Very nice! The new finish and little extra touches/details makes this PC build stand out from the crowd - and looks great as well. Felicitaciones! 😎

Too bad the Viper V330 was busted; just keep on looking for another PCI video card, there has to be one in Argentina that works, is affordable and better than the Trident 9680.

Windows 95 OSR 2.x USB support is very basic and rather poor - PC hardware manufacturers had to provide the drivers, (almost) nothing came from MS. With luck it will work out, otherwise it's back to Windows 98.

Gracias!!! 😎
I have a problem with the switch wiring. I have to switch ground cables too, but the switch is double. I need a quad switch. I thought it would not be neccesary but if I switch off the caddy I get a bios hard drive error. I will work on this soon. Well, there are many final details I would like to work on and not enough time. But sooner or later I will do all of it. It's just part of the fun.

But not everything is bad news. I bought a few days ago fifteen ram modules at less than two dollars. All 72 pin edo sticks between 4 and 8 mb. Well, those were seller's words. In fact, I found two 16mb sticks. One of them was broken. The system didn't post. I've managed to repair the stick replacing a smd broken resistor and now I have the mega ultra hiper furious amount of 64 mb!!! 😀 😀 😀 Yes, I know I said that I wanted to keep things as original as possible, but how can I resist maxing ram for my 256kb cache? Now I don't mind installing w98se. Perhaps with a few tweaks the system will be fast enough to be enjoyable. And if I could implement some turbo function, it will be flexible enough for most games / software.

I'm searching for a video card almost every day. It's really weird how pci video cards dissapeared from market. I remember 8 - 10 years ago market beeing crowded of this cards. Everything from voodoo to Geforce 4. Now the only common cards to get are similar to my Trident.

Well done! And just be patient, chances are good that something nice will pop up 😀

Just a hint: You can often get seemingly dead/defective memory modules back to life with a clean of the contacts. Some people use rubbing alcohol (don't use anything that's more agressive) and the pencil eraser trick seems to work just as well for other members here.

Thanks Tetrium! Yeah, I know that tricks but nothing. The resistor was broken but wasn't visible until I touched with the soldered and it drop off in two pieces.
I'll run memtest+ today just to see if everything is ok.

Really like all the little custom mods on this build. That's what sets it apart from a run-of-the-mill Socket 7 machine!
If you can't find another PCI TNT, consider a GF2 MX440. I believe it's just as fast and will work with older drivers too.

Thank you bjt! 😀
I'm glad you guys like this machine. I worked with so much love and dedication for it. I really like this machine. It brings me back so many chilhood memories. And I'm not exagerating if I say that I'm studying engineering because of this machine jajaja. When I was seven or eight years old I learned to disassemble it completely, installing cards, upgrading cpu and ram, and so on.

The damaged VGA was a Riva 128, not a TNT. By the way, I think a MX440 would be ultra overkill for this rig. I had one of those in my Compaq Deskpro Piii 933mhz. I think a TNT2 (or similar) would be the most high card to look for this rig.

Perhaps I will start working on turbo function today. Need to see if my motherboard has some multiplier / bus pins in common with another ones, so I can use directly the case switch. I don't mind if I can't switch on the fly, but I strongly want to use the original switch. I will not change it or use another thing. What would be slower? 50 mhz x 3 or 66mhz x 2? I think the 50mhz bus will be slower. I still have to see if the mmx works on 50 mhz x 2. The slower I got was those two combinations.

Well guys, I managed to implement turbo function!!! 😎
I was looking at fsb jumpers on motherboard. There are two rows with three pins on each. Silly me, I can connect one switch in one row and switch between 50 and 66 mhz. Let's take a look at it:

First I had to solder the turbo led to the led display because I wanted to have it working:

Then I connected to the motherboard:

Let's try it at low setting:

Good! 150 mhz! I assume it's 50mhz x 3. Now try it at high setting:

Allright! Full 200 mhz! At 150 mhz I got 112.5 points in Speedsys. At full speed, 150 points. It isn't really slow, and I don't think it would serves too much. But it's cool to have this working. The only issue is that I can't switch it when the system is on. It leads to a full system crash.

I still need to see the caddy switch.

See you! 😎

PD: I can lower the system performance a little more. When I set the multiplier at 2x, the system boots at 133. Speedsys shows it at 125 mhz, so I assume it's 50 mhz x 2.5. The score is 93.77, but I don't have any unused switch to implement it (I don't want to use more switches than the original one). I will see if I can do some wiring magic.

Guys I'm having some issues with the psu and/or motherboard. I've noticed a few days ago some burning smell when I was playing Commander Keen 1. I took a look at it and saw the power_good signal motherboard connector burning. The cable isn't hot at all, it's just the connector. I unplugged it and noticed that the pin inside the psu AT connector was a little lose, so my first thought was arcing. I managed to give it proper shape but no luck. It is getting just too hot. I tried changing the cpu for a Pentium 100 for lower consumption and test only motherboard, cpu and ram. Still too hot. I don't see anything wrong on the motherboard, although I didn't measure anything yet.

What could it be? What do you think guys?

Burning components is really bad.

I'd say don't keep on messing around till you figured out what's wrong but my guess would be some kind of short, hold on.

edit:
I'd consider (temporarily) replacing both PSU and motherboard for now. Inspect both motherboard (you already mentioned you did that though, back side too?) and PSU. It sounds like there was too much resistance in one of the wires or something. Personally I lay aside both PSU and motherboard in such cases and perhaps troubleshoot both motherboard and PSU separately (with components you can afford to loose!). AT s7 boards should still be very cheap to come by, there were just so many made.

I'm not a guru when it comes to circuitry, but I do know that burned spots are typically very bad.

Yes Tetrium. Of course It's very bad. I think I can discard arcing, I triple checked it and no arcing issues. I checked both sides of the motherboard and nothing weird or shorted or anything. In fact, the system works fine even with prime95 long testings. It's just that pin burning. Even the cables are cold.
I stopped using this rig two days ago when I found that. I was testing the Voodoos when I felt hot air coming from that zone. And well, I saw that.

I will dissasemble all the system and test it on cardboard with just cpu (Pentium 100 for lower current draw) and one ram stick looking for lowest power consumption. I will see if I can buy another AT motherboard to run some tests.

I'm not happy about this 😢

ramiro77 wrote:

Yes Tetrium. Of course It's very bad. I think I can discard arcing, I triple checked it and no arcing issues. I checked both sid […]
Show full quote

Yes Tetrium. Of course It's very bad. I think I can discard arcing, I triple checked it and no arcing issues. I checked both sides of the motherboard and nothing weird or shorted or anything. In fact, the system works fine even with prime95 long testings. It's just that pin burning. Even the cables are cold.
I stopped using this rig two days ago when I found that. I was testing the Voodoos when I felt hot air coming from that zone. And well, I saw that.

I will dissasemble all the system and test it on cardboard with just cpu (Pentium 100 for lower current draw) and one ram stick looking for lowest power consumption. I will see if I can buy another AT motherboard to run some tests.

I'm not happy about this 😢

I know, and I understand. Especially since you haven't been "in the field" for very long, must seem like a disappointment.

But this also comes with the territory. Actually, building computers in the long run turned out to be very therapeutic for me as I learned to just have patience with things and to take it easy.

I've seen so much disappointments while busy with my old rigs, it's so un-funny that I can laugh about it now (i know that won't make you feel any better, just trying to get my point across). Once I was trying to build my second 486 rig, using an ASUS SP3...something (both PCI and 1 VLB slot) and the very first time I applied power to the mobo (it was on my standard cardboard box on my testbench) some cap exploded like a firecracker went off right in front of me and the thing never displayed anything on the monitor. Heck, it didn't even give any beep codes afaicr. So much for a now not so easy to find ASUS Socket 3 board...literally went up in smoke and a loud BANG!

But it happens and don't forget that you gave it your best shot, but that doesn't mean that you failed, sometimes it's the stuff you're working on that fails you despite your best efforts, please, don't forget that 😉

In fact I been with computers for the last 10 years. But not with systems this old, on which aging starts to show badly enough.
But I'm not giving up. This rig is so beautiful to me that I will keep trying to finish it. Worst case scenario would be buying a new psu and a full mobo recap (which is quite easy on this motherboards because most electrolytics are through hole).

Now I'm working on an array of 5v fan blowers to load the psu and take some measurements. I don't have enough power resistors for this purpose. I hope the psu is bad and not the motherboard.

ramiro77 wrote:

In fact I been with computers for the last 10 years. But not qith systems this old, on which aging starts to show badly enough.
But I'm not giving up. This rig is so beautiful to me that I will keep trying to finish it. Worst case scenario would be buying a new psu and a full mobo recap (which is quite easy on this motherboards because most electrolytics are through hole).

Now I'm working on an array of 5v fan blowers to load the psu and take some measurements. I don't have enough power resistors for this purpose. I hope the psu is bad and not the motherboard.

Sorry, I wasn't trying to imply you had no experience or are severely lacking in it (I can see how my previous reply might have implied this), but these old systems sometimes seem to have a mind of their own. The rules are very different and you're more or less on your own here.

Working with gear this old is a challenge in itself, partly because in the end it's all up to you. Someone who has a new computer can simply bring it to some shop, throw in some stacks of paper money and have it repaired, any computer repair shop will know what to do. But suppose you'd bring your Pentium 1 to the shop. Chances are they will either simply start laughing at you and advice you to just purchase a brand new one "as the old one simply isn't wort repairing" or they will attempt to accept the challenge, but will either charge you a kidney or will simply not be capable dealing with this old gear and might ruin it.

The fact that you seem to be quite capable of handling a soldering iron is actually a big plus for you 😀 The aging of these older parts and the lack of knowledge about these older parts is part of the challenge. Websites with tons of info will in time disappear, lots of dead links, manufacturers who don't have the manuals, drivers and BIOS files on their website anymore. It's still computers but with these old parts, but it's a different game 😜

The fact that the plastic is burning means it must be an overcurrent condition. You may actually have a short circuit on the motherboard - I have seen these a number of times myself. AT PSUs (especially early ones) didn't have short circuit protection, so if there was a short the voltage would sag a little and they'd just deliver their full rated load, until something went bang or started burning. Since the ATX standard made PSUs shut themselves off if a short circuit was encountered, this has become less of a problem.

Assuming the connector is OK (both the motherboard side, the plug, and most importantly, that the wire going into the plug has been crimped correctly, you will probably have to start looking elsewhere for a short.

Easiest way to find this would be to leave the machine on for a while, switch it off just before it starts burning, then touch every component on the motherboard until you find one that's unusually hot (but be careful, I've burned my fingers doing this from components that are *really* hot!) That will almost certainly be the faulty component. Capacitors are the most likely to fail, so you could start with those.

@Tetrium: don't be sorry, I did understand you perfectly! No problem dude! 😀 But sure, this is not as easy as buying some new system and put it to work. We deal with the worst evil: aging. 😵 And I will not give up on this. I really love my rig and I'm very nostalgic and sentimental about it. It was my first computer ever. I'm probably beeing too irrational, but this is how I feel about this system and I want it working perfectly. I don't mind if I have to do a full recap or repair any part of the motherboard (if I could).

@Malvineous: The psu is ok. I measured all voltages and they are perfectly fine. +12.15v and +5.05v. I also tried to load with an array of fan blowers and lights (I calculed around 20 - 25 watts of load) and everything went perfect. I also tried the same thing but adding all my hard drives, and again, everything ok. No heat, no weird voltage drops. So I can assume the problem is in the motherboard. I don't see any performance or stability issues with it, so there is a chance that the problem isn't serious. At least the system is fully working, but I'm not using it until I find where is the problem.

EDIT: Guys, I have a doubt: What does exactly the motherboard regulator? I mean the heatsinked one. It only provides voltage to the CPU or is it powering anything else? I ask you because I'm testing the motherboard alone. No CPU, no RAM, no cards, no anything. The connector doesn't gets hot anymore, but the heatsink on the regulator is still freaking hot. If there is nothing pulling current off the regulator; why is it heating? Maybe oscilation issues? I double checked that my modded heatsink is properly seated and isn't touching anything that it shouldn't.

I think the problem maybe the the wire from you LED display to your motherboard FSB setting.. Maybe this connection is shorted.

That wire isn't connected at tests. And I measured it for shorts when I soldered it and nothing... Think this overheating problem is clearly before that. I had to make a custom DIY heatsink for the regulator. But I never thought all that heat was a malfunctioning issue.

Sounds like the regulator is operating at close to maximum when there's nothing in the board, and when you add the CPU that's enough to push it over the limit and cause the overcurrent.

The CPU would have lots of filter capacitors to ensure a very smooth voltage - maybe one of those has shorted? It might have a high enough resistance to simply draw a lot of current (and probably heat up a lot) instead of causing a direct short circuit and exploding.

I had the same issue: Crispy AT Power Connector

I'm not 100% sure what happened. My theory is that the pins on the AT power connector were oxidised with age and/or the springs weren't tight enough in the connector. Once overheating starts to happen it will only get worse, as the heat causes extra oxidisation, which causes extra heat.

I was running lots of cards and a Voodoo3 PCI at that time too, which didn't help.
Luckily that board also had an ATX connector, so I changed to using that. It's been fine since.
I believe there are more 5V ATX pins and each is rated slightly higher too. Unfortunately that's not an option in your case.
I think you are looking at replacing that AT connector or solder power wires directly to the board.

Ugh! That's a nasty burning! But how about the regulator getting too hot without load? Well, probably there is more than just one problem with my rig. I'll make a temporal connection with terminals so I can see if oxide is a problem in my case or not.