Really good examples. But, using a real old car is something like, you can not drive it at roads. You need to have a special paper first, and it is for a short time only, for filming purposes or transporting etc. It can not give you the real comfort. My mom has a 1965 fastback Mustang, I have bought it a better alternator https://www.ebay.com/itm/173286094641 well, but what about an air condition? Very difficult, I have lately heard some people are successful at fitting, but really difficult. https://www.classicautoair.com/shop/complete- … tioning-system/ I really need to work for it hard, later. Then what, at an emergency, can you simply call Ford from car audio system, no. There are in the end some limits. But, this is a car, not a computer, a computer needs digital highway, not an ordinary road. So, if a computer is slow, you can not change everything.

I have my oldest Compaq ProSignia 486, on board SCSI good, but slow 50 pin. I have Compaq QVision 1280/E with 2 MB ram, no more from Compaq for EISA. At most can be add a Matrox 4MB but very difficult to find. Any 16/9 resolution support is not present in anyway, so no good monitor support. There is an empty 40 pin 512kb ram place for the onboard Cirrus Logic GD5420, I will try to solder a Texas Instruments TMS45165DZ -70 ram at a spare time, maybe I can reach to 1 MB onboard, risky if things go wrong, board may die, here trying to force it is equal to playing Russian roulette, the place may be for an additional 512 kb or replacement place for the 4 piece 44256 chips on board .There is no mainboard documentation, 40 pin ram connections seems OK according to TI TMS45165DZ -70 chip datasheet, for the chip three VSS pins must remain at left side and three VCC pins remain at right, for the bottom picture pin number one orientation of the ram, is at the low right side. Total system ram 128, no more option, no speed increase. Processor, up to Pentium Overdrive 83 MHz, or AMD 5x86-133, or Cyrix 586-120 etc. Those are more than a 1 to 10 thousand times slower than todays processors, a good way to dig with a pin.

I am a good repairer nevertheless, at Pandemic days I have repaired an Intel i815 D815EEA, an AMD A7V, a Slot1 Asus P3V4X, a Matrox G100 by bios rescue, a NEC PCI USB 2.0 card, an Intel 850E 478 pin EMV850E mainboard etc. But, those are only for fun, how long did I use them after repair? Not much. Except Intel 850E chipset one, they can not enter into internet from known browsers, only Firefox continues to support on Windows 7 for 478 pin SSE2 processors.

In conclusion, those 486 to Pentium III mainboards are not possible to go into digital highway, no permission slip for them are possible. 😀 I hope, there won't be a a real bad sun explosion, which can kill from most newer to older, electronics stuff. I don't want to see, only old stuff left remained, but they may die as well. Those old one's are really too far away from handling our todays needs, have a go for a 4K film watch with them, no of course. Besides, I am a Radio, TV & Film studies graduate from year 2000. I don't want to watch anything from Betamax & VHS cassettes on a tube TV.

I want to close the discussion here, I really like old computer stuff as well like the people here at vogons.org, but I like to use new ones better.

New toys always welcome!

karakarga wrote on 2023-02-26, 18:25:

Really good examples. But, using a real old car is something like, you can not drive it at roads. You need to have a special pap […]
Show full quote

Really good examples. But, using a real old car is something like, you can not drive it at roads. You need to have a special paper first, and it is for a short time only, for filming purposes or transporting etc. It can not give you the real comfort. My mom has a 1965 hatchback Mustang, I have bought it a better alternator https://www.ebay.com/itm/173286094641 well, but what about an air condition? Very difficult, I have lately heard some people are successful at fitting, but really difficult. https://www.classicautoair.com/shop/complete- … tioning-system/ I really need to work for it hard, later. Then what, at an emergency, can you simply call Ford from car audio system, no. There are in the end some limits. But, this is a car, not a computer, a computer needs digital highway, not an ordinary road. So, if a computer is slow, you can not change everything.

I have my oldest Compaq ProSignia 486, on board SCSI good, but slow 50 pin. I have Compaq QVision 1280/E with 2 MB ram, no more from Compaq for EISA. At most can be add a Matrox 4MB but very difficult to find. Any 16/9 resolution support is not present in anyway, so no good monitor support. There is an empty 40 pin 512kb ram place for the onboard Cirrus Logic GD5420, I will try to solder a Texas Instruments TMS45165DZ -70 ram at a spare time, maybe I can reach to 1 MB onboard, risky if things go wrong, board may die, here trying to force it is equal to playing Russian roulette, the place may be for an additional 512 kb or replacement place for the 4 piece 44256 chips on board, there is no documentation, ram pins connections seems OK according to TI TMS45165DZ -70 chip datasheet for the 40 pin chip, VSS pins must remain at left side and VCC pins remain at right, for the picture pin number one orientation of the ram is at low right side for the attached picture at the bottom. Total system ram 128, no more option, no speed increase. Processor, up to Pentium Overdrive 83 MHz, or AMD 5x86-133, or Cyrix 586-120 etc. Those are more than a 1 to 10 thousand times slower than todays processors, a good way to dig with a pin.

I am a good repairer nevertheless, at Pandemic days I have repaired an Intel i815 D815EEA, an AMD A7V, a Slot1 Asus P3V4X, a Matrox G100 by bios rescue, a NEC PCI USB 2.0 card, an Intel 850E 478 pin EMV850E mainboard etc. But, those are only for fun, how long did I use them? Not much. Except Intel 850E chipset one, they can not enter into internet from known browsers, only Firefox continues to support on Windows 7 for 478 SSE2 processors.

In conclusion, those 486 to Pentium III mainboards are not possible to go into digital highway, no permission slip for them are possible. 😀 I hope, there won't be a a real bad sun explosion, which can kill from most newer to older, electronics stuff. I don't want to see, only old stuff left remained, but they may die as well. Those old one's are really too far away from handling our todays needs, have a go for a 4K film watch with them, no of course, besides I am a Radio, TV & Film studies graduate from year 2000. I don't want to watch anything from Betamax & VHS on a tube TV.

I want to close the discussion here, I really like old computer stuff as well like the people here at vogons.org, but I like to use new ones better.

New toys always welcome!

Three years ago I helped a friend of mine install a AC kit from vintage air on his 67 mustang. It's was a very simple straight forward job took us less than 3 hours start to finish. We finished a full 100% restoration of the car in less than two years just working on it when we had some spare time.

karakarga wrote on 2023-02-26, 18:25:

.....

I have my oldest Compaq ProSignia 486, on board SCSI good, but slow 50 pin. I have Compaq QVision 1280/E with 2 MB ram, no more from Compaq for EISA. At most can be add a Matrox 4MB but very difficult to find.

........

🤣. A Compaq QVision 1280/E is also VERY hard to find.

I meant very old cars, 1900 around. Thanks for the info, it looks difficult a bit. I will see....

Yes, I think so, QVision 1280/E is also very expensive, finding a 4 MB Matrox EISA card is almost impossible.

karakarga wrote on 2023-02-26, 22:51:

I meant very old cars, 1900 around. Thanks for the info, it looks difficult a bit. I will see....

Yes, I think so, QVision 1280/E is also very expensive, finding a 4 MB Matrox EISA card is almost impossible.

Any time. What I liked about the Vintage Air kit is it's completely bolt in and it comes with every bracket and all the hardware. There is no fabricating or custom work needed at all.

karakarga wrote on 2023-02-26, 22:51:

I meant very old cars, 1900 around. Thanks for the info, it looks difficult a bit. I will see....

Yes, I think so, QVision 1280/E is also very expensive, finding a 4 MB Matrox EISA card is almost impossible.

Hello,
Would you be so kind and make a good picture of your QVision 1280/E?

or is the discussion now only about cars? If yes, please feel free to ignore the request.

Just got my first PODP5V83 after long hunting... And of course it won't do 100MHz as I hoped, posts but crash instantly when booting/after boot.

Now I'm thinking will I take the road to modify the CPU (original heatsink is long gone anyway) or just keep it that way...

By the way, it's missing that fourth capacitor.

Wow, this thread has two things that interest me:

The Pentium Overdrive (which I have running at 100mhz seemingly stable but am curious what folks here can squeeze out) and classic Mustangs, of which I have a ‘68 that I’m restoring 😀 The exterior and all mechanical parts are done. Now it’s just the interior left. Woot!

Got the POD to 4 volts and it's almost stable at 100MHz but in long term testing it will eventually crash.
Bought another one and hoped it would overclock easier but of course with my luck it's the same thing. Won't even boot at 100MHz.

I think I'll just give up with these and use at default clocks...

Any thoughts on whether this would affect the stability of the BIU and not necessarily the core?

I've got a POD63 where it is particularly sensitive to bus speed and won't run stable at or above 33MHz FSB even with the multiplier forced to 1x

Hello there.

I received a couple of PODP5V83 and, despite both booting as-is without issue at 40 MHz x 2.5, I couldn't get Quake timedemo to run without crashing.

So I first decided to try the simple Schottky diode mod between the regulator and the CPU grounds, but I couldn't go past 3.7 V. One or two diodes in serie didn’t make any difference. Same thing with a variable resistor, even crancked to the max.
The small extra voltage wasn't enough to run a stable system.

Then I went with @feipoa 6 A diode solution between the CPU 5 V output (I managed to desolder and lift the leg of the regulator instead of cutting it) to the regulator 3.3 V output. And I now get sufficient voltage for my system to be perfectly stable.
I must say that it took a lot of time and effort to manage to solder the legs without removing the heatsink, but I think the latter would have been more of a hassle.

This is a version of the Overdrive that hasn't a fourth capacitor in the corner.
Hence I could use the extra clearance to fit the big 6 A diode in that space to make the result tightier. Another benefit is that it is cooled a bit by some of the fan airflow.

I ran different benchmarks/stress tests while measuring the voltage and I was surprised with the fact that its average value varied more or less than when in idle based on the type of test.

At idle, the voltage stays around 3.9 V.
While running Doom (score: 1410 realticks) or Quake (23.5 FPS), I get about 3.85 V.
When running Topbench (score: 317), it rises to 4 V.

I'm not sure what could explains this. I would have thought that the variability would go in the same direction between the CPU being idle and being under stress, independently from the type of benchmark.

Diode dropout voltage varies with current, and also with the junction temperature.
The variation is direct with current (higher current = higher voltage drop) and inverse for the temperature (higher temperature = lower drop). You are probably seeing a combination of the 2 effects.

Ok, but then I would have expected the current/voltage drop to increase for both the Doom/Quake AND the Topbench stress tests.
Or based on what you said, when running Topbench, the diode temperature rises to such a level that the voltage dropout is less than when being idle, even if more amps are pulled. I will try to measure the diode temp while running these different tests then.

Can I confirm that you've cut the Vout lead of the onboard VRM, such that the only power going to the CPU is through from the diode? I want to ensure that you aren't running the diode in parallel with the VRM's output.

If my memory is correct, I was getting about 4.10 V to the CPU while running Quake.

I’ve lifted both the Vin and Vout legs of the regulator from their CPU pads. But the GND is still attached. Is that a problem?
My understanding was that the diode is a full substitute of the VRM to downstep the voltage. It could actually be removed completely, no?

xmamat wrote on 2024-10-31, 11:06:

I’ve lifted both the Vin and Vout legs of the regulator from their CPU pads. But the GND is still attached. Is that a problem?
My understanding was that the diode is a full substitute of the VRM to downstep the voltage. It could actually be removed completely, no?

Yes, the 6 A diode is a replacement for the onboard VRM. Just that your voltage getting to the CPU seems too low. What do you measure for the voltage going into the diode? Should be 5 - 5.10 V.

Yes, you can completely bypass the diode and wire 5 V into the Pentium, but I haven't run any tests on this to see if it will work and how long the CPU will last.

Leaving GND connected to the VRM should be fine.

feipoa wrote on 2024-10-31, 11:46:

Yes, you can completely bypass the diode and wire 5 V into the Pentium, but I haven't run any tests on this to see if it will work and how long the CPU will last.

Sorry, I meant to remove completely the VRM, not the diode 😀
I would not risk sending that much voltage to the CPU even if it’s 5 V tolerant.

What would be the consequence of injecting the diode output to the 3.3 V without disconnecting the regulator Vout from the CPU since it wouldn’t receive anything in the Vin anyway?

I’ll double check how much I get on the diode input. But I think I already did before the mod on the VRM Vin and it was below 5 V indeed.

I didn't quite understand your question in the middle paragraph, but it sounded like running the diode in parallel with CPU's load, in parallel with the VRM's output. I think the diode would burn out as the VRM wants a 3.3 V across it, while the diode wants 0.7 V. Forcing 3.3 V across the diode without a current limiting resistor should fry the diode in short order.

Curious what voltage you are measuring before the diode. Based on your numbers, I'd guess that only 4.6 V is getting in. Sometimes when the connectors wear out on ATX or AT PSU's, or the motherboard connector, you can see this much voltage drop. Normally the connector will feel a little warm.

On input, I measured 4.7 V at idle, 4.65 with Doom/Quake and 4.8 V with Topbench. So a 0.8 V drop based on previous output measures.
The diode is about 40 °C all the time.

I got the same results with another motherboard.
It looks like the problem is that my test PSU is a bit weak. I measured only 4.85 V on a the 5 V rail at idle.

Assuming you are measuring the voltage of the PSU before it leaves the connector, then it might be time to find another AT PSU. If you get 4.85 V after it leaves the connector, then sometimes you can loose voltage on the actual connector when they aren't making good contact.

My favourite AT PSU is the one which output 5.10 V, which normally accounts for any contact loss or loss from trace impedance.