I decided to manage the cables in my main retro system earlier today, now that I have an AG430HX on the way. Might as well make sure I have the room for a semi-modular setup.

I eliminated an extra PATA cable, managed to use the power cables for the GOTEK and IDE-SD combo I have in the top drive slot to keep the floppy cable out of the way, and managed to pull cables together in a way that didn't try to bugle the back of the case out. I know my combo cable for the CD audio looks nasty, but it'll be useful when I have the Yamaha and AWE64 running side by side.

Revived a PC-Chips M748MR motherboard I was given 10 years ago along with a Pentium II CPU and RAM, all of which I used for a little bit and then stored for all these years. Thought it was dead (my POST card didn't say anything probably because it wasn't making contact with the crusty old PCI slots) but I hooked up my logic analyzer - a poor man's oscilloscope - and saw activity on the BIOS data lines, so it wasn't entirely dead. Gave it new SDRAM and it booted right up.

Immediately noticed two mistakes I made 10 years ago:
1) Assume the CPU is a 233 (66*3.5) because the jumperless BIOS defaulted to that setting. I looked up the sSpec and it turned out to be a 350 (100*3.5).
2) Wire up a USB bracket with shielding to the ATX form card connector, resulting in magic smoke and no USB power. This board has the same stupid design as other PC-Chips boards where the shield pin is wired to 5V, which fries the Vbus circuit and now it's only outputting about 2.5V. I plan on fixing it by cutting the Vbus rail and taking 5V from elsewhere, but for now, jumper wires will do.

The AMI BIOS is quite bad - even on the latest version, it doesn't auto-detect the CPU bus speed (as mentioned above) and just hangs with >32GB hard drives. Just out of curiosity, I looked for other similar boards to try and borrow a BIOS from, and ended up finding a 128GB patched Award BIOS for the Matsonic MS7380SG which works great. (The ECS P6SET-ML BIOS boots once, then after a reboot it emits a distress signal and only works again after reflashing)

The board itself is a bit finicky - the stock BIOS often has trouble detecting hard drives, the Matsonic BIOS sometimes fails to boot after I set the CPU frequency, hardware monitoring reports up to 2.8V on the 2.5V line, and I'm struggling to get Windows XP installed (for some emulation-related research work, long story). The caps look fine but I'll probably end up recapping it anyway. I've already taken a soldering iron to this board for IDE pin 20 removal surgery because all my IDE cables are 39-pin.

Almoststew1990 wrote on 2020-10-24, 16:53:

The Violet V.... *can't think of a cool name beginning with V* .......Violet Verisimilitude (great word) is still going strong! Now with my DX4-100, STB Lightspeed 128 and SB 2.0 (because of Phil's video the other week).

I challenge you to paint your keyboard too! =)

Not exactly today, but on my 98se machine I installed an upgrade copy of office 2000 premium over office 97. But since I didn't want to burn a million CDs to burn 4 iso images, I did it by having one iso image at a time on a 1GB flash drive and mounting the image and installing it that way. disc 1 installed fine, but restarting after disc 2 asked for the CD which I couldn't provide since I couldn't mount the image but clicking cancel didn't seem to hurt anything. Disc 3's setup looked more like office 97's installer than office 2000's installer. disc 4 didn't have an installer and it was just clipart and stuff like that.

After last time's success, today I repaired a Radeon VE that also had bad memory. This time the chip donor was a borked Seagate 7200.12 disk (parts are pretty much the only thing F3 Seagates are good for).

And then made some more useful DIMMS, 128 and 256MB from 2 single sided 64 and 128MB modules respectively.

PROTIP: If you are thinking of doing this yourself, don't. If you really want, then Capton the edge connector and use lots of flux. And remember to edit the SPD to suit the new layout.

Finally decommissioned the mother's friend's C2D E6500 (2.93GHz) with low end chipset motherboard with not activated win 7 (boggles!) which luckily begun to die but I got the data moved to far more newer computer with SSD, much faster and with windows 10 activated.

Means I can recycle the case which I don't care but that good for now, for vintage computing at last.

Cheers!

RichardG867 wrote on 2020-10-25, 04:35:

Revived a PC-Chips M748MR motherboard I was given 10 years ago along with a Pentium II CPU and RAM, all of which I used for a li […]
Show full quote

Revived a PC-Chips M748MR motherboard I was given 10 years ago along with a Pentium II CPU and RAM, all of which I used for a little bit and then stored for all these years. Thought it was dead (my POST card didn't say anything probably because it wasn't making contact with the crusty old PCI slots) but I hooked up my logic analyzer - a poor man's oscilloscope - and saw activity on the BIOS data lines, so it wasn't entirely dead. Gave it new SDRAM and it booted right up.

Immediately noticed two mistakes I made 10 years ago:
1) Assume the CPU is a 233 (66*3.5) because the jumperless BIOS defaulted to that setting. I looked up the sSpec and it turned out to be a 350 (100*3.5).
2) Wire up a USB bracket with shielding to the ATX form card connector, resulting in magic smoke and no USB power. This board has the same stupid design as other PC-Chips boards where the shield pin is wired to 5V, which fries the Vbus circuit and now it's only outputting about 2.5V. I plan on fixing it by cutting the Vbus rail and taking 5V from elsewhere, but for now, jumper wires will do.

The AMI BIOS is quite bad - even on the latest version, it doesn't auto-detect the CPU bus speed (as mentioned above) and just hangs with >32GB hard drives. Just out of curiosity, I looked for other similar boards to try and borrow a BIOS from, and ended up finding a 128GB patched Award BIOS for the Matsonic MS7380SG which works great. (The ECS P6SET-ML BIOS boots once, then after a reboot it emits a distress signal and only works again after reflashing)

The board itself is a bit finicky - the stock BIOS often has trouble detecting hard drives, the Matsonic BIOS sometimes fails to boot after I set the CPU frequency, hardware monitoring reports up to 2.8V on the 2.5V line, and I'm struggling to get Windows XP installed (for some emulation-related research work, long story). The caps look fine but I'll probably end up recapping it anyway. I've already taken a soldering iron to this board for IDE pin 20 removal surgery because all my IDE cables are 39-pin.

You sir are amazing. You diagnosed a motherboard using an oscilloscope, upgraded it with a different BIOS, and plan on replacing the +5v USB rail. That's just awesome. Please post pics when you actually rework the USB power. I want to see that.

andrea wrote on 2020-10-25, 21:37:

After last time's success, today I repaired a Radeon VE that also had bad memory. This time the chip donor was a borked Seagate […]
Show full quote

After last time's success, today I repaired a Radeon VE that also had bad memory. This time the chip donor was a borked Seagate 7200.12 disk (parts are pretty much the only thing F3 Seagates are good for).

And then made some more useful DIMMS, 128 and 256MB from 2 single sided 64 and 128MB modules respectively.

PROTIP: If you are thinking of doing this yourself, don't. If you really want, then Capton the edge connector and use lots of flux. And remember to edit the SPD to suit the new layout.

That's impressive. If you don't mind, what made you want to "build" DIMMs yourself?

Freaking finally got my Terriblefire TF534 working. It kept just getting a black screen at boot. (building this from scratch) I did some very close inspections with the USB microscope and found that one leg on a single RAM chip was not soldered fully and had like a human hair gap.. Haha. Well, it works now. 😁 Sooo excited.

TechieDude wrote on 2020-10-25, 23:47:

andrea wrote on 2020-10-25, 21:37:

After last time's success, today I repaired a Radeon VE that also had bad memory. This time the chip donor was a borked Seagate […]
Show full quote

After last time's success, today I repaired a Radeon VE that also had bad memory. This time the chip donor was a borked Seagate 7200.12 disk (parts are pretty much the only thing F3 Seagates are good for).

And then made some more useful DIMMS, 128 and 256MB from 2 single sided 64 and 128MB modules respectively.

PROTIP: If you are thinking of doing this yourself, don't. If you really want, then Capton the edge connector and use lots of flux. And remember to edit the SPD to suit the new layout.

That's impressive. If you don't mind, what made you want to "build" DIMMs yourself?

I needed a 256MB DIMM and was too cheap to buy one. So I did an experiment with the useless 64MB sticks first to see if my idea worked and, if it did, to have some sort of fallback in case I later borked the 128MB modules.

andrea wrote on 2020-10-25, 21:37:

<snip>

PROTIP: If you are thinking of doing this yourself, don't. If you really want, then Capton the edge connector and use lots of flux. And remember to edit the SPD to suit the new layout.

You can't show us those nice results and then say _that_... 😉

Where did you pick up these skills? Professionaly? Years of experience? Any guides you might recommend?

Cyrix200+ wrote on 2020-10-26, 09:21:

andrea wrote on 2020-10-25, 21:37:

<snip>

PROTIP: If you are thinking of doing this yourself, don't. If you really want, then Capton the edge connector and use lots of flux. And remember to edit the SPD to suit the new layout.

You can't show us those nice results and then say _that_... 😉

Where did you pick up these skills? Professionaly? Years of experience? Any guides you might recommend?

Nowhere 😀.
It all started I want to say around 2008 I guess, where i got for free 3 Athlon XP machines (back then these were old but still usable as dailies, a bit like C2Ds today), that died of bad caps. So I learned to solder out of necessity.
As for guides, on Youtube there is (or at least used to be there) a full Pace videocourse on soldering that's pretty nice.

This job was done with a cheap Yihua 937 solder station and a way too big fake Hakko chisel tip. What I did not say was that i had to reflow the chips like 3 times before I got them to pass memtest.

Over the last few days I did some BIOS modding for my Diamond Micronics C400 motherboard, which is identical to the BCM QS440BX. While Diamond stopped shipping BIOS updates at version 1.06, BCM continued to improve the BIOS for their board - but only for the later revisions with a 2 MBit flash chip. Seems they didn't want to bother with the space constraints on the 1 MBit chip anymore.

Since Diamond used the 1 MBit variant, I could not use the much newer and improved BCM BIOS version 2.10 on my board. So I salvaged a 2 MBit chip from another (dead) Slot 1 board and did a hot-flash, which to my surprise worked on the first attempt.

While I was at it, I modded the BIOS:
- Used the patched BCM QS440BX image from wimsbios with 128 GB HDD support as a base
- Inserted the full-screen logo (Diamond/Micronics) that I had extracted from the original C400 BIOS image
- Created and inserted a Diamond logo to replace the EPA logo
- Added various microcode updates so I can for example run my Tualeron 1400/100 in a modified Slotket
- Modified the BIOS string to show the Diamond C400 name instead of BCM

gex85 wrote on 2020-10-26, 12:57:

Over the last few days I did some BIOS modding for my Diamond Micronics C400 motherboard, which is identical to the BCM QS440BX. […]
Show full quote

Over the last few days I did some BIOS modding for my Diamond Micronics C400 motherboard, which is identical to the BCM QS440BX. While Diamond stopped shipping BIOS updates at version 1.06, BCM continued to improve the BIOS for their board - but only for the later revisions with a 2 MBit flash chip. Seems they didn't want to bother with the space constraints on the 1 MBit chip anymore.

Since Diamond used the 1 MBit variant, I could not use the much newer and improved BCM BIOS version 2.10 on my board. So I salvaged a 2 MBit chip from another (dead) Slot 1 board and did a hot-flash, which to my surprise worked on the first attempt.

While I was at it, I modded the BIOS:
- Used the patched BCM QS440BX image from wimsbios with 128 GB HDD support as a base
- Inserted the full-screen logo (Diamond/Micronics) that I had extracted from the original C400 BIOS image
- Created and inserted a Diamond logo to replace the EPA logo
- Added various microcode updates so I can for example run my Tualeron 1400/100 in a modified Slotket
- Modified the BIOS string to show the Diamond C400 name instead of BCM

That's seriously awesome I wish I had the skills to pull that off.

chrismeyer6 wrote on 2020-10-26, 13:00:

That's seriously awesome I wish I had the skills to pull that off.

It's not black black magic 😉

There are a few different tools that you need:
- CBROM (fairly powerful command line tool for editing Award BIOS images, but many tasks are easier with the graphical Award BIOS Editor)
- Award BIOS Editor: http://awdbedit.sourceforge.net/ to extract and insert parts of the image, modify the strings, configure BIOS menu items, etc.
- CTMC microcode utility (to split the extracted microcodes from various BIOS images into separate files that can then be combined into a new file that contains all the updates)
- Pixelformer to create the 4-bit BMP Logo
- EPAcoder v1.50 (to convert regular 4-bit BMP files to the "EPA" format that is used for the BIOS logos)

The process is roughly as follows (no warranty if you brick your board):
1. Obtain latest BIOS from manufacturer or an already patched image that removes HDD size limitations etc. from a site like wimsbios.com
--- Logo modifications ---
2. Extract EPA logo with CBROM or Award BIOS Editor and save as Windows BMP
3. Craft your own EPA logo with Pixelformer or any other tool that can save to 4-bit BMP
4. Convert the logo with EPAcoder
5. Insert the converted file into the BIOS image
6. Repeat steps 2-5 for full-screen logo
--- Microcode updates ---
7. Extract the microcode part from the original BIOS image
8. Obtain a BIOS file for a Tualatin-capable board like the Abit ST6 and extract its microcode part
9. Use CTMC to store the microcode updates in separate files on a per-cpuid-basis
10. Copy all the microcode updates you want to include into a single file
(10a. For Asus boards there seems to be a special header needed in the microcode file, be aware of that if your board is an Asus)
11. Use Award BIOS Editor or CBROM to replace the microcode part with your newly crafted file
--- Other updates ---
12. Edit BIOS menu settings, strings etc. to your liking
--- Flash ---
13. Use AWDFLASH to actually flash your modified image

The real art of BIOS modding however begins with removing the HDD limits etc., this is something far beyond my knowledge...

Thank you for the step by step for the software part. I wish my soldering skills were up to the task and I use the word skills loosely. Most of my soldering is sweating pipes.

Not exaxtly retro yet, but I modded the bios of my Radeon 7970 to run at the clocks of the 7970 Gigahertz Edition. No change in voltage required.

After replacing the thermal paste, the card runs at basically the same fan speed and temperature as the regular clocked version, and total system power consumption is up by about 20 watts under load. That's a much better result than the reviews from back in the day, making the card look like a loud inefficient power guzzling hog.

Edit: on the other hand, if you download one of the modded bioses available online, they use a higher voltage, and the card truly becomes a loud hot mess, increasing the total system load power consumption by 100 watts!

Is it fully stable at the higher speeds and the stock voltage?

chrismeyer6 wrote on 2020-10-26, 14:57:

Is it fully stable at the higher speeds and the stock voltage?

Yes, I tested it first by overclocking in Afterburner.

Excellent. That's a nice increase with not much extra heat

I got a bunch of stuff locally:

There's a rather nice 386 SX-40 which looks like it was upgraded as there's an 8 bit ATI VGA/EGA card in there, a 44 MB MFM drive, other hardware that seems to fit better in an XT or early AT, and surprisingly the battery holds a date and hasn't leaked (will be removing it regardless). Floppies both work too, high density units.

Doesn't work - spins up lights flash, then spins down. I put in the CHS parameters in BIOS, the drive spins down and then gives a C drive error. I'll play with it some more later.

So the VLB I/O cards don't detect IDE drives (at least on the supplied motherboards) otherwise fine. The 8 bit floppy controller gives a 1701 error, the 16 bit I/O cards seem to work fine as well as the two ISA video cards. The Mach32 initially showed garbage, but I reseated it and it worked. It also displayed properly with the second motherboard. Motherboards are 486 VLB, one with a 33MHz DX chip, the other with an OverDrive 486 DX 33 (shows as a 66Mhz in tests - will be removing that one for my personal collection and replacing a standard DX2/66, as I have a few). They both need batteries, one had a barrel removed (extremely minimal leaking, I think I might be able to solder to the battery points) and the other one is an integrated RTC. Which is soldered on. And it's the type that won't let you get past POST on a dead battery. Fun!

The other box is a working P166 MMx - that'll be eBayed to cover the costs of all the other stuff.

I also got a chance to work with my IBM Blue Lightning - it's been sitting in my basement since I moved. I'll start a separate thread on that as I have specific questions.

All in all not a bad haul 😉