The missing parts are for a voltage regulator, though what voltage it generates I'm not sure of.

Tiido wrote on 2020-05-27, 04:41:

The missing parts are for a voltage regulator, though what voltage it generates I'm not sure of.

3.3V for cards that require that voltage?

I actually think negative voltage generation, the inductor and diode orientation seems to suggest that.

In any case I think it is unlikely to be required for ordinary cards.

All better cards use -12V present on PCI for the analog section and lack of it will result in probably very distorted sound or no sound at all.

Yes. But the -12V is present in the PCI connector, should come from the motherboard and no need to create in on a riser card. Also from which voltage it should be created?

With a better picture we should be able to tell for which voltage it is...

I googled this specific card and found some hits also without any components.

I guess this might be for a 3.3V. regulator. Since we are looking at 5V connector (if I get the orientation right) it might there as a provision for a handful of early PCI cards that used 3.3V before universal/3.3V standard was created.

Thinclient may have a PSU that only puts out +12V, +5V and +3.3V but not the -12V which is only ever used by COM ports and some sound cards.
One inductor leg going to ground and diode cathode going to the other leg strongly suggests the SMPS is wired to inverting configuration, in a positive regulator the inductor is in series with the output rail and diode connects differently.

Seeing where the output connects would solve the mystery for sure, my bet is on -12V rail.

Tiido wrote on 2020-05-29, 13:13:

Thinclient may have a PSU that only puts out +12V, +5V and +3.3V but not the -12V which is only ever used by COM ports and some sound cards.
One inductor leg going to ground and diode cathode going to the other leg strongly suggests the SMPS is wired to inverting configuration, in a positive regulator the inductor is in series with the output rail and diode connects differently.

Seeing where the output connects would solve the mystery for sure, my bet is on -12V rail.

I'm inclined to agree that this has to do with the -12V rail. First of all, both my Audigy 2 and my AW744L have a contact on the card edge for -12V connected (side B, pin 1) with traces leading away from them. Following those traces is difficult because of the dark solder mask. That said, at least on the AW744L, the -12V isn't used by the Sigmatel STAC9708T codec, because according to its data sheet, it only takes +5V.

As for whether the PCI slot provides -12V, I'll need to find my multimeter to check. Remember that the AW744L cards did not work in this motherboard when I put them into the PCI slot without the adapter either. If -12V is what these cards lack, then the fact that they did not work without the adapter implies that the PCI slot does not provide it.

The input to the board is a boring +12V/5A DC wall wart with a simple barrel connector. The CA10 board has two DB9 serial ports on it, each driven by a TI GD75232 chip. That chip requires +5V, +9V (range 7.5 to 15V), and -9V (range -7.5 to -15V). This implies that the PSU provides -12V to the board, but maybe not to the PCI connector? The maximum current in the PCI spec for -12V is 100 mA. Maybe the board couldn't guarantee a current that high, and they didn't route it to the PCI slot as a result?

I have attached clearer shots of both sides of the right-angle adapter. The -12V rail leads to the unpopulated R4 resistor pads, so this adapter as-is would never provide -12V, whether or not the PSU is supplying -12V.

I have a Neoware CA10 and it did not come with a riser card, I had to buy a cheap one on eBay, I got a "PCI Riser Card 1U 32bit REVERSE DIRECTION" and it has no place for capacitors or any thing else and "seems" to work just fine with the sound card I am using, I am using a ESS sound card, did you disable "all" audio options in the BIOS, there is a almost hidden MIDI option in one of the BIOS sub menus that you should disable, find the "Super IO Device" submenu for that MIDI setting, also you should reserve an IRQ for your PCI card , find the "Resources Controlled By" and set to manual, then in the "IRQ Resources" submenu set all of the IRQs to reserved except the one used by your PCI card, on my CA10 I have it set to "IRQ-5 assigned to PCI Device" for my ESS sound card, also you may have to set the "Reset Configuration Data" to Enabled once to clear out the old configuration settings, and BTW I also disabled the LAN but not sure if that helps.

Something else I noticed is that enabling the "AGP Fast Write", "AGP Master 1 WS Write", and "AGP Master 1 WS Read" seemed to help with 3DMark2000 scores when running under Windows 98.

will1384 wrote on 2020-05-29, 15:05:

I have a Neoware CA10 and it did not come with a riser card, I had to buy a cheap one on eBay, I got a "PCI Riser Card 1U 32bit REVERSE DIRECTION" and it has no place for capacitors or any thing else and "seems" to work just fine with the sound card I am using, I am using a ESS sound card

Nice to see someone else trying a similar build! Remember, as I said earlier, I tried my AWL744L in the slot without a riser card of any kind, and I still have tons of noise and almost no volume. That's equivalent to using a riser card that merely passes the signals through, like yours.

will1384 wrote on 2020-05-29, 15:05:

did you disable "all" audio options in the BIOS, there is a almost hidden MIDI option in one of the BIOS sub menus that you should disable, find the "Super IO Device" submenu for that MIDI setting, also you should reserve an IRQ for your PCI card , find the "Resources Controlled By" and set to manual, then in the "IRQ Resources" submenu set all of the IRQs to reserved except the one used by your PCI card, on my CA10 I have it set to "IRQ-5 assigned to PCI Device" for my ESS sound card, also you may have to set the "Reset Configuration Data" to Enabled once to clear out the old configuration settings, and BTW I also disabled the LAN but not sure if that helps.

Yes, yes, and no. I disabled all audio options, including the MIDI, and I reserved IRQ 5. I am keeping the LAN enabled.

will1384 wrote on 2020-05-29, 15:05:

Something else I noticed is that enabling the "AGP Fast Write", "AGP Master 1 WS Write", and "AGP Master 1 WS Read" seemed to help with 3DMark2000 scores when running under Windows 98.

I did not have WS Write or WS Read turned on; I will see if that makes Quake 2 run better. Thanks!

LightStruk wrote:

Nice to see someone else trying a similar build! Remember, as I said earlier, I tried my AWL744L in the slot without a riser card of any kind, and I still have tons of noise and almost no volume. That's equivalent to using a riser card that merely passes the signals through, like yours.

I wonder if its the sound cards that make the difference, I have a Sound Blaster Live 5.1 and a AOPEN AW744L that I could try, I had problems with both sound cards on other computers however, the Sound Blaster Live 5.1 is a DELL card and the drivers can be a bit hard to setup, and the AOPEN AW744L caused some games under DOS to lock up the computer, I will try both of them on the CA10 later today.

will1384 wrote on 2020-05-29, 16:52:

I wonder if its the sound cards that make the difference, I have a Sound Blaster Live 5.1 and a AOPEN AW744L that I could try, I had problems with both sound cards on other computers however, the Sound Blaster Live 5.1 is a DELL card and the drivers can be a bit hard to setup, and the AOPEN AW744L caused some games under DOS to lock up the computer, I will try both of them on the CA10 later today.

That's very generous of you! If you do try another card, you will know if you have a noise problem as soon as the computer turns on. I can hear noise caused by PCI bus activity pretty clearly in the BIOS, before any driver or OS has had a chance to intervene.

I think the sound card does make a difference... at least on the HP T5720 it does.

I have a bunch of different Yamaha YMF724 and YMF744 cards. A lot of them seem to work without the capacitors on the riser while a few, including the AOpen ones, do not. I've tried comparing the different cards and was never really able to tell why this is the case. My guess is that there is something I am missing because of them being multi-layer cards.

If you aren't getting -12v form the PC slot that would explain it on your CA10 though.

I installed a SB0100 and the drivers for DOS were a pain to setup, but it works, and no hiss and nothing strange, I did have to let the BIOS control the IRQs before I was able to get DOS sound working, I tested Quake 1, and Wolfenstein 3D, I had no sound problems in ether game.

And I tried a Ensoniq ES1370 sound card, with the old Ensoniq DOS drivers, and had no hiss or sound problems, I will say that the music under Wolfenstein 3D sounded strange but I think that is the bad sound card emulation, Quake 1 sounded just fine.

And I tried that AOPEN AW744L and had no hiss or strange sounds, Wolfenstein 3D had all the sounds, but Quake had no sound, Doom had no sound, and Rise of the Triad had no sound.

Could it be that your Neoware CA10 is different than my Neoware CA10, I took a picture of my CA10 and uploaded at imgur.com

and direct link:
https://i.imgur.com/CNqDDIC.jpg

Could it be your cables or speakers?

will1384, do you have the 1 GHz model or the 800 MHz model? I have the 800 MHz model. The most obvious difference visually between mine and yours is that mine lacks the DVI port. Otherwise our setups are very similar, including replacing the original flash storage with an SD IDE adapter!

There's R4 that should connect -12V from edge to the slot on the riser, put a solder bridge or 0ohm resistor on it and there should be low current -12V functionality on the output once more. 100mA is enough for the COM port level translator chip and shouldn't be a problem for the few opamps or something on some sound card if they use -12V for something.

LightStruk wrote on 2020-05-30, 01:28:

will1384, do you have the 1 GHz model or the 800 MHz model? I have the 800 MHz model. The most obvious difference visually between mine and yours is that mine lacks the DVI port. Otherwise our setups are very similar, including replacing the original flash storage with an SD IDE adapter!

The CA10 units I have came with XP embedded installed on a 512MB disk on a module card that I replaced that with a SD to IDE adapter, and have 1Ghz CPUs, and had 512MB memory in (2) 256MB memory DIMMs, but I removed one of the 256MB DIMMs for software compatibility.

An update for anyone watching this thread - I finally checked the voltages on the PCI slot, and the +12V signal rail is live, but the -12V rail is not.

I also checked the pins of the serial ports, and at least two pins had voltages like -10.5V.

Ideally, I would still rather populate the riser card correctly, since that's less cumbersome than running a bodge wire from somewhere on the motherboard. But that requires knowing what to populate it with.