mbandalauk wrote on 2025-08-30, 13:45:

Hello again, […]
Show full quote

Hello again,

Now that I’m back, I can share my findings using the PCM-3370 on the backplane. In short — it works flawlessly!

I mounted the PCM-3370 on a PCA-104-ISA REV.A1 sourced from AliExpress. The SBC can be powered directly from the ISA connector, and everything seems to run fine. Both Windows 98 and MS-DOS detect the onboard ESS sound chip without issues. The audio is a little noisy, but nothing that really bothers me.

My only nitpick was with the PSU connector: mine has a 24-pin plug, and as you can see in the second picture, a capacitor is in the way. Because of that, the connector doesn’t fully seat. Fortunately, since it’s actually a 20-pin + 4-pin design, I was able to detach the extra 4 pins without any problem. (It might be worth considering a small layout adjustment in a future revision — if that capacitor were just 2 mm farther from the edge, the connector would fit properly.)

Performance-wise, the SBC is excellent. With its Celeron 400, it boots Windows 98 in under 10 seconds. Since it’s an Intel processor, your PC/104+ adapter should work fine with it, right? If you tried your PCM-3353 with the Chinese PCA-104-ISA, would you get the same results?

Best,
Manuel

Awesome! Thx for the IDSEL numbers!

After assembly, I did notice the 24-pin ATX didn't fit. It did fit on a previous version, but after I moved those caps it no longer did. This didn't bother me, though as the 24-pin ATX was unnecessary to the design, and most current ATX supplies can detach the extra 4 pins as you mentioned. Moving those caps again isn't out of the question on a revised board.

Glad to hear it's all working. You mention a noisy sound output? In my testing, there was some slight background hiss coming from the ESS chip. However, it seemed to be no worse than my genuine ESS sound cards, so I considered that a reasonable compromise. During the design work, I really wanted an ISA sound chip with a built-in digital out, that way I could have implemented a simple DAC, and the outputs would have been truly noise free. The options were super limited, and nothing would have fit into that small footprint on the board.

I did add a simple RC filter to help reduce the noise. My schematic shows 1k over 8nf, but I believe I went with 1k over 10nf or slightly higher. So, 8nf ≈ 19khz F3, and 10nf ≈ 15khz F3. (R25 + C90) and (R88 + C89). Unfortunately, the chip itself is the source of the noise. But, if you were so inclined, you could increase the capacitor values. I was mostly testing with a battery-powered JBL until and was satisfied with the results. Definitely not audiophile grade, but reasonable when I compared it to my genuine ESS cards.

As for my PC104+ boards... I fear the limitation is in the AMD Geode CPU at this point, and it would perform just as poorly with a PC104 ISA adaptor. So, I'm keeping my eyes out for a better PC104+ board, or something like an older 486. I don't consider the PCM modules to be a bad purchase as I needed a real PC104+ board to test the socket fit. I'm happy with the results so far.

Please feel free to offer more feedback on the backplane. Of the few that have sold, I'm curious about how you plan to use it, and I might be able to offer help.

PS. If you were interested in one of these adaptor cards mbandalauk, let me know!

Also, I see in your pic that you have the ISA adaptor card in the ISA slot? In case you weren't aware, you can also use that ISA card in the SBC slots. Just FYI. 😎 Although with the ISA only adaptor, having the back slots for ISA cards is probably more reasonable.

Tevian wrote on 2025-08-31, 05:58:

Awesome! Thx for the IDSEL numbers! […]
Show full quote

mbandalauk wrote on 2025-08-30, 13:45:

Hello again, […]
Show full quote

Hello again,

Now that I’m back, I can share my findings using the PCM-3370 on the backplane. In short — it works flawlessly!

I mounted the PCM-3370 on a PCA-104-ISA REV.A1 sourced from AliExpress. The SBC can be powered directly from the ISA connector, and everything seems to run fine. Both Windows 98 and MS-DOS detect the onboard ESS sound chip without issues. The audio is a little noisy, but nothing that really bothers me.

My only nitpick was with the PSU connector: mine has a 24-pin plug, and as you can see in the second picture, a capacitor is in the way. Because of that, the connector doesn’t fully seat. Fortunately, since it’s actually a 20-pin + 4-pin design, I was able to detach the extra 4 pins without any problem. (It might be worth considering a small layout adjustment in a future revision — if that capacitor were just 2 mm farther from the edge, the connector would fit properly.)

Performance-wise, the SBC is excellent. With its Celeron 400, it boots Windows 98 in under 10 seconds. Since it’s an Intel processor, your PC/104+ adapter should work fine with it, right? If you tried your PCM-3353 with the Chinese PCA-104-ISA, would you get the same results?

Best,
Manuel

Awesome! Thx for the IDSEL numbers!

After assembly, I did notice the 24-pin ATX didn't fit. It did fit on a previous version, but after I moved those caps it no longer did. This didn't bother me, though as the 24-pin ATX was unnecessary to the design, and most current ATX supplies can detach the extra 4 pins as you mentioned. Moving those caps again isn't out of the question on a revised board.

Glad to hear it's all working. You mention a noisy sound output? In my testing, there was some slight background hiss coming from the ESS chip. However, it seemed to be no worse than my genuine ESS sound cards, so I considered that a reasonable compromise. During the design work, I really wanted an ISA sound chip with a built-in digital out, that way I could have implemented a simple DAC, and the outputs would have been truly noise free. The options were super limited, and nothing would have fit into that small footprint on the board.

I did add a simple RC filter to help reduce the noise. My schematic shows 1k over 8nf, but I believe I went with 1k over 10nf or slightly higher. So, 8nf ≈ 19khz F3, and 10nf ≈ 15khz F3. (R25 + C90) and (R88 + C89). Unfortunately, the chip itself is the source of the noise. But, if you were so inclined, you could increase the capacitor values. I was mostly testing with a battery-powered JBL until and was satisfied with the results. Definitely not audiophile grade, but reasonable when I compared it to my genuine ESS cards.

The attachment schematic.jpg is no longer available

As for my PC104+ boards... I fear the limitation is in the AMD Geode CPU at this point, and it would perform just as poorly with a PC104 ISA adaptor. So, I'm keeping my eyes out for a better PC104+ board, or something like an older 486. I don't consider the PCM modules to be a bad purchase as I needed a real PC104+ board to test the socket fit. I'm happy with the results so far.

Please feel free to offer more feedback on the backplane. Of the few that have sold, I'm curious about how you plan to use it, and I might be able to offer help.

PS. If you were interested in one of these adaptor cards mbandalauk, let me know!

Hi – yes, the sound is a bit hissy. I’m no audiophile either, but as a reference I’ve been comparing it with my WeeCee mini PC, which uses a CS4237 sound chip. That one sounds a bit cleaner, especially under Windows 98. Still, the hiss is barely noticeable once you’re actually using the SBC, so it’s not really an issue for me.

I might replace those capacitors at some point, but for now I’m happy with the backplane. I did know about using the EISA slots—I just forgot to mention it in my previous post. My SBC worked fine in both EISA slots, and it detected the sound chip without any problems. However, I preferred using the ISA slot since I wanted the SBC to be powered directly from the slot, which felt safer.

Honestly, I just wanted a backplane like yours for fun. The PCM-3370 works great, but I also plan to try another SBC I have: the IEI PCISA-3716EV-R4 VER 4.1. I’m just waiting on a P3-S 1400 Tualatin processor for it. If everything works as expected, I’ll be very happy!

Hi Tevian,

Thank you again for sending me your PC/104+ adapter PCB. I have just sourced the two PC/104 main connectors, which I thought would be all I needed to populate in order to test my PCM-3370 (although I may be mistaken). I would eventually like to fully populate the PCB, so would you be able to share a BOM? Some of the SMD footprints are unlabelled. While it is possible to make educated guesses, since some components appear obvious, I would prefer not to risk making any mistakes.

Thank you,
M

mbandalauk wrote on 2025-09-19, 15:46:

Hi Tevian, […]
Show full quote

Hi Tevian,

Thank you again for sending me your PC/104+ adapter PCB. I have just sourced the two PC/104 main connectors, which I thought would be all I needed to populate in order to test my PCM-3370 (although I may be mistaken). I would eventually like to fully populate the PCB, so would you be able to share a BOM? Some of the SMD footprints are unlabelled. While it is possible to make educated guesses, since some components appear obvious, I would prefer not to risk making any mistakes.

Thank you,
M

Check the private message logs here on Vogons. I sent the iBom file on one. I can also email it later.

As I mentioned in the DM, some of the values are missing in the iBom file as well. For instance, for the voltage status LEDs, you'll have to choose the values for each voltage based on what 0603 LED you buy. I can give you all the values I choose, but if you buy a different common 0603 LED, you might need different values. It's not critical, though. I'll post them if you really need them for reference... I'm attaching the iBom file here as well. You might have missed that I sent it earlier in the DM.

I purchased another PC104+ board. I got a PC104-P3-90C. It's a Pentium 3 system and I'm eager to test it. It came with no cabling, and I see no manual online for it. I was luckily able to find one pic that shows the connector functions, but I'll have to reverse engineer the pinouts. That could take a while. 😬

Thanks, apologies—I did not see the PM, not sure what happened there. That PC104-P3-90C looks like it has some very unconventional connectors. I really hope you managed to get the PIII version and not the Celeron. The same thing happened to me when I got my PCM-3370: I thought it was a PIII, but it came with a Celeron 400. Not a bad CPU, but definitely not a PIII!

It was $45 on eBay. A great buy, but I don't know yet if it's P3! Will know later tonight after work when I can mess with it.

😤😒😒

Well, I snooped the pinouts, got the monitor working as well as the keyboard. I was able to get into the BIOS and boot into DOS a few times. The BIOS is very custom with a ton of weird options... I attempted to plug it into the backplane to see if it could see the sound chip, but after I messed with a few BIOS options, I seem to have bricked the module!!

The BIOS battery was dead, so I removed the soldered-on one and replaced it. Very strange behavior now! All it will do now is very, very slowly count the RAM on power up. First, the low MEM, then the high MEM, which takes a couple of minutes! Like something is causing it to slow down or hang. It never goes past this, and I can't get into the BIOS now. I removed the battery and shorted the terminals, but the darn thing is a paperweight. I have no idea what I changed, but I can't imagine it was that critical.

It was working for a few minutes, but I have no idea what to do. I see no CMOS clear jumper, and I still can't find a proper manual for this PC104-P3. Ugh... I'm going to bed.

Hi Tevian,

Have you tried using any Nvidia PCI graphics cards? I have tested the three cards shown in the first photo: a Voodoo 3 2000, an Nvidia FX 5500, and a Zotac GT610 (which is an Nvidia-based card). The Voodoo 3 works flawlessly under Windows 98, and although Windows XP detects it, the unofficial drivers are quite temperamental. Since I am not particularly interested in using the Voodoo on XP, I focused on testing the other two cards instead.

Unfortunately, neither of them works. The FX 5500 has official drivers for both Windows 98 and XP, but it is not detected at all in Device Manager. The Zotac GT610, which is a high-end PCI card with XP driver support, also fails to be recognised. I have tried both of the PCI mappers you included, but the cards are still not detected. Both cards have been tested and confirmed to work on a standard desktop motherboard.

For reference, I am testing these on the PCISA-3716EV-R4 (second photo), which has proven to be excellent with your backplane. My setup uses a Tualatin Pentium III-S 1400 MHz CPU and 512 MB of RAM. It runs both Windows 98 and XP very smoothly, and as mentioned, the Voodoo 3 performs perfectly. However, the other two cards are not detected at all.

Do you have any advice or suggestions?

Thanks,
M

Interesting...

When I tested the backplanes, I first installed an older ATI All In Wonder graphics card into PCI (right slot). I made sure that the ATI card had the monitor and not the SBC. Also I made sure the BIOS had the option selected to use the PCI slot as the primary graphics.

I never tries something like dual monitor, SBC + PCI graphics though.

The main issue with getting the PCI cards working was, first, the proper IDSEL. If this is wrong, the machine will not detect the card. And, second, the proper INT# routing. This second part is not required for PCI function. For instance, using the SBC as the primary GPU, an SLI Voodoo 2 setup will work since the Voodoo 2 cards don't require an interrupt. However, things like primary graphics and network cards do require interrupts. In my testing, I can install a network card, have it detected and install the drivers under Windows, but have no network connection since the card can't report new incoming packets to the CPU.

The fact that the Voodoo 3 card works tells me the slot is functioning. Although, are you using all these cards as primary graphics, and have the BIOS option switched?

The best PCI graphics card I have is the ATI which worked well in my testing I would need to purchase a cheap newer PCI card to test further.

I don't have a good answer right now why it wouldn't work as long as another primary graphics cards worked under the same conditions. I'll see what I can do to test a newer PCI graphics card.

One thing of note, is the lack of SMbus support on the backplane. I don't think this is critical for GPUs, but I mirrored the PCISA and Allen Bradley backplanes as neither carry the SMbus lines to the PCI slots in favor of local termination at the PCI slot itself.

I don't think the SMbus is a critical component here. It's was in the PISA spec, but the PCISA backplane I have simply terminated the pins so I assumed it wasn't supported on PCISA SBCs.

Anyway, I'll see if I can get a PCI graphics card and test further, but I'm leaning toward some kind sort of BIOS issue/conflict.

Or..... The lack of 3.3V on the PCI slots!

Okay, if the card requires the 3.3V lines, this could be an issue. I never tested this, but the 3.3V net has a jumper near the 7-SEGs. It's a tiny solder jumper (JP5) that would connect the 3.3V bus power to the 3.3V PCI pins. These are normally left floating and I added them only for the sake of not knowing if they'd be needed since the backplane is hardwired for 5V I/O.

I know some card chips need a local 3.3V voltage rail to function, but they need to be 5V tolerant on the I/O at the same time.

I HAVE NOT TESTED THIS, but the solder jumper is there and only connects the dedicated 3.3V PCI pins to the 3.3V power supply line per the PCI spec.

I can test further when I get one of these cards...

Um... If I were testing I'd install the card and probe if the GPU IC is getting power. If not, it might require the 3.3V lines. This is tricky and you'd need to know where to poke.

Unless your willing to deal with the consequences, hold off until I get a card to test. 😅😅

This is the FX 5500 I am using — it works perfectly on a regular Socket 370 motherboard. The BIOS on the PCISA-3716EV-R4 includes an option to set the PCI slot as the primary display output. Regardless, I can switch displays from the Windows 98 display settings, although I have not yet tested both monitors working simultaneously.

In Windows XP, using the Voodoo 3 causes the system to freeze. This appears to be due to unstable Voodoo 3 drivers under XP. It does work, but only at lower resolutions — again, a driver-related issue. As mentioned before, the Voodoo 3 runs flawlessly on Windows 98, which was my main goal.

For XP, however, I wanted to use a more suitable graphics card. I do not currently have any ATI cards, but I will try to find one to test. I am also considering shorting jumper JP5 since it is easily accessible, but could you please confirm whether there is any risk of damaging the card or anything before I try that?

A couple of things...

I did find a "faster" PCI card in my stash. It's a GeForce MX4000 card that works fine with my PCISA-C400R-RS-R20 using the (PCISA ALT) Mapper card. This is a newer card, but is working fine without the 3.3V rails. I verified this by probing JP5, and it shows zero volts.

As far as risk, I don't suspect any major risk... I found this post -> Adding 3.3v to pci adapter(Released) about a 3.3V addon PCB, and also this VT vid about the Asus P5A super socket 7 mobo + GeForce 5500 not having the 3.3V -> https://youtu.be/U8W6A-VKroE?t=418

The backplane is wired for 3.3V as per the PCI pinout.

By soldering JP5, you connect the 3.3V ATX supply to the PCI 3.3V pins. The only pin left floating is the 3.3V AUX STBY on pin A14. I suspect standby 3.3V power isn't needed for most cases, but a wired jumper could be soldered.

I initially left it unconnected since both my example backplanes also left this open. PCI cards that needed 3.3V probably came later, and/or their use with backplane systems wasn't a priority as the PCISA spec was becoming obsolete. However, the universal type PCI cards (5V/3.3V) can work in both, but I'm discovering that some require the 3.3V for devices on certain cards, like RAM...

I've purchased a similar GeForce 5200 FX PCI, which seems to require the 3.3V. Those GT520,610 cards are a bit pricy!

Even without that card in hand, my gut is leaning towards jumpering JP5 as this mirrors the mods and 3.3V addon PCB I see online. My only question was whether the use also meant the need for 3.3V I/O (which isn't possible on my backplane), but from what I read, this is a non-issue for universal PCI types.

I did jumper JP5, even though my cards don't require it, and everything looks good. No adverse effects to report, and my cards perform no differently.

If you want to wait, I ordered the card, and it should be here by this weekend, but you might as well give it a try. 👍

Hi Tevian – no worries, everything’s sorted now. I shorted JP5, and both the FX5500 and the GT610 are working perfectly. I still need to properly install the drivers for the FX5500 on both Windows 98 and XP to test it fully.

However, I was able to install the NVIDIA drivers for XP for the GT610, and it worked like a charm! I even tested Deus Ex and Half-Life at high resolutions, and they ran fantastically. Turns out the 3.3V rail was the key. I also retested the Voodoo 3, and it’s still working perfectly—no issues there.

Thanks!

Awesome!

I'm glad I decided to add that in. ✌️