I actually did test with the other jumpers off, not all at once but one by one. Set the multimeter to continuity testing, connected the leads to 1 and 2 on j11, and removed all the other jumpers one by one then replacing them. The continuity didn't break at all.
I'll look at VOLDET though I don't have much hope.

stamasd wrote on 2022-01-25, 01:46:

Crap. I think I won't be able to run 3.3V CPUs on this board. I did some tracings and measurements, and I think that despite the […]
Show full quote

Crap. I think I won't be able to run 3.3V CPUs on this board. I did some tracings and measurements, and I think that despite there being a pad and holes for a VRM, installing the regulator chip will not work. Let me explain why.

****
Except that on this board, I cannot see any such bridge. There are no 2 pads connected by a wire, nothing that looks that it could be opened up. Neither on the front nor on the back. See the close-up pictures above and try to find one, I couldn't. I actually examined a much larger portion of the board, both front and back, than is shown in the pictures. Nothing.
****

What about J22 (The one that is near the cpu socket) ?

Oh shit. J22, I completely overlooked that one, because it's on the opposite side of the CPU socket. Yes, I have to try that one, but that has to wait until I get home tonight. Thanks for pointing it out to me. 😁

There's also a J26 which has a bridge, but that one's down too far, between the chipset and the BIOS chip. Which makes me wonder if there may be a J23 and a J25 somewhere also. 😀

(edit) yes there is a bridged J25 just below J24. Which leaves only J23 missing.

(edit2) J23 is also a bridged jumper, located at the other end of the motherboard between the last PCI and first ISA slot, below the I/O controller. Not a good candidate for CPU voltage control either.

J22 and J25 are the usual suspects at this point. Because they are the closest to the CPU. It would only make sense to put something that controls CPU power close to the CPU.

Belated update, but I was unsuccessful.

I undid all of those bridges and replaced with jumperable headers. None of them did anything for the voltage. The connection between 1-2 on J11 remains, even with all those bridges open.

I guess it's impossible to run anything other than a 5V CPU on this board.

So after looking through more of my CPU stack, I came up with 2 "best" candidates:
1. DX4OPPR100, which I know runs stable at 120MHz (40x3)
2. PODP5V83 pentium overdrive.

I'll test them both, but I have the feeling that the pentium overdrive will win.

I don't know if the PODP is overclockable at all, it would be awesome if it ran on 40MHz FSB for 40x2.5=100MHz... but I won't hold my breath.

The routeing on the back from J11 pin 3 to TC5 runs very close to J11 pins 2&3 and I can't see any resist between pad and trace. Any chance a slight misalignment has led to a short? Although if there were, then that would have pulled Vadj from the 3.15V you measured up to 5V. So probably not. Also, it's possible (maybe not likely) that voltage jumpers may have been done in parallel; I've seen boards where you have to move (say) 3 jumpers to switch from 5V to 3.3V. So it's possible that testing by removing and replacing jumpers one by one won't pick that up. OTOH, I don't immediately see any of the other jumpers with large power traces.

I'm starting to grasp at possibilities, but maybe remove J11 and check there's nothing shorting out beneath it. I've seen slivers of solder end up in odd places before.

I figure since they've fitted R55 and R57 that it must have been designed to be able to switch from 5V to whatever the regulator was set to. I couldn't make out in the photo what values they are, do they make sense for generating 3.3V? With Vout stuck at 5V then the voltage divider ratio sounds right (3.15/(5-3.15)). If there was a design level defect then (surely) they just wouldn't have fitted those resistors, and waited to fix the problem on the next board revision. It does seem odd that information about the board says it can do 3.3, 3.45 and 4V, but there doesn't seem to be any way on the board itself to chose a range of output voltages.

Getting a bit more extreme, and this is probably a good way to break things, could get a bench supply with a current limit and set it to, maybe, something like 0.1V/10A (so not much power) and see if it's a weak short between J11 pin 2 and AT 5V that can be burnt out, or if they've just completely forgotten to isolate Vcore from +5.

I've measured those 2 resistors, I have the values written down somewhere... IIRC one was 250 ohms and the other 430 ohms. The 3.15V midpoint is the right value to generate a 3.375V at the regulator output, which is a viable one within specs of both 3.3V and 3.45V CPUs. It was often done this way. As for other voltages mentioned in the manual, they may have manufactured different revisions of the board with slightly different resistors for different spilt voltages. There are as far as I can tell at least 3 revisions of the board, -S3, -S4 and -S7. My board is marked -S4 but the BIOS says -S7, go figure...

Accidental or not, the short is there. I'll remove J11 eventually to look under, but for that I need a better desoldering tool (which is ordered and should get here sometime next week). There's only so much I can do with just a soldering iron and a braided copper wick.

As for burning any hidden traces with a high current pulse, I've thought about it... and decided against. Worst case scenario, it's not just a 1-2mm trace between the 2 jumpers; there probably is a whole +5V plane in one of the inner layers, and if pin2 is connected directly to that plane (Why? who the hell knows) then a current pulse may cause much more damage than just interrupting that connection. And it may not even work for that. I'd prefer to keep this motherboard functional, with its limitations... because damage done inside the inner layers would be non-fixable.

stamasd wrote on 2022-01-27, 19:32:

I've measured those 2 resistors, I have the values written down somewhere... IIRC one was 250 ohms and the other 430 ohms. The 3.15V midpoint is the right value to generate a 3.375V at the regulator output, which is a viable one within specs of both 3.3V and 3.45V CPUs. It was often done this way. As for other voltages mentioned in the manual, they may have manufactured different revisions of the board with slightly different resistors for different spilt voltages. There are as far as I can tell at least 3 revisions of the board, -S3, -S4 and -S7. My board is marked -S4 but the BIOS says -S7, go figure...

I might not be recalling correctly, but I thought the 3 pin regulators adjust their output voltage so that the voltage at the adjust pin is Vref below Vout. So in this case for Vout to be ~3.3V, then Vadj should be ~2.05V when the regulator is fitted and working. In this case we know that Vout is currently fixed at 5V and Vadj at 3.15V, so I think that tells us the ratio of R1 and R2. Which as you say looks right for what should be a 3.3V output when the regulator is there. In fact, a quick double check is that 5/3.15 ~= 3.3/2.05. With no regulator then both Vadj and Vout should be 0V as they get pulled to ground, but that's the fault of the board.

As for burning any hidden traces with a high current pulse, I've thought about it... and decided against.

Not just me then. I was thinking around 0.1V/10A on the grounds that the voltage limit would protect any components, and a power plane wouldn't even notice 10A if it is a design fault and the pin is fully connected, but it might be enough to break a small connection. Fairly high risk though, with any fixes being very bodge like (running jumper wires between capacitors and the like), so yeah, probably not worth it.

If I don't find anything obviously fixable under J11, I'll leave it be. I have decent 5V CPUs to use with it.

Speaking of which, the POD5V83's fan is dead; that's not a problem, I have plenty of small fans I can strap on top of it. The only issue is that if this overdrive senses its fan module is not working, it will underclock itself by changing the multiplier from 2.5 to 1. I think I saw in the past a method for preventing that but couldn't find the info with a search; I think the fan sense pin (one of the 3 round pads at the top) has to be pulled to ground with a resistor, but can't remember the exact details. Without that, it will run at only 33MHz not 83MHz.

Re: Pentium 200 MMX Overdrive - Fan replacement?

If we got that right, there's a sense pin that is grounded when the fan spins.

Excellent. 😀

Yeah that worked. I placed a piece of copper foil on top of the fan-sense and ground pads, and secured it with some kapton tape to the CPU. It's a temporary fix until I make something permanent, but it tested fine. Did some benchmarking, it runs approximately 25% faster than the DX4/120MHz. But for some reason Quake won't run. Quake should run, after all it's a Pentium; it has enough memory too (16MB). But it doesn't.

Good work ! I tried finding the original board manual or info but is not on Protechs archive website going back to 1997.

Yeah for a 486-class system, I think I pushed it as far as it goes. I tried overclocking that PODP5V83 with 40MHz FSB; it boots, but some benchmarks fail. I'll stay with 33MHz/83MHz.

(update) Quake runs only if I boot without himem/emm386. Weird.

stamasd wrote on 2022-01-28, 01:08:

Yeah for a 486-class system, I think I pushed it as far as it goes. I tried overclocking that PODP5V83 with 40MHz FSB; it boots, but some benchmarks fail. I'll stay with 33MHz/83MHz.

(update) Quake runs only if I boot without himem/emm386. Weird.

That is odd ! You should be able to have Himem.sys running even though Quake uses DOS4GW wihich does not need it. Know some versions of DOS4GW do not like EMM386 but Himem.sys ....hmmmm???

This is BTW the Quake bundled with Phil's DOS benchmark pack. It runs fine on another board, with both HIMEM and EMM386 active (a socket 7 board).

(edit) it does work with HIMEM only. It's funny because on the socket7 board it worked with both HIMEM and EMM386 off the same drive. The exact same files. It works on a P200MMX but not on this overdrive - same Quake, same EMM386.

I thought I'd put here some pics of the passive cooler I have on the PODP5V83 right now. This overdrive was in a sorry state when I got it, the fan and heatsink had been ripped off, many badly bent pins, one pin missing and one pin short. I unbent all of the pins carefully, turns out the missing pin is one of the unused ones on the outer ring of pins, and the short one is long enough to make contact.

I plan on adding a heatsink and fan to it, but the only heatsinks I had were too big for it: 5x5cm, fit perfectly on top of a regular Pentium but this one has the extra VRM components on top so it won't fit. I ordered smaller heatsinks (4x4cm) but until they come I used this contraption.

I have several blocks of machined pure copper in various sizes. I stacked 2 of them on top of the CPU, best-fit. And a couple of the aluminum heatsinks on top of that. No fan. The copper weighs about 700g total, and due to its mass has an awesome heat dissipation capacity. Running the overdrive through intensive testing for 20-30 minutes, the CPU and the copper get barely warm.

I use those copper blocks as temporary passive coolers for testing quite often, they work great. Even for otherwise hot CPUs like K-6. I wouldn't recommend it for long durations in this case, but 20-30 minutes is perfectly fine.

One downside is that all of the accumulated heat takes some time to dissipate after one use. Fortunately I have multiple pieces of those blocks and can swap another cool one in, while the warm one cools down on top of an anvil in my garage. 😀

One more bit of info about the 2nd IDE port not working. When I inspect system resources in DOS with msd.exe, it shows that IRQ14 is assigned to "hard disk controller", but IRQ15 is "reserved" by the BIOS. I assume that's why the 2nd IDE isn't working.

I've played with the BIOS, which is kinda weird. There is only one setting for the secondary IDE: enabled/disabled; it doesn't work if it's set to either.

Also, in the "green PC" section of the BIOS, there is a section for "make IRQs available", and one can list up to 4 of them. Whether IRQ15 is listed there or not, it still shows as "reserved by BIOS" in DOS, and the 2nd IDE doesn't work.

I think that if I want more than 1 IDE channel, I might have to use an add-on card. Do any PCI IDE controllers even work with this class of 486 motherboards?

stamasd wrote on 2022-01-30, 15:28:

One more bit of info about the 2nd IDE port not working. When I inspect system resources in DOS with msd.exe, it shows that IRQ1 […]
Show full quote

One more bit of info about the 2nd IDE port not working. When I inspect system resources in DOS with msd.exe, it shows that IRQ14 is assigned to "hard disk controller", but IRQ15 is "reserved" by the BIOS. I assume that's why the 2nd IDE isn't working.

I've played with the BIOS, which is kinda weird. There is only one setting for the secondary IDE: enabled/disabled; it doesn't work if it's set to either.

Also, in the "green PC" section of the BIOS, there is a section for "make IRQs available", and one can list up to 4 of them. Whether IRQ15 is listed there or not, it still shows as "reserved by BIOS" in DOS, and the 2nd IDE doesn't work.

I think that if I want more than 1 IDE channel, I might have to use an add-on card. Do any PCI IDE controllers even work with this class of 486 motherboards?

Have I understood correctly that any device connected to the 2nd IDE port prevents system booting from a device connected to the 1st IDE port ?
If the 2nd IDE was disabled in BIOS settings would the 2nd IDE port work in Windows or with IDE DOS driver?

I don't have windows installed on this machine. And what DOS IDE driver do you have in mind? I don't know of any.
Also, the boot failure if a device is attached to the secondary IDE happens regardless of whether the secondary IDE is enabled in the BIOS or not.

stamasd wrote on 2022-01-31, 08:17:

I don't have windows installed on this machine. And what DOS IDE driver do you have in mind? I don't know of any.
Also, the boot failure if a device is attached to the secondary IDE happens regardless of whether the secondary IDE is enabled in the BIOS or not.

Ah, so drivers won't help