So far so good. How large can these partitions be? I'm using a sata SSD.

Got windows installed tonight. Will play with installing things tomorrow.

BTW. I love the little beeps this things makes.

Yikes. I guess I need to start over. Windows installed okay but nothing else will install in windows. Direct x, daemon tools, service packs all give odd errors when I try to install them.

Maybe I made the partition too large? (250gb)

fluxrez wrote on 2024-01-07, 18:33:

Yikes. I guess I need to start over. Windows installed okay but nothing else will install in windows. Direct x, daemon tools, service packs all give odd errors when I try to install them.

Maybe I made the partition too large? (250gb)

I'm not sure, tbh! I think the largest I tried was a 120GB ssd. That worked perfectly at least. Did you use my slipstreamed version of Win98SE with the TBPLUS drivers?

Also: kudos on the case - looks awesome! 😁

Yes. I used your infused version of windows. I only took screenshots of the service pack errors. The other software errors all said the data was corrupt.

I also noticed during the first steps fdisk said the partition size was 48gb... So some oddness there.

I have a 2nd SSD that is 128gb. I'll swap to that.

All Part of the adventure!!

radivx wrote on 2023-12-12, 09:30:

For those thinking of going for the Coolermaster NR200P.
I am working on a 3D printable retro style kit for it based on the MACASE-cases:

https://imgur.com/a/HE0CajI

That looks really cool.

@fluxrez @Eivind

I've always had problems with the unofficial service pack 3.
TBPLUS should work fine on it's own without SP3.

I was unable to install anything in WIndows with curruption errors.

I wonder if the version of FDISK on the virtual floppy drive is an old version? When you go to specify a partition size its only 5 digits?? meaning maximum of 99 gb

I am trying XFDISK....

I find it's easier to do the formatting and file copying of the win98 folder on a modern with linux.

I then move it over to the old system, use a win98 boot floppy, go to c:\win98 and run setup /im /is /p i from there.

@Eivind

Does the ITX-llama have ACPI implemented?

Duffman wrote on 2024-01-08, 06:52:

Does the ITX-llama have ACPI implemented?

Nope.

I tried reinstalling with a new SSD. 128gb. Used premade windows installer but still got a blue screen. Rebooting worked, no safe mode was needed. After install I get similar glitches. Files don't install. Files don't copy..

Maybe I'll try the SD card option.
Could it be the power supply?

Oh, also the fan isn't spinning. It's a 5v noctua. Jumper is set to 5v.

I have been messing around with my board for a week or so now. I've run into a number of issues. Power delivery seems to be a factor, but I don't quite understand why. I started with a 200W mini ATX PSU I had lying around. I could not get any graphics card to work reliably with it. I swapped over to an identical PSU and it was exactly the same. I then pulled an ancient 300W supply out of and old machine and I was able to get one of the cards (an i740) to work 100% of the time. However, no other cards will work (they all test good in a different machine). This setup does not need anywhere near this much power, so I suspect it's more of a PCB distribution/filtering issue? I checked the power rails at the AGP and USB connectors with a scope and they seemed fine. There was a bit of noise, but nothing too concerning. I had ordered a picoATX type of supply and it just arrived. It also works just fine with the i740, but no other card will display anything (the system POSTs, just no video).

Another problem is the corruption issue you are seeing. It's not the destination drive that was the issue for me. It was the source. Installing Windows from the USB drive seems to go fine, but reading from a USB-hosted drive became very unreliable after that. Certain files either behave badly (BIOS gives some access denied error) or the data is corrupted. I slowed the clock down to 100MHz and it got a lot better, but still sometimes acted weird. I was able to copy the driver archives I needed without corruption and installed them. However, sometimes I still get access denied on a random file. I may reformat the flash drive and see if that changes anything. If there was a brownout or other weird issue, it's possible that the flash memory got messed up and is making the controller act weird. I haven't tried disabling the BIOS support for USB disks and falling back to the Windows 98 USB drivers yet.

I may try to add some capacitors somewhere on the AGP side of the board to see if that changes anything. USB only uses 5V and that rail looks rock solid, but HF stuff can be weird. At one point, I had video working on the original PSU with a Rareon 9800Pro. I decided to try this board because it had an external molex connector for extra power delivery and I suspected power might be a factor. This worked for exactly one day and then the next day it stopped working. The card still works fine in another machine, but not this one. So I feel like something transient is going on.

fluxrez wrote on 2024-01-08, 18:03:

I tried reinstalling with a new SSD. 128gb. Used premade windows installer but still got a blue screen. Rebooting worked, no safe mode was needed. After install I get similar glitches. Files don't install. Files don't copy..

These file errors, are they from and to the ssd itself, or from or to a USB or SD device?

fluxrez wrote on 2024-01-08, 18:03:

Maybe I'll try the SD card option.

You can certainly try that (warning: it's quite a bit slower than an ssd) - if everything works fine then it looks to be some kind of sata issue. One step at a time! 😀

fluxrez wrote on 2024-01-08, 18:03:

Could it be the power supply?

Hard to say, but my gut says no, unless you're using a terrible one... 😉

fluxrez wrote on 2024-01-08, 18:03:

Oh, also the fan isn't spinning. It's a 5v noctua. Jumper is set to 5v.

Did you enter the BIOS setup and play with the fan settings there? Also, did you try different fan connectors?

rknize wrote on 2024-01-08, 20:27:

I have been messing around with my board for a week or so now. I've run into a number of issues. Power delivery seems to be a factor, but I don't quite understand why. I started with a 200W mini ATX PSU I had lying around. I could not get any graphics card to work reliably with it. I swapped over to an identical PSU and it was exactly the same. I then pulled an ancient 300W supply out of and old machine and I was able to get one of the cards (an i740) to work 100% of the time. However, no other cards will work (they all test good in a different machine). This setup does not need anywhere near this much power, so I suspect it's more of a PCB distribution/filtering issue? I checked the power rails at the AGP and USB connectors with a scope and they seemed fine. There was a bit of noise, but nothing too concerning. I had ordered a picoATX type of supply and it just arrived. It also works just fine with the i740, but no other card will display anything (the system POSTs, just no video).

I find it weird that power should be an issue - especially if you checked the power rails at the AGP connector with a scope. If there was insufficient bulk capacitance, I think you'd see drops in voltage.
I tested the boards with a Radeon 9520 and a Voodoo3 3500 and never saw any of these issues, not sure about the power draw of these though.

rknize wrote on 2024-01-08, 20:27:

Another problem is the corruption issue you are seeing. It's not the destination drive that was the issue for me. It was the source. Installing Windows from the USB drive seems to go fine, but reading from a USB-hosted drive became very unreliable after that. Certain files either behave badly (BIOS gives some access denied error) or the data is corrupted. I slowed the clock down to 100MHz and it got a lot better, but still sometimes acted weird. I was able to copy the driver archives I needed without corruption and installed them. However, sometimes I still get access denied on a random file. I may reformat the flash drive and see if that changes anything. If there was a brownout or other weird issue, it's possible that the flash memory got messed up and is making the controller act weird. I haven't tried disabling the BIOS support for USB disks and falling back to the Windows 98 USB drivers yet.

Just so we'll be able to deduce the issue here - are the corruption issues you're seeing only regarding USB flash drives? Not affecting SD or SATA?

rknize wrote on 2024-01-08, 20:27:

I may try to add some capacitors somewhere on the AGP side of the board to see if that changes anything. USB only uses 5V and that rail looks rock solid, but HF stuff can be weird. At one point, I had video working on the original PSU with a Rareon 9800Pro. I decided to try this board because it had an external molex connector for extra power delivery and I suspected power might be a factor. This worked for exactly one day and then the next day it stopped working. The card still works fine in another machine, but not this one. So I feel like something transient is going on.

Adding a couple of big bulk capacitors probably wouldn't hurt, just to confirm or disregard that particular issue!

@fluxrez and @rknize: I'm so sorry you're having problems with your boards - like I've said many times, I did check every single board was working fine before shipping them out - but I guess I should have tested more thoroughly, with more different devices. Maybe I was just lucky with the selection of my flash drives and video cards...? 🙁

Eivind wrote on 2024-01-08, 21:02:

I find it weird that power should be an issue - especially if you checked the power rails at the AGP connector with a scope. If there was insufficient bulk capacitance, I think you'd see drops in voltage.
I tested the boards with a Radeon 9520 and a Voodoo3 3500 and never saw any of these issues, not sure about the power draw of these though.

Yeah, I agree. There may be more than one issue happening and perhaps there is something unique with my board. I also inspected the solder work under a microscope. I did find one AGP trace that looks odd, but I tested it for continuity and it was fine. The legs on the bridge chip also look fine. I suspect my no-video issue is just the card not getting detected at all. There was very little DC drop across the board. Running the CPU at lower/higher frequencies didn't change anything, though I suspect that has no effect on the PCI bus speed.

Eivind wrote on 2024-01-08, 21:02:

Just so we'll be able to deduce the issue here - are the corruption issues you're seeing only regarding USB flash drives? Not affecting SD or SATA?

It seems so. Copying files around the SATA drive works OK. When a ZIP file gets copied from the USB drive corrupted, it always fails CRC at the same offset. It is only reading from the USB drive that is a problem. Then again, I installed Windows 98 from this same drive and it worked fine. Very odd. Some interaction with the IDE driver? I don't know how the USB drive presents itself to Windows in this mode. The SD card seems to work fine, but I only have DOS installed there.

Eivind wrote on 2024-01-08, 21:02:

Adding a couple of big bulk capacitors probably wouldn't hurt, just to confirm or disregard that particular issue!

Yeah, it's possible that there is something happening at a very HF that my scope can't see. I need to do some reading on PCI/AGP to get some idea of what to look for.

rknize wrote on 2024-01-08, 21:15:

Yeah, I agree. There may be more than one issue happening and perhaps there is something unique with my board. I also inspected the solder work under a microscope. I did find one AGP trace that looks odd, but I tested it for continuity and it was fine. The legs on the bridge chip also look fine. I suspect my no-video issue is just the card not getting detected at all. There was very little DC drop across the board. Running the CPU at lower/higher frequencies didn't change anything, though I suspect that has no effect on the PCI bus speed.

The CPU draws very little power, only like 3-4 W or something if I remember correctly.

rknize wrote on 2024-01-08, 21:15:

Eivind wrote on 2024-01-08, 21:02:

Just so we'll be able to deduce the issue here - are the corruption issues you're seeing only regarding USB flash drives? Not affecting SD or SATA?

It seems so. Copying files around the SATA drive works OK. When a ZIP file gets copied from the USB drive corrupted, it always fails CRC at the same offset. It is only reading from the USB drive that is a problem. Then again, I installed Windows 98 from this same drive and it worked fine. Very odd. Some interaction with the IDE driver? I don't know how the USB drive presents itself to Windows in this mode. The SD card seems to work fine, but I only have DOS installed there.

For good measure, maybe you could try running some kind of disk test on a USB drive from DOS? Just curious to see if Windows has anything to do with this.
Also - what kind of size of USB drive are we talking about here? And how big a partition and what program was used to create that partition?

rknize wrote on 2024-01-08, 21:15:

Eivind wrote on 2024-01-08, 21:02:

Adding a couple of big bulk capacitors probably wouldn't hurt, just to confirm or disregard that particular issue!

Yeah, it's possible that there is something happening at a very HF that my scope can't see. I need to do some reading on PCI/AGP to get some idea of what to look for.

The AGP slot power rails are literally peppered with capacitors, many small ones (0.1uF) as well, so I'd be surprised if HF power dips are the issue. But I wouldn't rule anything out, of course. You don't have access to a Radeon 9200/9250, just for comparison with my testing?

Eivind wrote on 2024-01-08, 21:35:

The CPU draws very little power, only like 3-4 W or something if I remember correctly.

Yeah, I could see from my bench supply that the entire board (including the i740) draws less than an amp on average when using the picoATX. So something like 10W.

Eivind wrote on 2024-01-08, 21:02:

For good measure, maybe you could try running some kind of disk test on a USB drive from DOS? Just curious to see if Windows has anything to do with this.
Also - what kind of size of USB drive are we talking about here? And how big a partition and what program was used to create that partition?

I can try that. I'm using an 8GB SanDisk card. I was also using another 8GB Verbatim card at one point, but not when I was experiencing these issues.

rknize wrote on 2024-01-08, 21:15:

The AGP slot power rails are literally peppered with capacitors, many small ones (0.1uF) as well, so I'd be surprised if HF power dips are the issue. But I wouldn't rule anything out, of course. You don't have access to a Radeon 9200/9250, just for comparison with my testing?

I don't have a 92XX, no. I recently sent a bunch of AGP cards to the scrapper, but most of them were middle-of-the-road nVidia cards that needed work (leaky caps). I'm still not clear on why the 9800Pro worked fine and then suddenly stopped working. I also have a Voodoo3 3000 that was kind of working when I first started testing and now doesn't work at all. I was trying them again yesterday and neither is detected now. The only difference is that the Voodoo card sometimes prevents the system from posting (no beep). Other times it posts, but there is no video. It's true that all the cards I am trying are old and probably tired. While the caps on them show no outward signs of distress, it could be that they are marginal and my other test system is able to overcome this.

It would be handy if the BIOS beeped a different way to complain if the card is not detected. If it is detected but just isn't producing video, it might point to the card itself being unhappy rather than the bus.

One other interesting datapoint: I have noticed that from time-to-time, the ATTINY stops managing the LED properly and acts strangely. I can toggle the power, but the LED will not light up in any color. In one instance, the LED was flashing at around 30Hz or so. It showed the correct colors (red when off, green when on), but was flashing. I had to completely remove power from the board to get it to work properly again. I noticed it appeared to be using 3.3V (I would expect the AT to be on the 5V standby?). I added a 10uF tantalum bypass cap across the AT. We'll see if that makes any difference long term. It didn't help the AGP detection issues at all, but I didn't think it would. I'll add some caps under the AGP slot itself and see if it changes anything.

I was tracing out the AGP slot and taking some measurements with the intent of adding some bulk capacitance to the 3.3V rail. I noticed that, for example, pin 9 only gets 3.3V when there is a card in the slot and also that it measures more like 3.1V. I am wondering if this supply is derived or if pin 9 is getting supplied by the card itself via one of the other 3.3V pins? I guess I'll wait for the schematics to be released. 😀

Edit: I checked the other 3.3V pins and they are the same. To be more specific, the inboard side of FB8 seems to be the 3.3V rail and the outboard side of the ferrite bead is switched (sees ~3.1V when the i740 is present). When I put the Voodoo or Radeon card in there, the 3.3V supply does not appear on any of the expected AGP pins. It rises to maybe 0.5V.

The board arrived today. It looks like a masterpiece 😉. You are a god 😁.

I'm going to test the board on graphics
1) Geforce 4 MX 440 SE 64MB
2) Ati Mobility 8MB
3) Radeon 7500 64MB
4) Riva TNT 2 32MB

All cards are new and never used so I assume if something doesn't work it won't be the graphics.

rknize wrote on 2024-01-09, 18:19:

I don't have a 92XX, no. I recently sent a bunch of AGP cards to the scrapper, but most of them were middle-of-the-road nVidia cards that needed work (leaky caps). I'm still not clear on why the 9800Pro worked fine and then suddenly stopped working. I also have a Voodoo3 3000 that was kind of working when I first started testing and now doesn't work at all. I was trying them again yesterday and neither is detected now. The only difference is that the Voodoo card sometimes prevents the system from posting (no beep). Other times it posts, but there is no video. It's true that all the cards I am trying are old and probably tired. While the caps on them show no outward signs of distress, it could be that they are marginal and my other test system is able to overcome this.

There seems to be something fishy going on hardware-wise, for sure. And it might very well have to do with power...

rknize wrote on 2024-01-09, 18:19:

I was tracing out the AGP slot and taking some measurements with the intent of adding some bulk capacitance to the 3.3V rail. I noticed that, for example, pin 9 only gets 3.3V when there is a card in the slot and also that it measures more like 3.1V. I am wondering if this supply is derived or if pin 9 is getting supplied by the card itself via one of the other 3.3V pins? I guess I'll wait for the schematics to be released. 😀

Yeah I think I should get my act together and ready my schematics and layout soon now! 😁 All 3.3V power pins on the AGP slot comes from the main 3.3V rail from the ATX connector - some passing through FB8 and some through FB9.

rknize wrote on 2024-01-09, 18:19:

Edit: I checked the other 3.3V pins and they are the same. To be more specific, the inboard side of FB8 seems to be the 3.3V rail and the outboard side of the ferrite bead is switched (sees ~3.1V when the i740 is present). When I put the Voodoo or Radeon card in there, the 3.3V supply does not appear on any of the expected AGP pins. It rises to maybe 0.5V.

So... reading your comments and thinking about this, I suspect the ferrite bead FB8 is either broken or not making good contact with its pads.
First of all - the ferrite bead itself should only drop the voltage a miniscule amount, like 0.01V in my measurement. The fact that you're both not seeing any voltage there when there's no AGP card inserted, and a drop of 0.2V when there is looks to me like that rail "VCC_3V3_AGP" is being backfed from the other 3.3V AGP rail, the "VCC_VIO_AGP" through the card itself. Which is definitely not good, but might explain the card-dependant erratic behavior you're seeing - as these are supposed to be separate rails providing power to the card.
I'll attach a screenshot of the AGP schematic here for now.

Using a voltmeter, try measuring resistance through the ferrite bead, and play around with touching either the bead leads themselves, or the tiny pads it's attached to.
You could also try desoldering the FB8 bead and bridging the gap with solder directly.

rknize wrote on 2024-01-09, 18:19:

One other interesting datapoint: I have noticed that from time-to-time, the ATTINY stops managing the LED properly and acts strangely. I can toggle the power, but the LED will not light up in any color. In one instance, the LED was flashing at around 30Hz or so. It showed the correct colors (red when off, green when on), but was flashing. I had to completely remove power from the board to get it to work properly again. I noticed it appeared to be using 3.3V (I would expect the AT to be on the 5V standby?). I added a 10uF tantalum bypass cap across the AT. We'll see if that makes any difference long term.

I have seen something similar earlier when testing, but couldn't quite nail down the issue or reproduce it reliably. I suspect there's some sort of timing/race condition going on in my ATTiny code - I'll try playing around with it this weekend!
The chip uses 3.3V, yes - but that's directly generated from the +5VSB rail from the ATX through the U10 LDO.