H.W.Necromancer wrote on 2022-06-26, 07:03:

Today activity is an attempted to repair Radeon HD2600 PRO but it failed. 😓 The card is running untill you install the drivers but than the OS will crash on boot. I found one missing elyt capacitor - replaced and replaced the other electrolytics but it did not help. Thinking if reflowing the chip makes any sence or not - wanna try my best before "scorching" the GPU with hotair.

Have you tried it in multiple motherboards? Sometimes that is a symptom of an incompatibility, alot of those late AGP cards had weird issues with certain AGP chipsets.

You can also try turning down the AGP slot to 2x or 4x and see if makes a difference.

It's also very important for certain chipsets to have their drivers installed, otherwise the AGP slot will be unstable.

Kahenraz wrote on 2022-06-26, 05:17:

Get a TV with VGA input or an adapter and split the video signal using a passive cable. You'll get a better picture without all of that luma chroma garbage.

I've seen people modding TVs to have VGA inputs specifically for CRT EmuDriver usage, I suppose that could be an option depending on the availability of the service manual and whether using something to lower the VGA output resolution to 15KHz is an option for him (there are DOS TSRs that can do this, but game compatibility is hit or miss as demonstrated in Part 2 of The 8-bit Guy's TV RGB Modding video).

There were also VGA to TV converter boxes that could output through RGB SCART (the AverMedia AverKey iMicro is one of them, although its "SCART" output is a DE-15 VGA-style connector. Could probably mod a CRT TV to accept an RGB input through BNC jacks like a PVM does and use a VGA-to-BNC breakout cable for that), could be useful if RGB modding the TV is an option (depending on the scarcity of such converters).

Kahenraz wrote on 2022-06-26, 16:10:

Check for missing surface mount components. If all else fails, replace the rest of the capacitors and see if that helps. Low-ESR are the best choice for video cards.

I had a GeForce 4 MX that exhibited graphical corruption, very characteristic of faulty memory. It ended up being the capacitors. After replacing them, I could even overclock the memory without any problems.

Repair shops hate me. I fixed defective memory on my video card with this one neat trick.

Good work! Few months ago, I saved a box of GPUs from a scrap yard and there were some GF440 as well. I repaired most of them and one or two GF4MX went back to life after capacitors replacement too. They are cheap but still I like to save those for the future.
--
By the way if you are more experienced in GPU diagnostics you may be able to help. If there are artifacts just right on the POST screen - for example vertical stripes or dots, or horizontal. Do you think this is GPU or the memory problem? For example, now I have Gf4Ti 4200 with thin vertical lines. Once I tried to hot-air reflow one Radeon 9600 with horizontal grainy stripes - did it twice, the card surprisingly survived my scorching but no change. Still faulty. But later, I started to think those might by faulty memory and reflowing the chip is useless. (??) Any idea will be appreciated...there might be better card later worth to save. So good to learn.

Have a nice day.
H.W.N.

Kahenraz wrote on 2022-06-26, 16:23:

You can also try turning down the AGP slot to 2x or 4x and see if makes a difference.

It's also very important for certain chipsets to have their drivers installed, otherwise the AGP slot will be unstable.

Thank you! This is not the problem here, but you are right. I had got this lesson very painfully years ago when those systems were modern and up to date and I had been struggling with my FIC VA503+ motherboard AGP...Oh, those times. I miss sometimes.

Kahenraz wrote on 2022-06-26, 16:23:

You can also try turning down the AGP slot to 2x or 4x and see if makes a difference.

It's also very important for certain chipsets to have their drivers installed, otherwise the AGP slot will be unstable.

Thank you. I have tried this - and finally I found my 3650 (very similar card) and it works in the same system. I have tried 3 very different motherboards and two OS versions to be sure...

H.W.Necromancer wrote on 2022-06-26, 07:03:

Today activity is an attempted to repair Radeon HD2600 PRO but it failed. 😓 The card is running untill you install the drivers but than the OS will crash on boot. I found one missing elyt capacitor - replaced and replaced the other electrolytics but it did not help. Thinking if reflowing the chip makes any sence or not - wanna try my best before "scorching" the GPU with hotair.

EDIT - progress:
1) Thank you to all of you who put some hints here. It may help to other people with AGP struggle
2) Unfortunately/ fortunately, with 99% sure I have diagnosed the problem - there is a cold joint almost for sure. The previous owner had installed a 2-screws aftermarket cooler with a diagonal steel spring (see the pictures above).
Those coolers are infamous known for putting an uneven tension on the PCB, which after a long time, supported by temperature and gravity tend to twist the PCB in a shape of a "propeller blade".
Here it is not so much - so I missed that fact. I have 2 cards in my spare parts box where it was fatal.
3) After my futile recap attempt, I just simply very firmly grabbed the card and bent it by my hands back to the straight shape by my hand and started the PC. I hold my hand there during the boot. Then released.
The OS booted and the card styed stable and nd passed even a stress test in 3D Mark 2005 and 2003 with no problems! However, it again stopped working after a restart. So there is a problem preventing the driver to load, however the chip itself seem to be OK, same as the other parts. I was not able to repeat the process. But I know what want to know and I don´t want to stress the pcb for now more.
4) I removed the unholy cooler, heated up the PCB a bit by a hairdryer and let it flat - it is much less twisted now.
5) I will try the card later again with a cooler from my working 3650 (in needs cleaning anyway and it has same design as 2600) - it has 4 screws and springs - much better for the PCB. In case the card is not working this way it needs a careful reflow - 99% it is a problem under the chip. I might ask someone with an infra station to help. I spent too much time with that card and want to save it.
--
Sharing this to help the others as well. Inspect the card and try simply stuff first before you plunge in to soldering and recap like I did .... 😀)

Have a nice summer Vogoneers

I see a lot of used motherboards that are clearly warped around the CPU socket due to the stress of a heavy cooler sitting on it for years at a time. I'm not sure what a good solution would be to address this. You would think that a metal brace covering a large surface area of the board would be able to prevent it from flexing, but this exact kind of thing still failed to prevent many of Microsoft's past Xbox designs from dying.

H.W.Necromancer wrote on 2022-06-27, 10:22:

2) Unfortunately/ fortunately, with 99% sure I have diagnosed the problem - there is a cold joint almost for sure. The previous owner had installed a 2-screws aftermarket cooler with a diagonal steel spring (see the pictures above).
Those coolers are infamous known for putting an uneven tension on the PCB, which after a long time, supported by temperature and gravity tend to twist the PCB in a shape of a "propeller blade".

This shouldn't happen unless the screws are overtightened. The flex from the aluminum brace is stronger than the material of the PCB, and it will flex over time and become warped. This is also a good reason to use a GPU brace, to take the weight off of the I/O bracket and the slot. There is a reason that the PCIe slots meant for GPUs on higher end boards are reinforced with a metal guard.

Kahenraz wrote on 2022-06-27, 11:09:

I see a lot of used motherboards that are clearly warped around the CPU socket due to the stress of a heavy cooler sitting on it for years at a time. I'm not sure what a good solution would be to address this. You would think that a metal brace covering a large surface area of the board would be able to prevent it from flexing, but this exact kind of thing still failed to prevent many of Microsoft's past Xbox designs from dying.

More likely to be the high clamping force from S423/S939 and on when the socket grabbing style was dropped.
Easily fixed with a sturdy back-plate, but hey that's like $0.45 more!

Noticed when assembling a few "cheap" modern LGA1700 systems that the MB bend is insane even with Intel's stock cooler.
Cheap and thin motherboards combined with strong clamping force meant that I actually had problems screwing the MB down in the case if I mounted the cooler first.
The bend was probably 15-20 mm or so if placed on a flat surface.

bofh.fromhell wrote on 2022-06-27, 11:24:

More likely to be the high clamping force from S423/S939 and on when the socket grabbing style was dropped. Easily fixed with a […]
Show full quote

Kahenraz wrote on 2022-06-27, 11:09:

I see a lot of used motherboards that are clearly warped around the CPU socket due to the stress of a heavy cooler sitting on it for years at a time. I'm not sure what a good solution would be to address this. You would think that a metal brace covering a large surface area of the board would be able to prevent it from flexing, but this exact kind of thing still failed to prevent many of Microsoft's past Xbox designs from dying.

More likely to be the high clamping force from S423/S939 and on when the socket grabbing style was dropped.
Easily fixed with a sturdy back-plate, but hey that's like $0.45 more!

Noticed when assembling a few "cheap" modern LGA1700 systems that the MB bend is insane even with Intel's stock cooler.
Cheap and thin motherboards combined with strong clamping force meant that I actually had problems screwing the MB down in the case if I mounted the cooler first.
The bend was probably 15-20 mm or so if placed on a flat surface.

Gamers Nexus just did a Video all about LGA 1700 and how messed up Intels latching system is on that socket, TL;DR Intel needs to either return to the dual latching they had on X299 or move to a system like Threadripper uses where pressure is even across the entire socket.

https://www.youtube.com/watch?v=Ysb25vsNBQI

LGA 1700 is really poorly designed retention wise and even worse when half the CPU dies are not being cooled due to uneven pressure on the IHS from the cooler due to socket warping.

Kahenraz wrote on 2022-06-27, 11:09:

I see a lot of used motherboards that are clearly warped around the CPU socket due to the stress of a heavy cooler sitting on it for years at a time. I'm not sure what a good solution would be to address this. You would think that a metal brace covering a large surface area of the board would be able to prevent it from flexing, but this exact kind of thing still failed to prevent many of Microsoft's past Xbox designs from dying.

H.W.Necromancer wrote on 2022-06-27, 10:22:

2) Unfortunately/ fortunately, with 99% sure I have diagnosed the problem - there is a cold joint almost for sure. The previous owner had installed a 2-screws aftermarket cooler with a diagonal steel spring (see the pictures above).
Those coolers are infamous known for putting an uneven tension on the PCB, which after a long time, supported by temperature and gravity tend to twist the PCB in a shape of a "propeller blade".

This shouldn't happen unless the screws are overtightened. The flex from the aluminum brace is stronger than the material of the PCB, and it will flex over time and become warped. This is also a good reason to use a GPU brace, to take the weight off of the I/O bracket and the slot. There is a reason that the PCIe slots meant for GPUs on higher end boards are reinforced with a metal guard.

The mentioned cooler has not a very good design. There are just rubber rings between the metal columns and the PCB. And no springs. The steel strong diagonal spring is holding the cooler on the top and it is almost impossible to have a good contact with the chip surface if the screws are not tightened fully. The process of the deformation took some time, and the card looked a bit twisted when I found it. Of course, I don ´t know the conditions of previous usage and storage...
You are right with the motherboards - especially 775 and later are sensitive - at least what I have seen. Fortunately, usually works even with such a deformation. But not always...

Does anybody know what the stock CD ROM on the Gateway 2000 DX4-66 LP1 was? I finally got the hard disk and floppy drives working (though it refuses to boot my Win95 diskette for some reason, DOS 6.22 works fine) but I cannot get the CD drive to detect. I'm wondering if this is even stock. When I got the system it had a SB16 Vibra CT2800 and a Sony IDE CD ROM connected straight to the motherboard with a RedHat Linux install CD in it so evidently the previous owner was trying to get Linux onto this 486 circa 2000. Alot of people are saying that the 66V (the full height desktop version of this that had VLB instead of PCI) came with 1st Gen SB16s w/ proprietary CD ROM drives. Right now the CD drive just blinks for about 10 seconds on start up and thats it. No detection, nothing.

I've already tried moving the CD ROM to the primary, putting the HDD and CD on the same channel, and swapping out a known good CD drive. This is definitely just a detection/support issue.,

Does the BIOS on this even support IDE CD ROMs on the onboard controller? I tried connecting the CD ROM to the Soundblasters IDE controller but then the BIOs started throwing a fixed disk conflict error up.

Its been a hot minute since I messed with anything older than a Pentium MMX, recommendations on my best path would be great.

TheAbandonwareGuy wrote on 2022-06-27, 19:16:

Does anybody know what the stock CD ROM on the Gateway 2000 DX4-66 LP1 was? I finally got the hard disk and floppy drives workin […]
Show full quote

Does anybody know what the stock CD ROM on the Gateway 2000 DX4-66 LP1 was? I finally got the hard disk and floppy drives working (though it refuses to boot my Win95 diskette for some reason, DOS 6.22 works fine) but I cannot get the CD drive to detect. I'm wondering if this is even stock. When I got the system it had a SB16 Vibra CT2800 and a Sony IDE CD ROM connected straight to the motherboard with a RedHat Linux install CD in it so evidently the previous owner was trying to get Linux onto this 486 circa 2000. Alot of people are saying that the 66V (the full height desktop version of this that had VLB instead of PCI) came with 1st Gen SB16s w/ proprietary CD ROM drives. Right now the CD drive just blinks for about 10 seconds on start up and thats it. No detection, nothing.

I've already tried moving the CD ROM to the primary, putting the HDD and CD on the same channel, and swapping out a known good CD drive. This is definitely just a detection/support issue.,

Does the BIOS on this even support IDE CD ROMs on the onboard controller? I tried connecting the CD ROM to the Soundblasters IDE controller but then the BIOs started throwing a fixed disk conflict error up.

Its been a hot minute since I messed with anything older than a Pentium MMX, recommendations on my best path would be great.

Are you loading any drivers when booting DOS? VIDE-CDD.SYS should work for most IDE CD-ROM drives.

In my experience optical drives are not detected by the BIOS on 486 and older boards, so only hard drives show in the drive list during POST. You won't know if it works until you try to load the DOS driver.

EDIT: regarding whether the drive is stock or not, you can check the date code in the drive and compare it with the dates of the other components, it won't be 100% certain but it could give you a rough idea.

Does the BIOS have the option to boot from CD? Not all 486 did

Is it possible to boot from a CD without support from the BIOS?

Kahenraz wrote on 2022-06-27, 20:26:

Is it possible to boot from a CD without support from the BIOS?

There are small apps that enable that (boot from floppy then pass control to the CD), but I've never tried it, so can't advise. It needn't be a big app, a small app like a TSR could do it.

If your CDROM is 8x or slower you also have to be questioning whether or not the drive itself is boot capable or not. It's kind of a sliding scale at 1:9 at the 1x end to 9:1 at the 8x end as the years went on for proportion of drives boot capable or not. Then there's been some latter day natural selection going on, i.e. the 8x (and 2x, 4x 6x) that won't boot got yeeted sooner than the 8x that were fully capable.

BitWrangler wrote on 2022-06-28, 01:24:

If your CDROM is 8x or slower you also have to be questioning whether or not the drive itself is boot capable or not. It's kind of a sliding scale at 1:9 at the 1x end to 9:1 at the 8x end as the years went on for proportion of drives boot capable or not. Then there's been some latter day natural selection going on, i.e. the 8x (and 2x, 4x 6x) that won't boot got yeeted sooner than the 8x that were fully capable.

I wasn't trying to boot from the CD directly. I was trying to get the CD ROM to detect so I could install Win95 from it. I'm using a Win95 OSR2 Boot Floppy to actually boot, its just the Win95 setup data is actually on CD. I don't feel like creating 31 floppies to install Win95 floppy disk edition.

Anywho this setup is def aftermarket and I did get it working. The motherboard IDE controller and BIOs has absolutely no CD-ROM support. I don't think it even shipped with a CD drive. I want to say the previous owner had no idea what he was doing but TBF this would have been a weird issue to resolve in 2000 unless you had something like Scott Muellers "Upgrading and Repairing PCs" or an internet connection to help with trouble shooting AND a lot of spare parts to rule out parts failure or incompatibility. I wouldn't have wasted time on a 486 then either.

To anyone reading this in the future with the same problem with their Gateway 4DX2-66 LP1:

To get the IDE controller on the Soundblaster working you need to disable the onboard secondary IDE channel, and then connect the CD Drive to the SB's IDE controller jumpered for Master (CS does not work). The OAK CD Rom driver works fine after that. I also disabled all the serial and COM ports to reduce the chance of address conflicts as (I at least) never use those on my retro PCs. The Windows 95 OSR2 Boot Disk also does not function correctly with this machine for whatever reason, use a Windows 98FE Boot Disk with the 95OSR2 CD-ROM.

I'm trying to set up a CICD system that will deliver GZDoom, LZDoom and OpenMW for Windows XP x64. I also need a VNC server for testing, I don't want to walk from one room to another all the time.

You may disagree that Windows XP x64 is a retro setup and you might be partially correct. The retro core of this setup is the SoundBlaster Audigy 2 ZS which is supported very nicely under that OS. I've chosen that system because it can do 64bit arithmetic natively and since I'm going to be compiling a lot of things, this will come in handy sometimes. Vista on the other hand killed DirectSound3D which I find to be detrimental to my goal of building a setup that uses EAX.

moog wrote on 2022-06-28, 07:52:

I'm trying to set up a CICD system that will deliver GZDoom, LZDoom and OpenMW for Windows XP x64. I also need a VNC server for testing, I don't want to walk from one room to another all the time.

You may disagree that Windows XP x64 is a retro setup and you might be partially correct. The retro core of this setup is the SoundBlaster Audigy 2 ZS which is supported very nicely under that OS. I've chosen that system because it can do 64bit arithmetic natively and since I'm going to be compiling a lot of things, this will come in handy sometimes. Vista on the other hand killed DirectSound3D which I find to be detrimental to my goal of building a setup that uses EAX.

XP/XP64 is retro in my book, along with Vista, they both run on similar hardware which is firmly in the retro arena.