By the way, I've tried setting up my OPTi 82C929-based card again on my P8B75-M and still no success.
The card itself is working okay, as I'm able to set everything up (SB mode, WSS mode, MPU401) on AIMB-865 (ICH5, PCI-ISA) without errors.

Current status of 82C929 (using CS4231A as codec):
- SB mode outright fails complaining about unable to set up SB DSP.
- WSS mode can be set up if using a different port than 530h (e.g. 604h), provided the I/O has been mapped and can be accessible. F40h can be used, and is better because that range can be mapped alongside its configuration ports in a single slot (which resides in F80-F9F range, or F8D-F92 to be precise).
- MPU401 configuration also fails. Not sure if it's because of IRQ2/9 (which is usually claimed by ACPI as ACPI SCI IRQ). On AIMB-865 it's possible to move it to APIC territory by enabling APIC ACPI SCI IRQ, though this will break Win98's ACPI (which is better off disabled anyway).

It's possible to move GPIO registers from 500h elsewhere, such as 480h like RUBY-9719VG2AR, but doing this will not make ports 530/534 accessible for WSS purposes.

Also, WSS ports must be reachable for proper setup of either mode, or the setup will fail citing no compatible codec detected.
Not sure what else might be needed for 82C929 to work in SB mode. All I/O ranges normally used by Sound Blaster (220h, 388h, 330h) should be accessible as my CT2290 already works okay.

I did try using the 82C929 card on RUBY-9719VG2AR and it doesn't work there, either, mainly because of the WSS port issue.

dartfrog wrote on 2025-06-13, 08:11:

I don't want to speak too soon, but spoiler: On the IT8888 front; I've ordered a few pcbs from oshpark. Exciting times.

I also created a BGA to QFP adapter so it's possible to use the IT8888F (QFP footprint) PCI card with the IT8888G (BGA chip)

How's the progress?
Life..?

myne wrote on 2025-07-13, 10:13:

dartfrog wrote on 2025-06-13, 08:11:

I don't want to speak too soon, but spoiler: On the IT8888 front; I've ordered a few pcbs from oshpark. Exciting times.

I also created a BGA to QFP adapter so it's possible to use the IT8888F (QFP footprint) PCI card with the IT8888G (BGA chip)

How's the progress?
Life..?

I've been waiting for components and chips. I've had to order and reorder the IT8888's from multiple sources. I was sent the wrong components from a few sources... and they tried saying they sent the right ones... It's been a whole thing.

I've got a solid source now for the IT8888s, or so they say. I have a tracking number that claims I'll have some samples on Monday (tomorrow). Here's hoping.

I'll have to make a new revision to the adapter PCB and the PCI card as they both have minor problems but nothing major and nothing stopping me from using these as I can tell atm.

You can try my PS/2 expansion superIO mod. First you have to check SuperIO datasheet how many PS/2 lines it supports and check the pins if they are free. In a bad luck case they might use them as GPIO for something else. In case of Asus MB-does its SuperIO support PS/2 at all? They still might used some older one with PS/2 but there are also some newer/smaller/cheaper superIOs (like nuvoton) that lacks this legacy stuff like PS/2, FDC, etc...

>LSS...
about the bad 78xx regulators, I quite wonder about it, I don't remember when I need replace some that would behave badly. Did you checked at least with DMM the output if there are some vdrops? Or better with scope if they are not oscillating? Maybe rather some surrounding capacitor went bad that makes 78xx unstable? Maybe as you replaced it with some modern version that is less prone to oscillation it masked the problem. Also some 79xx regulators needed some minimal load (not sure, like 5-10mA) to be stable.

BTW how about the PCIe bridge prototype board, something new?

Edit: heh just reload the page opened yesterday and see photo of the board 😀

I got some IT8888Fs from a couple different aliexpress sellers, I have no idea if they work yet though. Been too busy to try assembling my PC104 board.

I'll try and get to it soon.

RayeR wrote on 2025-07-14, 06:42:

>LSS...
about the bad 78xx regulators, I quite wonder about it, I don't remember when I need replace some that would behave badly. Did you checked at least with DMM the output if there are some vdrops? Or better with scope if they are not oscillating? Maybe rather some surrounding capacitor went bad that makes 78xx unstable? Maybe as you replaced it with some modern version that is less prone to oscillation it masked the problem. Also some 79xx regulators needed some minimal load (not sure, like 5-10mA) to be stable.

I don't have a scope with me. I only measured the voltages with a multimeter when the card was not detected and at one time I saw something like about ~3.8V coming from the 7809 output so I wonder if there's something wrong with them.

Maybe the issue is elsewhere as replacing the regulators (as well as cleaning the contacts just in case) does improve the chance of the card being detected but some of the issues still persist like there's not enough power on the Wavetable header that WavetablePi would keep rebooting itself. For now I'm using something different that doesn't require as much power for MIDI.

As for capacitors, I don't see any caps failing apparently (e.g. bulging) on the CT2290 at the moment, however.

Maybe RPi can take some short high current pulses that makes trigger overcurrent protection/voltage drops. You can try to feed it from PSU cable directly?

TBH I don't think I'll do more on those Creative SB16s for the time being, as it seems there are some issues that manifest only on Creative sound cards, like the inability to play any 16-bit audio on NT3.51 (possibly affects NT4 also but fine at least with XP). That issue affects as far as AWE64 Gold and can happen with a relatively proper ISA bus environment like AIMB-865 (ICH5, PCI-ISA).

Non-Creative cards don't appear to be affected by that issue even if using NT's built-in Sound Blaster driver (SNDBLST.SYS). The fact that a truly SB16-compatible card like CMI8330 works without being affected by that issue on the same environment (AIMB-865) as when I used AWE64 Gold (CT4390) kind of ruled out the high DMA being the issue and narrowed it down to something Creative-specific.

AFAICT PicoGUS' SB2.0 mode currently works without any major issue, including those affecting Creative SB16s. Perhaps it will also not be affected by those when SBPro and later support becomes a thing.

Unfortunately for now it seems Intel's LPC atmosphere (up to 9-series PCH), is simply not friendly to some non-Creative cards, limiting dISAppointment's usefulness.
- PicoGUS can't work in GUS mode behind LPC, citing no memory detected. I wonder if I need to configure something memory-related for the LPC controller. The card itself is actually fine, however.
- On ASUS boards ISA PnP doesn't work properly.
- WSS cannot be placed at 530h, because for many boards GPIO configuration registers reside in this range (500h-57Fh) and while the registers can be moved elsewhere, the LPC still cannot access 530h after that.
- Some cards require more than one IRQ which further complicates things, as on some boards it's already a hassle to make either IRQ5 or 7 available. One of the main need of the extra IRQ on non-Creative cards would be Intelligent mode MPU401, which usually uses IRQ2/9, and can be a problem with some OS' ACPI that don't support APIC (e.g. Win98).

PS: A question about WSS. Does WSS really require a separate IRQ from SB to work, or that's also vendor-dependent? This would the potential issue of requiring a third IRQ with WSS-capable hardware. In CMI8330's case:
- UNISOUND supports setting up an IRQ for it, but it defaults to not set (IRQ0).
- Win9x's VxD driver does not use an extra IRQ for WSS but its WDM driver does.
- WinNT4 driver seems to require one and by default set it to IRQ11. This IRQ is often used as one of PCI INT# lines along with IRQ10. Tried that driver with WinNT3.51 but it doesn't appear to work with a vague "an unexpected error" message, even though I've reserved IRQ11 in BIOS moving the particular INT# line elsewhere (I'm using this card on AIMB-865 which supports this functionality).

LSS10999 wrote on 2025-07-14, 15:05:

- On ASUS boards ISA PnP doesn't work properly.

On Asus B85/Z87/Z97 boards access to port A79 (write data) is blocked somewhere rendering ISA PnP init sequence not working at all.
The solution is - PicoGUS, ESS1688/688 - half PnP card, where IRQ and DMA while set by software, the protocol does not require ports other than base port 220 to init. ES688 has no MPU401 in hardware. Some Aztech cards works also.
Es1869/1868 PnP cards can be initialized without access to port A79, using only ports available (279 and bypass key sequence).

LSS10999 wrote on 2025-07-14, 15:05:

- WSS cannot be placed at 530h, because for many boards GPIO configuration registers reside in this range (500h-57Fh) and while the registers can be moved elsewhere, the LPC still cannot access 530h after that.

I spend few days trying to run Impulse Tracker in WSS mode on ymf719. And while drivers (both - gusmax and itcs) found Crystal Codec (!) on any other than 530 ports, no success: just no sound. SETYMF was able to push test tone in 48kHz WSS mode though.
Also I have serious issues with hanging notes on (three) YMF719-S with Wolfenstein 3D, where all the music is Adlib. Seriously, Yamaha has flawed OPL3 implementation here.
I finally return back to ESS's 1869 where everything smooth on both Asus Z87 (using bypass key init utility) and Asrock Z87 with UNISOUND.
BTW ESS may not be initialized when default control port (6xx, 8xx, depending on card) are not decoded even with A79 port accessible. This is not issue, because UNISOUND sets desired control port (R240 for example) from SET BLASTER env var.

I had some free time and I got a little update, I figured out the ramp for the hotplate / mg chemical leadfree solder paste / stencil application. Looks okay.
Unfortunately I am still waiting on a component, it got "lost in the mail"... but I might just use what I have on hand. ie non-surface mount electrolytics in place of the surface mount footprints as well as the resistor networks. (using two sip5/sip9 resistor networks in place of the larger ones)

I'm getting impatient and I'm sure you all are too. Anyway progress has been moving just slowly. Should I say screw it and make a Frankenstein card, or wait for the right components? I'd be leaving out the optional clock gen circuit and using the "wrong" footprint components. I probably would be able to finish the soldering tomorrow/tonight.

FWIW I have 2 ISA cards I thought would be good test candidates. (I had an ISA graphics card too, but I guess I lost it. I also had a CNC card, but it's no longer working. 🙁 )

• PicoGUS v2.0
• MicroMedia 8bit ISA 5.25" Floppy controller

I have 3 PCs/Motherboards to test on.

• Lenovo M93p with Intel Core i7 4790 and Intel Q87?
• Nixsys (1 AGP, 1 PCI slot, no ISA slot - SL73Z (ICH5)- I forget what CPU maybe Pentium 4?- They don't sell it anymore, similar to NX-L70 (also don't sell anymore))
• P4SD-VX (Sony OEM) motherboard with Intel 865G and Pentium 4 2.8GHz

For homemade prototypes I always rather use SnPb solder and paste, it feels better and need lower melting temp...

So I'm curious how DMA will work and all that bridges/IO windows mess....

No point in waiting.
Revision 0.1 of anything, ever has always been a Frankenstein's monster.
Even if you wait, you will at some point have bodge wires, and other things hanging precariously.

Function > form

Alright, It's all soldered up. Give me a bit to clean up and setup the testing area. I've already preprogrammed the eeprom, but added pin headers and breakout board for ease of reprogramming if i got that part wrong.

Edit: I just want you all to know how much time, effort, and money has went into this project. It's been a lot for a random let's see if it will work. If it doesn't work at least we tried the IT8888 approach. If it does great. In any case thank my wife for all the support in this, without her, I would have never even tried.

Cheers to your wife!
A rare woman.

Looks like a solid 0.1.
Hopefully all issues are non-destructive.

myne wrote on Yesterday, 00:33:

Cheers to your wife! A rare woman. […]
Show full quote

Cheers to your wife!
A rare woman.

Looks like a solid 0.1.
Hopefully all issues are non-destructive.

Indeed. She thanks you!

Lol, so interesting you say that. I thought I messed up big time. Big pop from the Sony OEM motherboard but I realized I didn't have the card plugged into it. Phew. 2 mobo caps on that went, they were already leaking from the tops but i figured it might be fine... apparently not 🤣. Gonna fix it and mess with that one in a day or two.

The M93p doesn't like the card. I think it's sucking too much power from the pci slot? Not sure. It goes to boot, then stops. I'll have to investigate further on that one. Maybe I'll try the pcie to pci card on this, the pcie - pci card might give enough power?

So far It's completely fine in the old nixsys mobo, right now i'm installing windows XP with it plugged in no problems. Fingers crossed. oh btw this nixsys motherboard is a Celeron based one.

Not exploding is a great first step

myne wrote on Yesterday, 01:36:

Not exploding is a great first step

Even better 😀 (requires setting "Other PCI Bridge ?!" driver with the ite8888.inf)

Edit: I plugged in the PicoGUS, booted the machine and huge explosion... Cap on the PCI-ISA card's -12v rail blew and sent the PicoGUS into oblivion. The PicoGUS's "U1" chip exploded (SN74CB3T3257). 😐
I guess I must have sent -12v to somewhere I shouldn't have. RIP U1. Time to probe and see what's getting -12v.

Oooh!

One old smart guy teach us to cover the prototype by a felt hat before 1st power on (to catch all that flying shrapnels from the explosion instead get them hit into an eye 😀 Was the polarity of caps OK? or did get negative voltage where it shouldn't get? It's always good to measure the power rails and voltage regulators before pluggind a cad in... OK bridge seems to be detected. What can be set via INF? Does it display some used resourced in devmgr? Does at least outportb (0x80) display something on a POST card? No collision of ITE DMA and onboard DMA controller?

RayeR wrote on Yesterday, 05:06:

One old smart guy teach us to cover the prototype by a felt hat before 1st power on (to catch all that flying shrapnels from the explosion instead get them hit into an eye 😀 Was the polarity of caps OK? or did get negative voltage where it shouldn't get? It's always good to measure the power rails and voltage regulators before pluggind a cad in... OK bridge seems to be detected. What can be set via INF? Does it display some used resourced in devmgr? Does at least outportb (0x80) display something on a POST card? No collision of ITE DMA and onboard DMA controller?

That's a good tip. I covered it with a glass tupperware. That way I could see if anything lit on fire and protect the eyes. Good thing thing I did too, tiny flame that self extinguished on the PicoGUS's U1 🤣. Poor GUS. I did measure the voltages and everything looked good, but I measured at the components (caps) and not the ISA slot. That was dumb, but exciting!

The exact problem was that the A and B ISA slot sides are swapped. It's very funny and wow that's embarrassing. I have no idea how I managed to do that. I guess I can solder wires from the back of the card on the ISA slot to another seperate ISA slot with the sides swapped until I fix the schematic/PCB and order a new revision.

I would have probably figured the A/B ISA slot swapping if I measured the voltages at the ISA slot without blowing my GUS to bits. Big dumb by me. I was just so excited to test it. I ordered some replacements of the SN74CB3T3257, so GUS should be fine. I'll have to look the both cards over and see if I messed anything up too badly, I don't think so, there's really only the -5v, -12v and +12v that would be a problem looking at the correct ISA pinout, and besides the blown -12v cap, I think the ISA8888 card is fine. I'm unsure if the GUS blew from the +12v, -12v or -5v, probably the +12v/-12v. There is a tiny problem in the -5v generation section but it was mainly feeding the -12v/+12v voltages to the data lines ... that was fun i guess.

One big step forward, two smalls steps back. AH, so close though. At least we know the unloaded card can be seen in OS for now. Also, learned a few good lessons 🤣.