemented. Once I have all needed functions done I'll work in FIFO so that the MCU wouldn't need an IRQ on every data byte.

SB data passing is mostly implemented too but not to a point where games can detect anything.

One curious thing is that IRQ3 which I used for YMZ290 stopped working. And at the same time when I switched CPLD to FPGA, so my thought was that I have a timing issue since the FPGA is much faster and will have to clean it up somewhere... eventually nothing fixed music being gone, I could talk to all the chips properly and I could see IRQ3 line change as needed too... Then I tried the SW20PC and it behaved the same way, plot thickens. Then I took my YM71x card and ran a test in SETYMF and lo and behold, IRQ3 was not firing any interrupts ! I changed to IRQ 5/7/9 and everything works, FPGA was never at fault... I have no clue how IRQ3 died on the motherboard...

No work has been done on the ARM side for the time being. Effort has mainly gone into work related things...
First serious ARM thing will probably be making UARTs work and pass MPU401 data to the WaveBlaster connector for some funstuff ~

My second test card has been FPGA'd too, to verify that the problem wasn't with the first card also while I was trying to figure things out.

great stuff!

Hi Tiido.
I think I've found a bug in SETYMF utility. On my YMF718 card (OPL3SA2) I've obeserved problem with only two games (Dynablaster and Cool Spot) FM sound is playing fine, but SFX causes game to hang - FM still plaing in background, but game freezes. When I'm using setupsa.exe from Yamaha there is no problem at all. Rest of the games are perfectly fine with your utility.
Motherboard is Intel Ninja (420EX), CPU i486DX

Thänk you for the report, I will investigate. I even have the same motherboard to test things with.

Hi Tiido.

Does your card fix the mixer problem described here?:
Yamaha YMF71x SB Pro Mixer Bug Issue Fix

Also, does it have a proper lowpass filter like the SBPro2 CT1600?
Since Yamaha YMF71x based cards usually don't have the proper value caps to match the LPF on the CT1600.

My software does set things so that the SB level change range is maximal.

There's no SBpro2 style filter by default, all the filtering is made so that there's no meaningful attenuation in audible freq range on any sound source. You either have to change couple capacitors if want only the SB DAC output to get darker or adjust high freqs slider in SETYMF which will muffle all the sound sources (internal and external, which I imagine is how SB hardware functions, but I have no SBpro2 to actually see if that's the case and likely never will, those things are not worth the money you need to cough up for them...).

I will investigate the reported games problem today and hopefully I can come up with a fix.