Tore apart my TI-99/4A today so I could install the Pico9918 mod. Now I've got full F18A emulation and glorious 640x480 @60Hz output! It's a beautiful thing.

I think I'm done modding it... for now. 😜

I was thinking the partition Id byte corrupted.
9x might not care if fat16 is labelled fat32.
Dos will

RetroLizard wrote on 2026-10-24, 03:41:

classic_cola wrote on 2026-10-24, 03:38:

Appreciate that, I'll certainly keep in mind next time a CF starts ghosting me. […]
Show full quote

Trashbytes wrote on 2026-10-24, 03:03:

Going to go out on a limb here and suggest you test the CF adapter and CF card on another known good system .. I've had a few CF Adapters just give up the ghost yet still be visible to the host system, same with the CF cards .. they do have a reputation for just going dead but appear to still be working.

Appreciate that, I'll certainly keep in mind next time a CF starts ghosting me.

Fortunately, after much hair pulling it's back to life using FDISK option 5 to set the CF as the "fixed disk" - then deleting the partition completely, and repartitioning, has brought it back to life.

Ultimately this is probably a skill issue on my part; I prepared this partition in Windows 11 a week or two ago which probably did something weird to the partition layout, I thought FAT16 is FAT16 whether prepared in modern Windows or not. Turns out I'm probably wrong.

Weird thing is it actually worked for a week like this. Subsequent Win 11 reparations tonight didn't fix it, but FDISK on the actual Socket 7 machine did the job immediately.

Like I said, a skill issue 😁

Probably has to do with it being GUID or whatever the modern default is, rather than the MBR old computers expect. That would be my guess.

I wasn't aware that GPT would even work with older IDE based drives ..very weird if it was formatted that way. Just had a quick look and yeah unless the CF was formatted on Vista or better it wouldn't have been GPT as XP and earlier never supported GPT on removable media or drives.

Only XP 64 and later supported GPT for fixed bootable disks and Vista or better for removable media.

myne wrote on 2026-10-24, 00:15:

PcBytes wrote on 2026-09-30, 17:49:

Fixed an old derelict ASUS K7M using the "corpse" of the one I bought untested recently. Turns out Q16 was the culprit on the derelict one causing dead fans 🤣

Did have to dig up for the boardview to figure Q16 out though.

this help?

Not quite but thanks anyways. I don't have KICAD, just plain old BoardViewer.

RetroLizard wrote on 2026-10-24, 02:36:

PTherapist wrote on 2026-10-23, 18:29:

Today I fitted a PicoLoader to my GameCube.

It's a really nice idea, but damn they are so frustrating to install! It took me 6 attempts and 4 ruined ribbon cables but I finally managed to get it fitted correctly and it's working well.

I already had an SD2SP2 and was previously using the savegame exploit to access Swiss. The PicoLoader is a definite improvement over the savegame method, so it was ultimately worth all the effort getting it installed.

The ribbon cable is way too easy to destroy trying to install. And I've only tried it once!

In the end I actually cheated slightly to get it installed.

I unscrewed the disc drive from the metal shield and then fitted the shield first, double checking that the ribbon cable was in the correct position. Then I gently lowered the disc drive into place - so much easier to do when it's not attached to the metal shield, it allows you to be more precise without worrying about the shield ripping the cable. It does mean that you can't screw the drive back to the metal shield, but a small piece of tape on the top end is enough to secure it, as the case keeps it firmly in place anyway.

Trashbytes wrote on 2026-10-24, 07:29:

RetroLizard wrote on 2026-10-24, 03:41:

classic_cola wrote on 2026-10-24, 03:38:

Appreciate that, I'll certainly keep in mind next time a CF starts ghosting me. […]
Show full quote

Appreciate that, I'll certainly keep in mind next time a CF starts ghosting me.

Fortunately, after much hair pulling it's back to life using FDISK option 5 to set the CF as the "fixed disk" - then deleting the partition completely, and repartitioning, has brought it back to life.

Ultimately this is probably a skill issue on my part; I prepared this partition in Windows 11 a week or two ago which probably did something weird to the partition layout, I thought FAT16 is FAT16 whether prepared in modern Windows or not. Turns out I'm probably wrong.

Weird thing is it actually worked for a week like this. Subsequent Win 11 reparations tonight didn't fix it, but FDISK on the actual Socket 7 machine did the job immediately.

Like I said, a skill issue 😁

Probably has to do with it being GUID or whatever the modern default is, rather than the MBR old computers expect. That would be my guess.

I wasn't aware that GPT would even work with older IDE based drives ..very weird if it was formatted that way. Just had a quick look and yeah unless the CF was formatted on Vista or better it wouldn't have been GPT as XP and earlier never supported GPT on removable media or drives.

Only XP 64 and later supported GPT for fixed bootable disks and Vista or better for removable media.

I used MBR, which is still a secondary option in W11. Not sure if GPT is an option for these, although I'm using a CF dongle on the modern PC so the PC might think it's a USB flash drive if that makes any difference.

I think @myne has it here, Windows 95 might have been more forgiving about a byte related partition mismatch, whereas DOS might have rejected it.

classic_cola wrote on 2026-10-24, 14:47:

Trashbytes wrote on 2026-10-24, 07:29:

RetroLizard wrote on 2026-10-24, 03:41:

Probably has to do with it being GUID or whatever the modern default is, rather than the MBR old computers expect. That would be my guess.

I wasn't aware that GPT would even work with older IDE based drives ..very weird if it was formatted that way. Just had a quick look and yeah unless the CF was formatted on Vista or better it wouldn't have been GPT as XP and earlier never supported GPT on removable media or drives.

Only XP 64 and later supported GPT for fixed bootable disks and Vista or better for removable media.

I used MBR, which is still a secondary option in W11. Not sure if GPT is an option for these, although I'm using a CF dongle on the modern PC so the PC might think it's a flash drive if that makes any difference.

I think @myne has it here, Windows 95 might have been more forgiving about a mismatch between the partition type and the actual filesystem, whereas DOS might have rejected it.

True DOS do be like that.

It was just by chance that I didn’t burn out the Seasonic S12II in my retro PC—while testing the system outside the case, I eventually noticed heat radiating from the protective grille above the fan (the power supply was positioned vertically). The fan wasn’t spinning, and when I spun it by hand, the speed fluctuated and it made an unpleasant noise.
After unplugging and disassembling it, it turned out that there had originally been insulating spacers between the grille and the fan, which had deteriorated over time and warped to the point where the fan’s rotor was touching the grille (the grille is located under the casing along with the fan), and this was blocking its movement. After replacing them with new rubber washers, everything is working fine again, and although I don’t know how long this condition had persisted (the fan is inaudible even when running and the power supply had previously been installed in the case), the power supply didn’t burn out, and there were no visible signs of damage to the capacitors.
I’ve never encountered anything like this before.

Today I built and tested a new Asus TUSL2-C BIOS image. My base goal was to have a correct CPU detection for the VIA C3 chips.
It's based on the latest official 1014 build but with the following changes:

Removed the Intel network boot ROM (for CNR LAN cards) to gain some space
VIA C3 Nehemiah and Ezra-T are now recognized correctly (previously shown as "Intel Celeron").
VIA CPUs show their real name at boot, e.g. "VIA Nehemiah".
Newer and additional microcode for Pentium III and Celeron, including Pentium III-S Tualatin.
Add the i815 BIOS Tweaker from x86-secret.com, now dead
- The tweaker replaces the ChipAway boot-sector antivirus in the ROM
- Turn it on with "Boot Virus Detection" in Setup.
- It brings back AGP 4X and faster memory settings that the BIOS silently turns off when overclocking.

So basically it now correctly detects every Socket 370 CPU.

PS: yes the logos are broken, but it's from a faulty Radeon, not a BIOS bug.

Surfin' the BBS scene on my TI-99/4A like it's the 1980's!

It warms this old mans nostalgic heart. 😁

Ydee wrote on Yesterday, 08:42:

... (the grille is located under the casing along with the fan) ...

^ Placing the grille under the PSU case, so close to the fan, is bad practice for several aerodynamic reasons, not to mention the problem you had.

Seasonic rectified this on later S12IIs with a raised, stamped grille:

I'd move the grille to the outside of the PSU case, something like this:

Or, if once placed inside the PC the PSU fan is facing down, we can even completely remove the grille for optimal air flow.

TELVM wrote on Yesterday, 17:19:

Or, if once placed inside the PC the PSU fan is facing down, we can even completely remove the grille for optimal air flow.

I have an AT/ATX transition system where the ATX PSU fan faces up, not down.
Personally I'd keep the grille because the case might be moved/rotated and cables or other parts might move.

I had fun with routing AGP signals. And boy, am I happy that AGP died. The layout is just utter crap. the DBI_LO and DBI_HI signals are part of the strobe group where AD16-AD31 belong. But they are so far away on the connector from each other, that while all other signals of the group are around let's say 20mm trace length - the DBI_LO and DBI_HI are 31mm - an absolute joy to fiddle to match.

PCIe is just lovely compared to anything AGP.

Had to scrap an ASUS M2N32 SLi Deluxe. Dead NB and SB.

Spent a few hours recapping an ABIT BX133-RAID.
Never had one before, and this one was almost completely dead (sold as "untested".......).
CPU fan spun up but nothing else.
And I did something I don't normally do, I recapped it in stages.
First the worst caps, the 3 next to the ATX connector, so bad they had lifted from the board.
Still no post.
Next in line were the 8 larger caps next to the CPU, also obviously bad.
Post! but hangs during the process.
Next in line were the 3 smaller caps around the MOSFET's and the 2 next to the mem slots.
Again POST and it completes it, hangs when trying to enter the BIOS tho.
Swapping to a PCI card allows me to enter BIOS.
So next are the 5 caps next to the AGP port.
That solves the BIOS problem and the board now seems stable with an AGP card!
It's running memtest (no errors so far!) as I write this.

Also I measured all caps as I replaced them.
Not a single one was even remotely OK, most were barely showing as caps at all.

bofh.fromhell wrote on Yesterday, 22:38:

... I recapped it in stages ...

^ Quite interesting how fresh caps progressively resuscitated CPU, then Northbridge/RAM, then AGP. Nice work! 👍

TELVM wrote on Today, 00:06:

bofh.fromhell wrote on Yesterday, 22:38:

... I recapped it in stages ...

^ Quite interesting how fresh caps progressively resuscitated CPU, then Northbridge/RAM, then AGP. Nice work! 👍

I thought so to =)
Not that its a big surprise when swapping obviously broken components makes stuff work again.

Right now its running a 700E cu-mine @ 933MHz (133FSB) like its nothing.
Chugging SuperPI and Lightwave renders.
Well I'm pleased that I finally got me a BX133, I know its pretty much identical to the BE6-II 2.0 but BX boards with S370 are just cooler somehow =)

Now I'm eyeing a "Soyo KT333 Dragon Ultra" with an even worse caps situation.
I count 30+ visible bad ones.
And I have no clue where to find visually matching caps for it, gotta keep that silver look somehow.

TELVM wrote on Yesterday, 17:19:

^ Placing the grille under the PSU case, so close to the fan, is bad practice for several aerodynamic reasons, not to mention the problem you had.

It wouldn't be a problem if the material used for the pads were stable. Here, it looks like they used something similar to the “velvet coating” found on old IBM laptops and some mice, which over time turned into a sticky, gooey mess. But of course, placing the grille above the casing is a better idea.

bofh.fromhell wrote on Today, 00:29:

... Now I'm eyeing a "Soyo KT333 Dragon Ultra" with an even worse caps situation.
I count 30+ visible bad ones.
And I have no clue where to find visually matching caps for it, gotta keep that silver look somehow.

Replacement Silver Capacitors for Soyo SY-KT600 Dragon Ultra

^ Good advice on mobo capacitor selection, plus fellow Vogoner Baracoota sharing the awesome job he did: