Is your PSU 100% o.k. beyond all doubt?

-

You might be right about the power supply. It's from an old Dell Dimension from the early 2000's, and I have had one of those fail before on a different machine. I'll try out a different one when I get back from work. Do the original Pentium Overdrives really have L2 Cache issues?

ODP 's have an internal L1 which makes the boards cache an L2, have a couple of them and they do not like "write-back" enabled in bios iirc.

The AP4's bios has an option to set MB Cache to write-through. I tried that and it didn't help. Also I never had cache issues prevent a machine from posting; it was always after post when a cache issue would lock up the machine. At least that was the case with my SS7 machine when I try a fsb > 100mhz without disabling mb cache.

3dnow wrote on 2023-03-16, 14:22:

.....
However, a week after that (yesterday) it would no post at all no matter what I tried.
.....

What does that mean exactly "it would not post"?
What is the last message that is written to the screen?

Nothing on screen at all and no PC speaker beep.

Oh that sounds more like a board failing. Runs ok at 25Mhz but not at 33Mhz ? Had a soc 4 do similar (run ok lower speed clocks but not higher) and was bad caps.

Horun wrote on 2023-03-17, 02:21:

Oh that sounds more like a board failing. Runs ok at 25Mhz but not at 33Mhz ? Had a soc 4 do similar (run ok lower speed clocks but not higher) and was bad caps.

i second this, bad caps or maybe a vrm not powerful enough... the pentium overdrive can be quite power hungry

I busted out my test bench power supply and multimeter. The VRMs are fine; they both output their required voltages though they can get quite hot. There is a single electrolytic cap near the socket that is the culprit. There is a sizable voltage drop on that circuit during startup that is only present with the fsb at 33mhz. The cap is not breeched, but it isn't able to help the circuit maintain 5v during startup. It's a good thing these older motherboards have their components more spread out making them easier to work on.

While I wait for my new cap to arrive, I'll keep the fsb at 25 with the P Overdrive. It is still moderately faster than a 5x86 in the games I play a lot on that computer (whiplash, tomb raider, quake, screamer 2) thanks to the FPU. Also Pentium CPU's get a much larger performance boost when using a 3d accelerator card compared to a 486. IME the 5x86 gets about a 50% framerate increase with Glide over software. The Pentium Overdrive can get around a 75% increase (in Whiplash and GLQuake). Thank you for the help everyone! 😁

Are you replacing this capacitor?

I can´t believe it´s involved in maintaining 5V stable. This is the PSU´s job.

majestyk wrote on 2023-03-17, 05:59:

Are you replacing this capacitor?

pvi486ap4s1.jpg

I can´t believe it´s involved in maintaining 5V stable. This is the PSU´s job.

A capacitor is like a filter. It filters out noise by adding current to a circuit to prevent voltage drops and removes excess current during a power spike to prevent over-volting. The capacitor on this board cannot store enough energy to maintain proper voltage to the CPU at startup. It also explains why the computer could run for hours after many false starts in the past, the greatest amount of current is needed during startup as opposed to normal operation. Also my 600w evga power supply with a massive 20 amp 5v rail (100w is more power than the whole computer takes) could not get the machine to post at 33mhz fsb.

All true, but at startup during the first fraction of a second this capacitor, if it is fully functional, has a low resistance before it´s completely loaded so it´s an additional burden for the 5V rail.
If the 5V rail breaks down at startup, a new or better capacitor at this place will not help.
Later, when everything has reached some equilibrium, it indeed does improve local stability when transients / load shifts happen.
Are you sure this electrolytic is at the input (5V) side of the regulator? I would expect it to be at the 3.3V side.

p.s. This electrolytic is just about 5 cm away from the PSU connector. Only some high frequency filtering (1nF - 4.7µF) would make sense there.

On the other hand, it seems like the 3.3V regulator is the one on the left, so the right one could be a MOSFET for auto-switching between 3.3V/5V CPUs (reading the VolDet pin at the CPU). If the MOSFET is switching very fast / hard, a local stabilisation for the 5V rail _could_ be necessary to prevent transients.

Yes, a "transistor" for selecting 5v or 3.3V and a regulator. Unusual in this way.

Cheers,

majestyk wrote on 2023-03-17, 18:09:

On the other hand, it seems like the 3.3V regulator is the one on the left, so the right one could be a MOSFET for auto-switching between 3.3V/5V CPUs (reading the VolDet pin at the CPU). If the MOSFET is switching very fast / hard, a local stabilisation for the 5V rail _could_ be necessary to prevent transients.

The 3.3v regulator is definitely the one on the left since it gets very hot when the 5x86 is being used and not the overdrive.

I have also heard many things about the Asus PVI-486AP4, and not many of them good; so all of this frustration could be due to it just being one of those motherboards that does not properly support the 83mhz overdrive. I'm still going to replace that cap, and if it does not help, it's no skin off my bones. I did get the PO for the price of the slower 63mhz one.

This forum is really awesome! I used to browse it for days when working on previous builds before I became a member!

I have a thought. Maybe the transistor switch is bearing the heavy current of the pentium overdrive 83? Find a way to bypass it directly to a 5V rail so 5V full power can be delivered direct to CPU with a jumper if your manual have one?

Cheers,

HUZZAH!!! I got it to post with no modification! But you would not believe what I had to do.

#1: Set fsb jumper combination to 25mhz.
#2: Power on and go to bios. (Post should display 66mhz)
#3: While PC is running carefully set fsb jumpers to 33mhz.

#4: Press arrow keys to make sure the PC is not frozen. If all is well, then exit discarding changes.
#5: The PC will reboot displaying an 80mhz clock:

I ran several tests to confirm if the system was performing as it should be. You can check too with the benchmark results I posted previously.

At the time of this post the PC has been running the GLQUAKE demos for 4 hours now as a stress test. This confirms my theory that the Pentium Overdrive 83 does not get enough power at startup, but once it gets going it has all the power it needs. This is similar to how an A/C compressor needs capacitors to help it at startup. There has to be an underperforming cap on this board.

3dnow wrote on 2023-03-18, 17:39:

HUZZAH!!! I got it to post with no modification! But you would not believe what I had to do. […]
Show full quote

HUZZAH!!! I got it to post with no modification! But you would not believe what I had to do.

#1: Set fsb jumper combination to 25mhz.
#2: Power on and go to bios. (Post should display 66mhz)
#3: While PC is running carefully set fsb jumpers to 33mhz.
IMG_20230318_131553_737.jpg
#4: Press arrow keys to make sure the PC is not frozen. If all is well, then exit discarding changes.
#5: The PC will reboot displaying an 80mhz clock:
IMG_20230318_132033_090.jpg

I ran several tests to confirm if the system was performing as it should be. You can check too with the benchmark results I posted previously.
IMG_20230318_130735_132.jpg

That is really weird.
So that would mean that the onboard voltage regulator is too weak.
The solution then is to force 5 volts (the CPU has it's own voltage regulator)
I don't know how to do it on this board.
So you could just cut of the VOLDET pin of the P24T, it is T5
http://www.dickhardtstrasse.org/Hardwarehandb … ml?cpu_486.html