vutt wrote on 2024-02-24, 10:42:

Treated my 486 setup with new Rubycon caps. Original ones looked beaten. However after running them trough my bench LCR comicall […]
Show full quote

Treated my 486 setup with new Rubycon caps. Original ones looked beaten. However after running them trough my bench LCR comically old ones measured closer to 220uF then brand new Rubycons. Still new ones were well in 20% spec.
ESR @100khz was slightly higher for one old one at 0.23Ohm compared to rest with 0.15Ohm measurement. However I don't think it's critical.
I was slightly worried that old ones were rated only at 85C while one of them is located right next to voltage regulator with spec temp oner 100C. With my Am486 DX2-80 I measured regulator temp 75C.
Then again board is 30 years old and still working so 85C was good enough...

I also added brand new heatsink to CPU. Seems to be good enough. During multi hour stress test it never reached 50C. Farnell/Element14 is selling them with thermal adhesive pad attached. Quite convenient.

Also joined finally theretroweb.com community and made my first contribution by sharing better quality MB pic.
newcaps.jpg
qdi_v4s471.jpg

Cool 😀 What do you use for ESR testing? I've just got one of those little transistor testers and it's mostly guesswork except for caps that have visibly failed or are known to be the failure point already in the circuit. Surprising that those original caps have held up so well honestly, but maybe the caps on that board running a linear regulator are not driven as hard as they would be for a switching power supply.

Joakim wrote on 2024-02-24, 13:13:

I was going to fix the missing caps on my 8 mb voodoo 2 I got years ago. Did not get it to work. Gets stuck on 3dfx logo. So I […]
Show full quote

I was going to fix the missing caps on my 8 mb voodoo 2 I got years ago. Did not get it to work. Gets stuck on 3dfx logo. So I put an other voodoo 12 mb in there instead.

It did not work either... Had a look at it and yeah it also has a missing cap otherwise it looks fine. Sadly I don't have that cap at hand.

Ok then I try my other voodoo 2 12mb card. Also this did not work. It has a bent pin on one of the tmus. Its not shorted or so but it might be cause if the problem.

I am sure one of these two 12mb worked 2 years ago.

So now I'm a bit bummed out as I realize I don't have a single working voodoo card when I thought I had at least 1.

I need to cool my head a bit and try with an other computer that at least has a good way of transferring files and that is in a tower case.

What demos are good for testing voodoo 2s? What are the most stable drivers? All of the cards are Creative labs.

That's totally fixable and you can use some the mojo utility to find out which part is bad. Since all 3 are now bad you should probably look at how these are being stored - the QFP (quad flat pack) chips on the Voodoo 1/2/Rush cards have lots of pins with pretty weak solder and that's getting worse over time it seems. It's possible if the card is stored squished / twisted with other cards the solder on the legs could eventually come loose.
See this thread for some details on diagnostics for the Voodoo 1 - it's the same for the Voodoo 2 but many more moving parts / things to go wrong: 3Dfx Voodoo 1 - Low level hardware information and diagnostics thread
There's also some youtubers like Bits und Bolts, vswitchzero and Nitton Åttiofyra that I think all post on this forum and maybe could help.
I've got at least 1x Voodoo 2 card back to working (bought it as for spares / repairs) by finding the general area of the fault with mojo / basic diagnostics then using that to resolder just the area of the affected chip which got it running again.

smtkr wrote on 2024-02-24, 00:39:

Thermalwrong wrote on 2024-02-22, 02:47:

LCD panel transplant - the original looked like the LCD layer was delaminating so has these weird squiggles at the edges of the […]
Show full quote

LCD panel transplant - the original looked like the LCD layer was delaminating so has these weird squiggles at the edges of the display:
IMG_3010 (Custom).JPG

I went into my parts bin of LCDs and found a similar lq121s1lh02 - 800x600 LCD that runs at 3.3v, the laptop gives 5v to the LCD. I tested the backlight and the backlight has bright spots and dark spots all over it, rubbish. The fresnel sheets look like they got scratched or are starting to go moldy - they are being stored essentially outside right now 🙁

Tried something I've never done before which was putting a different LCD panel in place of the original damaged LCD panel. The mini grinder tool made it possible and the backlight sheets made it through the transplant operation mostly undamaged.
I had to cut lots of plastic from the frame to make it fit but it's looking good and lights up the whole pixel area of the LCD just a little adjustment needed still. There's an AMS1117 converting 5v into 3.3v so the LCD panel electronics don't burn up, and the new panel's pinout is different from the original so some minor rewiring of the LVDS pairs was needed, but it works.
The repaired LCD looks great except for a bright line at the top but that doesn't notice too much in use, compared to the squiggles or bright spots. This means I've got one very nice LCD screen from 2 useless ones 😀

I did an LCD transplant last year. The new LCD I put in had a refresh rate significantly better than the original, as well as better contrast. I did it for fun. I haven't used the panel since initial testing to make sure it works. I'm kind of curious if the faster refresh rate will be noticeable, or if the control board is too old to drive it fast. Something to test in the future, I guess

If that's from a TN panel to an IPS panel like on a Thinkpad or something, the difference should be quite immediately noticeable - I've been interested in LCD stuff for a long time so swapping a Thinkpad T440 to a good LCD panel for cheap was really worth it.

I did the final adjustments to the LCD panel fitting today and it looks great 😀 Two bad LCDs (one bad LCD panel, the other a bad backlight uniformity) make one just about perfect LCD. These were both basic 800x600 LVDS TN panels that by modern standards aren't great but back then were "Active Matrix TFT". Up to a few years back they were cheap and plentiful but now they're pretty rare so I've got to use up whatever parts I can:

Thermalwrong wrote on 2024-02-24, 15:55:

That's totally fixable and you can use some the mojo utility to find out which part is bad. Since all 3 are now bad you should p […]
Show full quote

That's totally fixable and you can use some the mojo utility to find out which part is bad. Since all 3 are now bad you should probably look at how these are being stored - the QFP (quad flat pack) chips on the Voodoo 1/2/Rush cards have lots of pins with pretty weak solder and that's getting worse over time it seems. It's possible if the card is stored squished / twisted with other cards the solder on the legs could eventually come loose.
See this thread for some details on diagnostics for the Voodoo 1 - it's the same for the Voodoo 2 but many more moving parts / things to go wrong: 3Dfx Voodoo 1 - Low level hardware information and diagnostics thread
There's also some youtubers like Bits und Bolts, vswitchzero and Nitton Åttiofyra that I think all post on this forum and maybe could help.
I've got at least 1x Voodoo 2 card back to working (bought it as for spares / repairs) by finding the general area of the fault with mojo / basic diagnostics then using that to resolder just the area of the affected chip which got it running again.

smtkr wrote on 2024-02-24, 00:39:

Thermalwrong wrote on 2024-02-22, 02:47:

LCD panel transplant - the original looked like the LCD layer was delaminating so has these weird squiggles at the edges of the […]
Show full quote

LCD panel transplant - the original looked like the LCD layer was delaminating so has these weird squiggles at the edges of the display:
IMG_3010 (Custom).JPG

I went into my parts bin of LCDs and found a similar lq121s1lh02 - 800x600 LCD that runs at 3.3v, the laptop gives 5v to the LCD. I tested the backlight and the backlight has bright spots and dark spots all over it, rubbish. The fresnel sheets look like they got scratched or are starting to go moldy - they are being stored essentially outside right now 🙁

Tried something I've never done before which was putting a different LCD panel in place of the original damaged LCD panel. The mini grinder tool made it possible and the backlight sheets made it through the transplant operation mostly undamaged.
I had to cut lots of plastic from the frame to make it fit but it's looking good and lights up the whole pixel area of the LCD just a little adjustment needed still. There's an AMS1117 converting 5v into 3.3v so the LCD panel electronics don't burn up, and the new panel's pinout is different from the original so some minor rewiring of the LVDS pairs was needed, but it works.
The repaired LCD looks great except for a bright line at the top but that doesn't notice too much in use, compared to the squiggles or bright spots. This means I've got one very nice LCD screen from 2 useless ones 😀

I did an LCD transplant last year. The new LCD I put in had a refresh rate significantly better than the original, as well as better contrast. I did it for fun. I haven't used the panel since initial testing to make sure it works. I'm kind of curious if the faster refresh rate will be noticeable, or if the control board is too old to drive it fast. Something to test in the future, I guess

If that's from a TN panel to an IPS panel like on a Thinkpad or something, the difference should be quite immediately noticeable - I've been interested in LCD stuff for a long time so swapping a Thinkpad T440 to a good LCD panel for cheap was really worth it.

I did the final adjustments to the LCD panel fitting today and it looks great 😀 Two bad LCDs (one bad LCD panel, the other a bad backlight uniformity) make one just about perfect LCD. These were both basic 800x600 LVDS TN panels that by modern standards aren't great but back then were "Active Matrix TFT". Up to a few years back they were cheap and plentiful but now they're pretty rare so I've got to use up whatever parts I can:

Yeah the thing with the 8 mb card is that mojo does not complain about anything. Before the resoldering of the caps it was showing only 4 mb but still some problem. Might be driver related or that the motherboard is bad. I never used its only pci slot before. Need a better test setup.

Thermalwrong wrote on 2024-02-24, 15:55:

vutt wrote on 2024-02-24, 10:42:

486 setup

I've just got one of those little transistor testers and it's mostly guesswork except for caps that have visibly failed or are known to be the failure point already in the circuit. Surprising that those original caps have held up so well honestly, but maybe the caps on that board running a linear regulator are not driven as hard as they would be for a switching power supply.

Those funny looking T3 transistor/LCR testers are dead reliable when it comes to testing capacitors. Yes, capacitors not in a SMPS circuit have luxurious easy life and dont degrade provided they havent been cooked and are not from very bad capacitor years when they tried tuning dielectric for low esr (first SMD caps on macs/amiga) - this is why there is little point replacing electrolytic caps in 50 year old vintage gear.

Spent a large amount of time stripping my Gateway keyboard and removing all of the human slime it came with. Now it's beautiful again! But it also means my time with the system is up, it needs to go to a new home. It's on eBay if anyone in the UK is interested 😀

Not finished yet, but:
https://www.vbforums.com/showthread.php?90243 … 885#post5633885

I replaced the fans on two of my retro systems with quieter, more expensive fans from Digikey. I had to do all the crimping and cable covering myself.

Also replaced the VRM TO-220 heatsink with a beefier one since it didn't come with the original heatsink anyway.

Today I dedicated some time to a motherboard, an ECS P4IBAS with Willy 1700 478/256/400/1.75V, initially I had some problems, but fortunately I solved it, from the initial 128 MB of RAM I went to 512, tomorrow I have to connect one VGA AGP 4X and more RAM, in total I can reach 1.5 GB, and finally I just have to connect an HD and load Windows 7.

Spent the last 24 hours trying to get Windows 98 installed on a Pentium 4 system with an SSD.

I was continuously reminded of why I prefer DOS builds.

Shponglefan wrote on 2024-02-25, 02:44:

Spent the last 24 hours trying to get Windows 98 installed on a Pentium 4 system with an SSD.

I was continuously reminded of why I prefer DOS builds.

What was the problem? Something with the SATA controller? Too much memory?

9/10 times I saw that bluescreen during an install. I had a bad cable somewhere.

I spent hours yesterday trying to figure out why my Socket 7 system was unstable. It had weird issues like being able to run Windows 98 setup but throwing an error on first boot. I was able to make it go away by disabling the processor cache, but I hadn't had a problem with this CPU before.

I was swapping CPUs around earlier in the day and had left it in the wrong configuration with the voltage was set too low. After I turned the voltage back up with a jumper, everything worked fine.

Kahenraz wrote on 2024-02-25, 03:55:

Shponglefan wrote on 2024-02-25, 02:44:

Spent the last 24 hours trying to get Windows 98 installed on a Pentium 4 system with an SSD.

I was continuously reminded of why I prefer DOS builds.

What was the problem? Something with the SATA controller? Too much memory?

Seems to be an incompatibility with SSD drives on this particular motherboard / controller. It resulted in all sorts of scandisk errors (incorrect file locations, etc.), random setup issues, and other issues. I've been documenting the saga here: Re: Quad boot with a Pentium 4 2.8 GHz (DOS / Win 95 / Win 98SE / Win XP)

That said, I finally may have found a solution in switching to PIO mode 3 (default was 4).

luckybob wrote on 2024-02-25, 04:07:

9/10 times I saw that bluescreen during an install. I had a bad cable somewhere.

I wish this was just a bad cable. But no, seems to be an issue with the default drive settings in the BIOS versus the SSD drives I was trying to use.

Today I went ahead and hooked up my Windows 98 gaming PC again. I recently got an original copy of Baldur's Gate 1 and wanted to install it, only to not be able to install due to the CD-ROM drive not being able to read certain files. I looked at the first disc and of course, it has lots of minor use scratches on it. All the discs do. I gotta take them to the local used game store tomorrow. Those guys have one of those CD polisher things that can make it look brand new again.

The monitor I am using today is a Gateway Diamondtron VX920. This is by far the biggest CRT monitor I've ever used at 19". It supports 100Hz refresh rate and up to 1600x1200 resolution, although I think the most pleasing resolution for me has been 1024x768 @ 75Hz. I got this monitor from work last year and it's been sitting around. It wasn't until today that I got a kitchen sponge - one of those that has the rough side - and scrubbed off the remaining anti-glare coating that made the image on the screen look like shit. Now that the front glass is clear, it looks amazing. Honestly, I wish there was a way to play my old game consoles on this thing, but it doesn't do Interlaced signals.

Been playing Quake II and Turok in the mean time until I can get Baldur's Gate working.

In case anyone's not familiar with this PC specs,
Processor: Intel Pentium III (600 MHz, Slot 1 'Katmai') [originally a 350MHz Pentium 2]
Motherboard: Asus P2B (one of the legendary classics)
Memory: 256 MB PC-100 SDRAM [originally 128MB]
HDD: 40GB IDE (yes, it still works just fine - pulled from an office Windows XP system that was being thrown out years ago)
AGP: Diamond Viper V770 32MB [originally a Diamond Stealth II G460 with the Intel i740 graphics chip, 8MB]
Glide: 2x 12MB Voodoo2 cards in SLI [not originally in system]
Sound: Creative Labs AWE64 Standard ISA card (for full DOS compatibility) [originally an AWE64 GOLD card]
Network: Realtek Gigabit Ethernet PCI
OS: Windows 98 Second Edition [originally First Edition]

I'll need to find an old CRT TV - ideally one with S-video and Component inputs for N64, Gamecube, and PS1 (S-video) and Wii and PS2 (Component). It would also be good for playing Rocksmith on my PS3 (using PS2 Component cable) as I won't have any display lag to deal with.

Shponglefan wrote on 2024-02-25, 04:14:

Kahenraz wrote on 2024-02-25, 03:55:

Shponglefan wrote on 2024-02-25, 02:44:

Spent the last 24 hours trying to get Windows 98 installed on a Pentium 4 system with an SSD.

I was continuously reminded of why I prefer DOS builds.

What was the problem? Something with the SATA controller? Too much memory?

Seems to be an incompatibility with SSD drives on this particular motherboard / controller. It resulted in all sorts of scandisk errors (incorrect file locations, etc.), random setup issues, and other issues. I've been documenting the saga here: Re: Quad boot with a Pentium 4 2.8 GHz (DOS / Win 95 / Win 98SE / Win XP)

That said, I finally may have found a solution in switching to PIO mode 3 (default was 4).

I had a similar issue due to LBA limitations with Windows 98. I saw that you had addressed this later, but wanted to share my thread as well.

Trying to figure out the cause of data corruption in Windows 9x

FFXIhealer wrote on 2024-02-25, 05:05:

Today I went ahead and hooked up my Windows 98 gaming PC again. I recently got an original copy of Baldur's Gate 1 and wanted to install it, only to not be able to install due to the CD-ROM drive not being able to read certain files. I looked at the first disc and of course, it has lots of minor use scratches on it. All the discs do. I gotta take them to the local used game store tomorrow. Those guys have one of those CD polisher things that can make it look brand new again.

Disc error correction os actually very, very good. I've never had a disc error due to scratches on the plastic side. However, the label side hides the reflective layer underneath, and I have lost many discs due to extremely minor damage there.

It's not the scratches on the plastic but on the label, which often destroys a disk. And these cannot be repaired.

Made a clean install of Windows 95 on the Olivetti M4-100 and noticed something that was right before my eyes all time. It has no L2 cache. So that's why it felt so uncharacteristically sluggish. Even more surprised it stayed in use as an office PC until the mid-2000s.

Edit: And another reinstall after it started blue screening when loading up the desktop. Sometimes I forget how finicky W95 can get.

Kahenraz wrote on 2024-02-25, 06:09:

FFXIhealer wrote on 2024-02-25, 05:05:

Today I went ahead and hooked up my Windows 98 gaming PC again. I recently got an original copy of Baldur's Gate 1 and wanted to install it, only to not be able to install due to the CD-ROM drive not being able to read certain files. I looked at the first disc and of course, it has lots of minor use scratches on it. All the discs do. I gotta take them to the local used game store tomorrow. Those guys have one of those CD polisher things that can make it look brand new again.

Disc error correction os actually very, very good. I've never had a disc error due to scratches on the plastic side. However, the label side hides the reflective layer underneath, and I have lost many discs due to extremely minor damage there.

It's not the scratches on the plastic but on the label, which often destroys a disk. And these cannot be repaired.

However, if a bit of grit got under the disc and then it was rotated, that's a very easy way to repairably kill the error correction by producing a scratch which follows the data track rather than cutting across it.

rasz_pl wrote on 2024-02-24, 17:56:

Thermalwrong wrote on 2024-02-24, 15:55:

vutt wrote on 2024-02-24, 10:42:

486 setup

I've just got one of those little transistor testers and it's mostly guesswork except for caps that have visibly failed or are known to be the failure point already in the circuit. Surprising that those original caps have held up so well honestly, but maybe the caps on that board running a linear regulator are not driven as hard as they would be for a switching power supply.

Those funny looking T3 transistor/LCR testers are dead reliable when it comes to testing capacitors. Yes, capacitors not in a SMPS circuit have luxurious easy life and dont degrade provided they havent been cooked and are not from very bad capacitor years when they tried tuning dielectric for low esr (first SMD caps on macs/amiga) - this is why there is little point replacing electrolytic caps in 50 year old vintage gear.

Yes those atmega based cheap devices are more than enough. I have one DIY built. However I went little bit overboard during Covid times with my home lab equipment. So I have EastTester ET4410 bench LCR meter. In this case ESR is more like indirect Cap condition indicator since it's more relevant in switching power delivery circuits. Socket 3 and socket 5 era MB used mainly linear power delivery.

Yesterday finally got decent AT case and already built ss7 system(k62+ 475, ms-5184, asus v3400, sb16). "333" - that's how it was. Probably there was a Celeron 333. However, such cases are from the p-mmx era(1997), as far as I know. It looked much worse when I bought it(last photo)