Kahenraz wrote on 2022-10-28, 10:28:

Today was a repair day! Successfully repaired an excellent Socket 7 motherboard, with onboard S3 graphics, as well as an integra […]
Show full quote

Today was a repair day! Successfully repaired an excellent Socket 7 motherboard, with onboard S3 graphics, as well as an integrated Yamaha sound chip with General MIDI.

Re: Intel TC430HX Socket 7 motherboard repair

20221027_233751_resize_79.jpg

20221027_233812_resize_55.jpg

retouch_1666939210370.JPEG

20221028_024835.jpg

2022-10-28 06_27_04_resize_85.jpg

That's good work, a nice all-in-one board.

Spent the evening setting up another tower, this one a Pentium 120.

Added a dirty TAG to my SIS496 based motherboard and wrote a utility to enable it's usage. Now can run with an 8-bit TAG without having the SIS chipset run with "Force Dirty" which kills RAM performance.

Shponglefan wrote on 2022-10-29, 01:36:

Spent the evening setting up another tower, this one a Pentium 120.

That's a beautiful case. It's also great that the CPU speed indicator is three digits. I usually only see them with two.

I also see your Gravis Ultrasound hiding at the bottom. 😀

Kahenraz wrote on 2022-10-29, 02:39:

That's a beautiful case. It's also great that the CPU speed indicator is three digits. I usually only see them with two.

Thank you.

This and the previous 486-build cases both have an extra digit. Though that first digit is limited to a "1" on both. I keep hoping to find a proper full three digit LED so I can build a 200+ MHz system with the appropriate speed on the front. 😁

I also see your Gravis Ultrasound hiding at the bottom. 😀

Heh, that would be my other GUS Extreme. 😉

Though I don't know if I'll keep it in this particular build or not. It seems to be a bit more temperamental with this particular system.

The dreaded task of straightening pins on Socket 3 CPUs rescued from a gold scrapper, got two AM5x86-P75-133 CPUs that had some crazily bent pins, the CPUs work just fine after putting the pins right again !!

Shponglefan wrote on 2022-10-29, 01:36:

Spent the evening setting up another tower, this one a Pentium 120.

She's a beauty.

Just received some more toys. One of them being Mustek Twain-Scan Gray 800. As it included 3 floppy disks, i immediately backed them up into archive.org: https://archive.org/details/mustek-twain-scan-gray-800

A couple of months ago I got a new 120GB hard drive and set up my Prescott build for dual booting Windows 98SE and XP. Back then after installing the two Windows(es?) I moved on to other projects so it's left without drivers. Today I had some issues with Powerslide (details here) and my W98BOX suffers from HDD rot, I moved on to completing the setup. This will be my backup W98 PC.

The interesting thing is, it has 1GB RAM and rumors say unmodded Windows 98SE shouldn't be able to boot with this much, however this one shows no issues.

I made an error when reassembling a system on my test bench. I had connected the ATX power supply "first" without verifying that it was switched off. While inserting the Slot 1 CPU, the system spontaneously sprang to life.

I was worried about possible damage to the board, the CPU, or other components, and started testing using my most reliable coffee-spill CF card. I got to the BIOS screen, but it would fail to boot into DOS. I tried another CF card for Windows ME. which booted, but the system started freezing later while testing a PCI card. It seemed similar enough to the DOS boot problem that I thought this could be an indication damage somewhere. I even tried to boot DOS from a floppy, and had the same problem with DOS freezing at boot.

I spent the next couple of hours troubleshooting, swapping the CPU for an identical one, trying a different CF adapter, using an IDE adapter, swapping out the motherboard for an identical one (to keep everything consistent), etc. etc.

I then tired booting the CF card in a laptop and it worked fine. So the CF card was fine, but not in this desktop board. And every part tested the same when I swapped it out. I tried booting to gparted to look at the drive partition, but it was shown as unformatted. It booted fine in the laptop, but Scandisk complained about being unable to read the last block and that there *may* be an error.

To make a long story short, none of the hardware was damaged. I had modified config.sys on my DOS CF card to limit the RAM to 16MB when testing on another system and forgot about it. For some reason, this caused my desktop board to lock up when booting from it. The problem with Windows freezing at boot was just a driver issue that appeared at an inconvenient time. The reason Scandisk reported an issue on my laptop was a conflict with an LBA addressing flag and not a hardware issue. I have no idea why gparted saw the CF card as unfortmatted; this remains a mystery, as I have scanned the drive in Windows ME and Windows 10 and there are no problems with the FAT partition.

What about the floppy disk? That disk ended up being damaged and an unreliable source of testing. Another disk I tested afterwards booted to DOS just fine.

There was no damage from the inadvertent power-on by CPU insertion. Just a comedy of errors that made it appear that way.

Played around with audio options for my Pentium build. I had wanted to install a GUS Classic along side something else, but discovered the hard way that the GUS Classic is too bloody large to fit on any of the Pentium/Socket 7 motherboards I have. All of them have components on the boards which interfere with overly long ISA cards.

So I went with the GUS Extreme / Roland MPU401-AT combo I had in the 486 I was building, and plan to use the GUS classic in the 486 along side something else. Maybe an Orpheus I since that has genuine OPL3, plus a wavetable for MIDI.

I had a similar problem with my GUS Classic, back when I had one. I ended up cutting a hole in the metal frame of the case for it to fit through. There was just enough of a gap between the frame and the front facia that it could fit.

I don't have the GUS anymore. But I still have the case.

Kahenraz wrote on 2022-10-29, 21:08:

I made an error when reassembling a system on my test bench. I had connected the ATX power supply "first" without verifying that […]
Show full quote

I made an error when reassembling a system on my test bench. I had connected the ATX power supply "first" without verifying that it was switched off. While inserting the Slot 1 CPU, the system spontaneously sprang to life.

I was worried about possible damage to the board, the CPU, or other components, and started testing using my most reliable coffee-spill CF card. I got to the BIOS screen, but it would fail to boot into DOS. I tried another CF card for Windows ME. which booted, but the system started freezing later while testing a PCI card. It seemed similar enough to the DOS boot problem that I thought this could be an indication damage somewhere. I even tried to boot DOS from a floppy, and had the same problem with DOS freezing at boot.

I spent the next couple of hours troubleshooting, swapping the CPU for an identical one, trying a different CF adapter, using an IDE adapter, swapping out the motherboard for an identical one (to keep everything consistent), etc. etc.

I then tired booting the CF card in a laptop and it worked fine. So the CF card was fine, but not in this desktop board. And every part tested the same when I swapped it out. I tried booting to gparted to look at the drive partition, but it was shown as unformatted. It booted fine in the laptop, but Scandisk complained about being unable to read the last block and that there *may* be an error.

To make a long story short, none of the hardware was damaged. I had modified config.sys on my DOS CF card to limit the RAM to 16MB when testing on another system and forgot about it. For some reason, this caused my desktop board to lock up when booting from it. The problem with Windows freezing at boot was just a driver issue that appeared at an inconvenient time. The reason Scandisk reported an issue on my laptop was a conflict with an LBA addressing flag and not a hardware issue. I have no idea why gparted saw the CF card as unfortmatted; this remains a mystery, as I have scanned the drive in Windows ME and Windows 10 and there are no problems with the FAT partition.

What about the floppy disk? That disk ended up being damaged and an unreliable source of testing. Another disk I tested afterwards booted to DOS just fine.

There was no damage from the inadvertent power-on by CPU insertion. Just a comedy of errors that made it appear that way.

Wow that sounds like a hell of an afternoon chasing ghosts. Glad to hear all of your hardware is working properly. I had a similar thing happen with a PCI-E GPU I slotted it into the slot and the system powered up. Scared the hell out of me I couldn't pull the plug fast enough as I could of sworn I had flipped the switch before the upgrade. Thankfully just like in your case all the hardware worked perfectly after that.

Because I use very quiet fans and CompactFlash cards that make no sound, it can be hard to tell if my bench system is on without looking really closely, for some kind of LED or the spin of the CPU fan.

One of the things I do to help me identify whether the power is on, at a glance, is to always keep my POST card inserted. It has a couple of very bright red LEDs on it, plus the hex indicator.

In this case, I was still assembling everything. Which is how I got mixed up. I try to always flip the switch on the PSU and then press the power button, to ensure that everything is off, before removing a board from the bench. This is also good practice in general, as it ensures that all of the capacitors are drained.

RandomStranger wrote on 2022-10-29, 17:05:

A couple of months ago I got a new 120GB hard drive and set up my Prescott build for dual booting Windows 98SE and XP. Back then after installing the two Windows(es?) I moved on to other projects so it's left without drivers. Today I had some issues with Powerslide (details here) and my W98BOX suffers from HDD rot, I moved on to completing the setup. This will be my backup W98 PC.

W98SE.png

The interesting thing is, it has 1GB RAM and rumors say unmodded Windows 98SE shouldn't be able to boot with this much, however this one shows no issues.

No it will boot, but your mileage may vary with program and game compatibility when Win98SE hits the 512mb it was designed to handle and tries to use the other 512mb it wasn't designed to handle. My suggestion is to apply the memory patch to avoid possible issues later.

You can also install more than 512MB of memory and use different tricks to cap that amount at 512 for Windows 98. This is useful when you want to dual boot into XP, where that extra memory is more useful.

I have found 2GB to be the most stable, even when trying to cap the memory. The system that I have tested with had strange problems in Windows 98 with 3GB or more.

There's a lot of pain above 2GB in 2000 to 2008ish systems. Basically their device memory aperture mapping sucks. Later they managed to confine it to above 3.5GB... and then when 64bit gave them acres of address space, do an actual decent job of mapping around.

BitWrangler wrote on 2022-10-30, 05:12:

There's a lot of pain above 2GB in 2000 to 2008ish systems. Basically their device memory aperture mapping sucks. Later they managed to confine it to above 3.5GB... and then when 64bit gave them acres of address space, do an actual decent job of mapping around.

98Se in particular has a lot of issues when you try to use both >512mb of memory with a >512mb GPU, the OS simply was not designed to map such large amounts of memory so if you are using 98 then stick to GPUs with less than 512mb of memory, preferably use GPUs with 256mb or less for best compatibility.

Then I guess the severity of the issue can also change based on the system.

Nexxen wrote on 2022-10-03, 07:29:

That message is a known issue with ram above 512 (or 768) with Win98. Unless you downgrade, it'll always display that. […]
Show full quote

retrogamerguy1997 wrote on 2022-10-02, 23:43:

Nexxen wrote on 2022-10-02, 21:59:

If ram is > 512 downgrade to 256.

unfortunatly, I don't have a 256mb stick, just two 1GB sticks

That message is a known issue with ram above 512 (or 768) with Win98.
Unless you downgrade, it'll always display that.

There is a registry hack. Look for Win 98 1gb registry hack, it'll come up here IIRC.
I could be wrong on sizes, you have a starting point.

RandomStranger wrote on 2022-10-30, 06:09:

Then I guess the severity of the issue can also change based on the system.

Nexxen wrote on 2022-10-03, 07:29:

That message is a known issue with ram above 512 (or 768) with Win98. Unless you downgrade, it'll always display that. […]
Show full quote

retrogamerguy1997 wrote on 2022-10-02, 23:43:

unfortunatly, I don't have a 256mb stick, just two 1GB sticks

That message is a known issue with ram above 512 (or 768) with Win98.
Unless you downgrade, it'll always display that.

There is a registry hack. Look for Win 98 1gb registry hack, it'll come up here IIRC.
I could be wrong on sizes, you have a starting point.

That's indeed a possibility, newer chipsets that can handle 1gb+ natively I would imagine have less issues than older systems that couldn't, but 98SE itself will be the reason for many of the known ones that occur regardless of the hardware used, which can be patched out with 3rd party fixes and workarounds that hide the extra memory from 98SE.