VOGONS


First post, by nickles rust

User metadata
Rank Newbie
Rank
Newbie

As part of my ongoing Socket7 quest I got an Epox MVP3C board. I flashed the bios with one from www.steunebrink.info/k6plus.htm (thank you Jan Steunebrink!) and after installing an AMD K6-3+ it ran great for a while then I had a catastrophic kb that killed the CPU and some other stuff. To effect a repair I first replaced the (obviously dead) main FET on the buck converter with an FDB6670AL. There are plenty of better power FETs available these days, I just had one of these handy. The original FET went up in smoke but from other pictures I've seen it may have been a B40NE03L. I tweaked the resitor network on the feedback to lower the base voltage to 1.9v (instead of 2). This worked well with a few sacrificial CPUs then I risked another K6-3+ and it's working well again. I noticed that the inductor was getting quite warm. The controller chip is a US3033. I can't find a datasheet for this chip that is not behind a pile of javascript/paywall/general-dumpster-fire-modern-internet but it looks like the main inductor should be around 4uH. The hot inductor is 11 turns on a blue/green (what looks like a) T50 core. I wound a new inductor that is 9 turns on a red/green T68 core. This is around the same value but runs much cooler.

The attachment epoxVRM.jpg is no longer available

So in case anyone else is using one of these boards with newer K6 CPUs, learn from my mistake and upgrade your VRM. Mainly, replace Q11 with something newer/better. You can also check the temperatures of other components and make adjustments as needed.

Reply 1 of 3, by majestyk

User metadata
Rank Oldbie
Rank
Oldbie

I hope you replaced all the green electrolytics in / around the VRM circuit first.
Otherwise it will not operate properly. The K6+ CPUs draw less power than Pentium, K6-2 or K6-III types so the FET should never be stressed. They only get steressed when the old electrolytics are still present. The resulting core voltage that contains high AC components can kill the CPU also.

I have repaired about 50 of these Epox mainboards and in all cases the green electrolytics (about 10) )had sky-high ESR and/or lost capacity.

Reply 3 of 3, by nickles rust

User metadata
Rank Newbie
Rank
Newbie

Thanks for the replies guys! I measured those caps and they actually seemed OK so I left them. I did add a few tants to the back side for good measure. You can see one in the picture on the +5v input to the buck converter. I used 9 turns of 16awg on the T68 core as described, which should result in a value of 3.2uH. The inductor now runs cool to the touch. The epox implementation looks pretty close to that datasheet. Nice find; thanks for posting it!