Since had worked with electronics, and little research on line, photos was invaluable of a original Deskpro 386/25e, I needed to map out the pin out of power supply on this motherboard received from ebay seller. Mystery of the 2 pins can be explained, one of them is for HD LED activity signal input to PSU that leaves one other pin for power good signal from PSU.

Completed mapping out the pinout for PSU input on the motherboard. Motherboard does not need -5V since -12V feed from PSU on motherboard is split off to feed a LM7905 regulator IC to generate -5V for the ISA bus and for serial port.

Hard drive LED activity is on one of the pins to light up the LED inside the PSU for hard drive activity. Now there is other pin is for power good signal from PSU that leaves me with 2 unknown pins, matter of a picking which is which for power good if correct motherboard will power up.

The PSU pin spacing is 3.96mm apart. The connector plug is female, needs 15 or 16 pins connector, one piece. Where can I get these? It is square pin not wide flat like the horrible IBM P5/P6.

Cheers,

Does a 11-hour non stop working on code to get a Learning Management System (Moodle) custom theme to work on iOS 12.4 count as a retro activity? I never felt myself so angry. I tested even for iOS 10 and 11.

Turns out a small patch was done but not without much research. I’m losing my magic.

Bruninho wrote on 2020-04-28, 04:16:

Does a 11-hour non stop working on code to get a Learning Management System (Moodle) custom theme to work on iOS 12.4 count as a retro activity? I never felt myself so angry. I tested even for iOS 10 and 11.

Turns out a small patch was done but not without much research. I’m losing my magic.

*oof*...
At least it’s not Blackboard Learn (competing LMS) That thing still gives me nightmares.

The replacement GeForce 6200 PCI board came in the mail today along with my OLED display boards (20 days via Speedpak...is not speedy whatsoever).

Discovered that the standard/default soldering tip on the Weller soldering station is really too wide to work on small electronic boards. I had to solder 4 header pins on the OLED board for the Thinkpad Gotek and it was a royal pain trying to put a blob of solder precisely on the pins that I needed. Well, at least it works.

Once the Gotek display was sorted out, managed to mess around with a few old OSes being tested on a pile of SD cards via the “open” Thinkpad 560E - none of the old UNIX installs work (AIX/PS2 is looking for a token ring card, AT&T SVr4 doesn’t boot) but got NT 3.51 installed. The Trident Providia worked but the ESS Audiodrive didn’t. Made the mistake of pulling the SD card (formatted to an ancient NTFS version) and putting it on my modern Win10 machine so I can copy some files over...eh, that for some reason corrupted the file system. Might need to reinstall on FAT32 the next time around.

The replacement GeForce 6200 PCI worked just fine on the t5720, but then I realize how mediocre the performance really was (tested it on UT2004, which is a DirectX 8 game) It was cheap, sure, but it wasn’t
going to go above 40 FPS. I think I have UT99 and Battlefield 2 here somewhere that I can fire up. Oh well, it was still better than the onboard SiS315.

Had a bit of a scare earlier - thought that the card died again when leaving it in the thin client will cause it not to boot. Turns out I tried to start the thin client with it plugged into the Thinkpad charger (whoops!). Swapped cables and it worked just fine. Oh well, let’s see how it performs with Win98 and older games. I’ll need to install the capacitors for the PCI riser if I want the AOpen YMF744 Cobra to work (not really a priority since my Thinkpad T21 does an okay job with OPL3 emulation.

Ran speedsys on one of my Compaqs.

I caused a minor fire

Nothing is as reassuring as a lightly smoking PC case...

I hate SATA power connectors.. 🤣

wiretap wrote on 2020-04-28, 09:26:

I hate SATA power connectors.. 🤣

More than molex? That's unpossible.

Much prefer Sata over molex if I can (i.e. an XP build)! Molex is a wriggly pain! And Sata snaps into place reassuringly...

(although I much prefer none at all - my main PC uses M.2 slots which are beautiful!)

Almoststew1990 wrote on 2020-04-28, 09:55:

And Sata snaps into place reassuringly...

idk, those pictures don't look too reassuring to me x3

I think I forced the sata cable in upsidedown because the vertical bit of the connector had snapped off the DVD drive. It was a bit stiff but still clicked into place like normal so I didn't think anything of it until it caught fire!

LewisRaz wrote on 2020-04-27, 20:23:

Spent the last few evenings gaming on my 486sx50. Really enjoying this machine and seeing what it can do.

I have noticed this banding in certain resolutions with this vga card. The issue remains on the 3 systems I have tested it on.

Has anyone seen this before?

Curious since I’m also building a 486 system, with an ISA card. How are you finding the speed in Duke3d and other games? Also, which ISA card do you have?

LewisRaz wrote on 2020-04-27, 20:23:

Spent the last few evenings gaming on my 486sx50. Really enjoying this machine and seeing what it can do.

I have noticed this banding in certain resolutions with this vga card. The issue remains on the 3 systems I have tested it on.

Has anyone seen this before?

Is this a Trident card? They tend to do that on TFT panels..

kolderman wrote on 2020-04-28, 09:28:

wiretap wrote on 2020-04-28, 09:26:

I hate SATA power connectors.. 🤣

More than molex? That's unpossible.

For convenience factor, SATA is better, but as far as design and current capacity - SATA is worse. SATA tends to sag down and cause high resistance on the pins which leads to oxidation or burning. Molex secures tightly and creates a better electrical connection. As far as ratings, the Molex connection is usually rated at 132 watts (11 amps) whereas the SATA connection is usually rated at 54 watts (4.5 amps). Of course this also depends on meeting the spec, and how badly the Chinese manufacturers cut corners. This became apparent during cryptocurrency mining when people try using Molex to SATA adapters to power GPU riser boards and melted them or started fires depending on the graphics card PCIe slot power draw.

appiah4 wrote on 2020-04-28, 13:44:

LewisRaz wrote on 2020-04-27, 20:23:

Spent the last few evenings gaming on my 486sx50. Really enjoying this machine and seeing what it can do.

I have noticed this banding in certain resolutions with this vga card. The issue remains on the 3 systems I have tested it on.

Has anyone seen this before?

Is this a Trident card? They tend to do that on TFT panels..

Yes it is indeed a trident 8900C.

dave343 wrote on 2020-04-28, 13:08:

LewisRaz wrote on 2020-04-27, 20:23:

Spent the last few evenings gaming on my 486sx50. Really enjoying this machine and seeing what it can do.

I have noticed this banding in certain resolutions with this vga card. The issue remains on the 3 systems I have tested it on.

Has anyone seen this before?

Curious since I’m also building a 486 system, with an ISA card. How are you finding the speed in Duke3d and other games? Also, which ISA card do you have?

The card is an 8900C.

Duke3D is on the limit of what you might tolerate. Doom is very playable however. I also tested simcity 2000 which crashed after around an hour after becoming sluggish. And Transport tycoon deluxe which struggled with scrolling around the map but was playable. I would easily have tolerated the performance of TTD and doom back in the day, but not duke3d.

A member from another forum has offered me a dxs 66mhz so it will be interesting to compare that to the 50mhz SX. The system does also have 256kb cache.

LewisRaz wrote on 2020-04-28, 14:01:

appiah4 wrote on 2020-04-28, 13:44:

LewisRaz wrote on 2020-04-27, 20:23:

Spent the last few evenings gaming on my 486sx50. Really enjoying this machine and seeing what it can do.

I have noticed this banding in certain resolutions with this vga card. The issue remains on the 3 systems I have tested it on.

Has anyone seen this before?

Is this a Trident card? They tend to do that on TFT panels..

Yes it is indeed a trident 8900C.

Known issue with no workaround, use a CRT or another card.. Sorry. 8900 is a fast card and I have one with 1MB RAM too, but the same issue prohibits me from using it..

Almoststew1990 wrote on 2020-04-28, 09:17:

I caused a minor fire

Nothing is as reassuring as a lightly smoking PC case...

I smell bacon!

appiah4 wrote on 2020-04-28, 14:16:

LewisRaz wrote on 2020-04-28, 14:01:

appiah4 wrote on 2020-04-28, 13:44:

Is this a Trident card? They tend to do that on TFT panels..

Yes it is indeed a trident 8900C.

Known issue with no workaround, use a CRT or another card.. Sorry. 8900 is a fast card and I have one with 1MB RAM too, but the same issue prohibits me from using it..

How does the 8900 compare to the Cirrus Logic 5429? Also 1mb ISA.

dave343 wrote on 2020-04-28, 14:29:

appiah4 wrote on 2020-04-28, 14:16:

LewisRaz wrote on 2020-04-28, 14:01:

Yes it is indeed a trident 8900C.

Known issue with no workaround, use a CRT or another card.. Sorry. 8900 is a fast card and I have one with 1MB RAM too, but the same issue prohibits me from using it..

How does the 8900 compare to the Cirrus Logic 5429? Also 1mb ISA.

I would take an ISA Cirrus Logic over a Trident 8900C any day . The 8900D is supposedly quite fast, but the previous models, not so much .
I used a VLB 5429 back in the day and had no compatibility issues, I can't comment on the speed of the ISA version or LCD compatibility . I did have a Trident 8900B before that and definitely would not have called it fast (then again OAK Technologies and Realtek ISA offerings were even slower) .

Bruninho wrote on 2020-04-28, 04:16:

Does a 11-hour non stop working on code to get a Learning Management System (Moodle) custom theme to work on iOS 12.4 count as a retro activity? I never felt myself so angry. I tested even for iOS 10 and 11.

Turns out a small patch was done but not without much research. I’m losing my magic.

Sounds like the first time I discovered the need for cache busting for mobile platforms. I spent a bunch of time thinking they were handling css selectors differently, but it was just a stale version of the css file.