I have an upgraded M530 board that is similar. Jan made a BIOS for it that supports the K6-3+ and it might also work on yours. That thread is here:

www.vogons.org/viewtopic.php?t=73635

There are several recent threads about problems trying to measure the voltage on an empty Socket 7. From the picture in the link you posted, it looks like plain linear regulators. The pair probably means it can supply dual voltages, but you would probably be limited to early/slow CPUs. With a VRM or socket adapter of some kind you could run the later chips like the K6-2 etc.

The built-in 512KB of cache and up to 75MHz bus is nice. I think you would be limited to 128MB of RAM. It looks like you also need a clock chip--be sure to get the correct kind.

Anyway, I like mine but whether it's worth your time is up to you.

128MB is max RAM of the i430VX, but it can only cache 64MB, so unless you're doing something that results in disk thrashing, don't install more than 64MB.

The M538 seems to share a lot of design with theM537 which is notorious for power delivery problems, with a linear 5V-3.3V regulator that often fails and frequently takes out other components on its way down. So not surprised to see you might be having issues there.

As for whether it's worth trying to fix - the question is if you're in it for the process (enjoying fixing, learning how to / how to do it better) or the end result (this board, working as well as it did before the power failure). In the latter case I wouldn't bother - it one of the lowest quality boards implementing arguably the least interesting Intel So7 chipset which is moreover very common, so it's not like better alternatives don't exist. However those exact same characteristics make it an ideal board to tinker with and/or to go outside your comfort zone when trying to fix things: worst-case you haven't destroyed much, and the really thin PCB makes it relatively easy to (de)solder components.

Thanks. I'm doing this mainly for fun and learning, but it would be nice as reward in the end to have the mobo survive and perform right.. it would be such a let down if the repairs will fail soon again due to inherent design issues. That's where I prefer to not invest much buying parts for a mobo that I'll throw away soon after.

If no reported compatiblity issues or crippling perf is known for this mobo, then I guess I can give it a try without investing too much into it.

There are two power modules on this board. Each made of two D2583 transistors by NEC. One module is preadjusted for 3.3v for VIO, and the other is connected to a configurable set of jumpers that connect resistors of various values to adjust the base of the Vcore transistor. VIO is measuring fine, but Vcore is stuck to 4.8V, which will basically destroy the CPU.

Looking at those transistor's datasheet, they are not even classified as adjustable voltage regulators.
https://datasheet4u.com/datasheet/NEC/D2583-944378
This drives me nervous for finding a replacement that will work with the existing set of resistors. I'm expecting a replacement could respond a little different, which will throw off the mobo's prescribed jumper settings for Vcore.

Ugh, that's ugly, even by PC Chips standards...

Explains why they burn out: that thing is rated at max 5A DC; if using as linear regulator, current in at source voltage will be basically equal to current out at set voltage, with the difference in power burnt off. So max power it could deliver is 5x2.8V=14W (at 2.8V at least). Now look at Pentium MMX CPUs' TDP. P166MMX draws max 13.1W, but everything faster goes well over 14W, so will draw more than 5A over the transistor.

Similarly, it's rated for max 10W dissipation. At 5V->2.8V i.e. a voltage drop of 2.2V, that means max 4.54A, so even a P166MMX (4.67A) will be overheating it slightly. The only way to use a Pentium MMX with this board within spec of the 'regulator' would be to set voltage at 2.9V (within spec for the P55C, if not nominal) as that would give you a voltage drop of 2.1V allowing 4.76A - just enough to handle a P166MMX without overheating. The P166MMX will run slightly hotter, but unlike this piece of underengineered mess, they are massively overengineered and can easily handle the extra 0.4W.

I share your concern about a replacement part, if choosing another audio amp transistor, you'd want one with the same DC gain. If going for an actual linear voltage regulator, I'd not be sure how to translate specs. You'd probably need to figure out what the jumper circuitry exactly does (probably voltage divider into a voltage feeding into the transistor) and look up a solution based on that.

If you're sure the regulator is bad, you could disconnect/remove it, and test the board with a single voltage CPU. If the board is still alive, then you could rig up a new/better source for Vcore. There are some examples here:

www.vogons.org/viewtopic.php?t=86611