So, two problems with the build.

1) The power switch is odd. It isn't a momentary contact switch. I have to keep it jumpered to keep the power going. I can work around that with a latching switch.

2) When I tried to power on the board in the case, I got a flame! I powered it off quickly.

It had all be going so well. The right board. The right case. I even found an old I/O shield that worked. The saving grace is that I took the board out and it still works.

Maybe I accidently shorted the board during installation, but maybe the power supply is bad. =(

Hi does this chassis have pre-mounted mobo standoffs? Maybe you shorted something on the flip side?

I’m really surprised for such a modern case to have latching power switch. Maybe it’s just broken?

Low-profile ISA cards are not a thing (well there are small cards but you still have to 3D print a bracket yourself). For example some OPTi931 sound cards fit half-height chassis.
For half-height PCI video your best bet is GeForce 2MX/4MX models.
There are plenty of low-profile PCI Ethernet cards (good USB alternative for data transfer)

Bruno128 wrote on 2025-03-07, 19:21:

Hi does this chassis have pre-mounted mobo standoffs? Maybe you shorted something on the flip side? […]
Show full quote

Hi does this chassis have pre-mounted mobo standoffs? Maybe you shorted something on the flip side?

I’m really surprised for such a modern case to have latching power switch. Maybe it’s just broken?

Low-profile ISA cards are not a thing (well there are small cards but you still have to 3D print a bracket yourself). For example some OPTi931 sound cards fit half-height chassis.
For half-height PCI video your best bet is GeForce 2MX/4MX models.
There are plenty of low-profile PCI Ethernet cards (good USB alternative for data transfer)

The case doesn’t have a latching switch now, but I have a few spare latching switches that I use when I put an AT board in an ATX case. Instead of connecting it to the AT-ATX power adapter, I’ll take it to the motherboard.

I’ll check the power supply out of the case and if is good, I’ll reassemble everything and be extra careful.

p.s. You say low profile ISA isn’t a thing? Nobody told this guy! https://theretroweb.com/expansioncards/s/compaq-v-34-isa

I put everything back together carefully. The ATX IO shield was loose. I remember a loose shield touching the edge of a motherboard causing shorts a long time ago. I pushed it in carefully. This minuet case is a pleasure to work on compared to the AST LPX case I was working with last week. That case found a way to slice my fingers on a daily basis. So much bleeding. Hard to use your phone with bandages on both thumbs! The AM2 board that used to be in the case had 8 screws holding it in. New board is shorter and only touched 6 of the stands. I counted 9 screws. So I think that answers the cause of the short. One of the hard drive screws must have been in there.

The motherboard still works, but the onboard video has a very purple tint. I think I know what got damaged now.

Had a hard time finding a storage device that works with the UDMA enabled. Kingston 16GB MSata works. Windows 98 is installing. Blessedly, the motherboard supports booting from CD-ROM, because installing from floppies wasn't going to be fun and I had trouble

And I'm whipping up a second set of 3d printed low profile brackets. Turns out that low profile brackets are mirror images of full height. Did not know that.

Since it seems like I got the onboard UDMA to working, I've got the two cards picked out:

1) Geforce 4000MX, because I got one, and the integrated video output seems messed up.
2) NEC based USB 2.0 card, because I want to wire up the front panel. The completionist in me wishes I had firewire too, but I if I have to skip something, that would be it.

douglar wrote on 2025-03-08, 21:30:

I put everything back together carefully. The ATX IO shield was loose. I remember a loose shield touching the edge of a mother […]
Show full quote

I put everything back together carefully. The ATX IO shield was loose. I remember a loose shield touching the edge of a motherboard causing shorts a long time ago. I pushed it in carefully. This minuet case is a pleasure to work on compared to the AST LPX case I was working with last week. That case found a way to slice my fingers on a daily basis. So much bleeding. Hard to use your phone with bandages on both thumbs! The AM2 board that used to be in the case had 8 screws holding it in. New board is shorter and only touched 6 of the stands. I counted 9 screws. So I think that answers the cause of the short. One of the hard drive screws must have been in there.

The motherboard still works, but the onboard video has a very purple tint. I think I know what got damaged now.

Had a hard time finding a storage device that works with the UDMA enabled. Kingston 16GB MSata works. Windows 98 is installing. Blessedly, the motherboard supports booting from CD-ROM, because installing from floppies wasn't going to be fun and I had trouble

And I'm whipping up a second set of 3d printed low profile brackets. Turns out that low profile brackets are mirror images of full height. Did not know that.

Since it seems like I got the onboard UDMA to working, I've got the two cards picked out:

1) Geforce 4000MX, because I got one, and the integrated video output seems messed up.
2) NEC based USB 2.0 card, because I want to wire up the front panel. The completionist in me wishes I had firewire too, but I if I have to skip something, that would be it.

Jeez, how did you manage to get UDMA working on that thing? I tried both some old platter HDDs and some smaller SSDs with no luck. If you set the RAM divider to 4 instead of 3 that CPU will probably OC to 300mhz BTW. Also, the onboard video is kind of crap and has poor DOS compatibility. It's an impressive achievement they crammed most of the subsystem (only the DAC IIIRC is in the Cx5530 companion chip) along with the memory controller and some other stuff onto the same die with the CPU, but its not a good graphics option outside of basic office stuff.

Jackal1983 wrote on 2025-03-08, 23:40:

Jeez, how did you manage to get UDMA working on that thing? I tried both some old platter HDDs and some smaller SSDs with no luck. If you set the RAM divider to 4 instead of 3 that CPU will probably OC to 300mhz BTW. Also, the onboard video is kind of crap and has poor DOS compatibility. It's an impressive achievement they crammed most of the subsystem (only the DAC IIIRC is in the Cx5530 companion chip) along with the memory controller and some other stuff onto the same die with the CPU, but its not a good graphics option outside of basic office stuff.

Well, perhaps I abbreviated too much. What I should have said was that I finally found storage that the onboard IDE controller was able to detect the device while the BIOS was set to UDMA = auto. We'll have to see if I get the driver to work in Windows.

Doing the third print on the slot covert for the Geforce for the 3rd time. Holes needed shifted to the side by 2.5 mm.

I need a pair of 9pin USB headers to supply the front USB ports on the case. I was thinking of using one of these to provide the USB headers pictured below. Looks like the device takes a 9pin USB header and makes it into 4 USB headers. The device brings in two sets of USB signals from the usb header. Power and Ground look common and travel on the back of the card, but but on the front,one set of USB data signals goes to the end most half of the J2 connector and the other set of data signals goes to the hub-chip that does the other 7 USB sets of data pins on the card. I've got a Adaptec AUA-2000C card. I can solder a 4 pin to it for an internal port. So I could probably do OK if I plug those 4 pins into to the side that goes to the USB HUB, and get three fully working USB headers and 1/2 working header on J2.

The ECS P5GX-M is a bit cranky under windows 98se. Boots fast, Runs OK, fussy drivers.

And , no, it's not been agreeable to the idea of UDMA under windows in the least so far. DOS benchmarks suggest PIO4.

I am going to try to adapt an ALS120 based Asound Gold AS007 to give an AC-97 pinout for the front panel so I don't have to rout cables in from outside the back of the case.

While I'm waiting for the parts to arrive that I need to finish the build, I started putting together a counterpoint to the anachronistic MediaGX build.

It's a Dell OptiPlex GX Pro with a Pentium Pro 200. It was a top of the line workstation in 1996, intel through and through. Boy is it a pain. I like to use 16GB Kingston Msata storage devices, because I got a bag of 20 of them for very little money and they can be pretty fast. The MediaGX picked it up no trouble and booted from CD. Even after installing the latest BIOS, there's no CD boot and doesn't properly support LBA. Doesn't work with EZ drive either. I'll get it installed eventually. Might end up going back to SCSI.

I finally got back to work on this.

I ended up running across a USB 2.0 card with internal headers. This let me skip that "in-case usb hub thingie" that was kind of inelegant.
https://theretroweb.com/expansioncards/s/hp-ku-204in-ver-10

I had a low profile Geforce MX 4000 laying around--- https://theretroweb.com/expansioncards/s/jaton-82208-v2

I had two sound cards to choose from:
Labway A151-A00 w/ Yamaha YMF719 (OPL3-SA3) / ST / Inmos TEA2025 (Audio Amplifier)
https://theretroweb.com/expansioncards/s/labw … a00-ide-power-a
Asound AS007 w/ Avance Logic ALS120
https://theretroweb.com/expansioncards/s/d-b-asonic-as007

The Labway fit in the case by the slimmest of margins and had a better chip and through holes for audio headers, so I went with that.

I printed low profile brackets for the cards.

Added audio headers:

And packed the parts in the case:

Now I just need to find a driver for a random PCI bridge device and review how to make Win98se work well without a floppy drive.

I'm getting 75fps in Doom and 28FPS in quake, both at 320x200.

I have regrets about the storage speed with the integrated IDE. Maybe I should have used a PCI Sata controller instead of the PCI USB controller.

I made some updates:

• removed the USB 2.0 card
• replaced it with a generic Sil3112 Sata card https://theretroweb.com/expansioncards/s/syba-serialata-f02
• remounted the Kingston msata device without the sata / pata bridge
• added an internal USB Hub and connected it to the rear USB port through a hole in a custom bracket on the syba sata card
• tweaked the memory settings and disabled the unused devices in the bios

This is more better:

Those are not great scores for a sata device, but it is using UDMA now, which greatly improves the overall Win98 experience. The USB 2.0 was also slowing things down and I have to say that for what I'm doing, transferring small files, USB 1.1 is adequate.

Nice clean front with all the ports working except for firewire:

Here is the cable through the back: Looks like I need to press the IO cover plate into place a little better

And here's how the case opens up:

Looks much cleaner once the drive caddy flips back in.

I still need to figure out what is causing the unknown PCI bridge

It's a pretty cool system. It's like having a 266Mhz 486 with reasonably modern BIOS.

Here is the guts with the drive caddy in place, much cleaner. I added a CD-Audio cable after I took this picture.

Here's the pesky device:

Any advice on how to track down that device?

Any advice on how to stop the device driver search from looking any a non-existent A: every time it starts?

douglar wrote on 2025-05-23, 14:03:

Any advice on how to track down that device?

Any advice on how to stop the device driver search from looking any a non-existent A: every time it starts?

I think that has to do with the IDE controller. Attached the last UDMA driver for the platform, the readme file mentions this device, but I can't remember if it worked for me.

I downloaded "Unknown Devices" to get some insight-- https://www.majorgeeks.com/files/details/unkn … wn_devices.html

The unknown PCI Bridge was in the CPU.

• I located cx5530.inf in the driver file that you gave me, right click, install.
• After the reboot, I had to gointo to the APM and say "Force APM 1.0" to make it happy.
• I was also getting errors from the Standard Drive controller because I had one of the channels disabled. I had to go into the driver and say I was only using one channel and that took care of that.

Also, I put this in the autoexec:
subst A: C:\drivers

And that took care of the annoying delay every time device manager went looking for the floppy drive that I don't have.

Now I just need to put together a wave table daughterboard for the sound card.

p.s. I guess if I had looked at the readme file ...

1Since Windows9x doesn't provide support for NS5530 I/O chipset,
2so it recognizes the "PCI Bridge" as an "Unknow Device". 
3We offer the "PCI Bridge Driver" to solve the issue,
4please follow the instructions as below step by step.
5
61.Go to the "Control Panel/System".
72.Go to Device Manager,and Pick up "PCI Bridge".
83.Click "Properties/Driver/Update driver"
94.When the Winzard pop up,Select the file e:\p5gx-m\5530\cx5530.inf to install driver.