Is it JB weld I noticed it was metallic whatever it was solvents even acid did nothing to it I used a hard rubber buffing wheel for the Dremel it also went through extra solder and made it nice and smooth, the ceramic that the CPU is made out of is very hard the buffing wheel doesn't do anything to it other than polish it. It's interesting that you measured those high voltages 4.90 V is probably more than CPU needs to run stable at 100mhz also you mentioned in another thread that you were able to chill a unmodified POD and get a successful Quake demo at 100mhz heat seems to be destabilizing the POD a different VRM to give 3.60-4.00 V may be the way to go, I have read of people putting different VRM on the mobo to OC AM5x86 and other 3.3 V CPUs, I like the external VRM idea too. In theory the POD should take 5 V but these are more or less the reject socket 4 cores

A POD83 should work with 5 V? Won't it overheat?

I decided to sacrifice the sharpness of one of my wood chisels to remove the thermal epoxy. It came off quite easily with a few blows.

How do you know this is JB Weld? The thermal epoxy that Intel used may just have the same colour as JB Weld, but be something entirely different. I think Omega Engineering makes some high quality thermal epoxy that is grey colour. They also make one with slightly less thermal conductivity than the grey one, which is white colour. The problem I have with all of these epoxies is the shelf life. Usually 2 years max.

I don't know if was JB weld for or not just seem to metal based. As for the POD at 5V I keep forgetting that the socket 4 PODs are 3.3 V not 5 V like the first Pentiums 60/66mhz, but according to literature of the time the socket 3 and 4 PODs were built to same specs of the 5 V generation of Pentiums so that also gets me mixed up sometimes. I actually didn't know about socket 4 until about a year ago when I found a socket 4 desktop in a dumpster finally found the proprietary cache module for it too.

I have determined that the VRM on the POD83 is normally fixed at 3.50 V. The 4 SMD components on the cermaic package are all 3 uF capacitors. To increase the voltage from 3.50 V using the POD's voltage regulator, splice in a variable resistor (trimmer) in series on the VRM's GND pin. I used a 1K trim pot. This allows voltage adjustment from 3.50 - 5.0 V. The problem is that the POD's VRM can only output so much current, which seems to limit the coltage increase to 3.73 V on my system (at load). Even if you increase the voltage to 3.8+ V, under load, the voltage drops to 3.70-3.73 V. Note that the red clip in the photo is so that I can measure the voltage on the VRM's centre pin.

For this modification, I selected one of my POD chips which doesn't quite cut it at 100 MHz. Before the mod, DOS Quake would not load. After upping the voltage to 3.73 V, DOS Quake loads and begins the timedemo, however it crashes shortly thereafter. So this mod may work for you if you only need an extra 0.1 to 0.2 V. If you need more voltage, it might be best to change the POD's VRM entirely, preferably to something which can output up to 5A. Note that the value of the trimmer is approx. 30-ohms with a 3.73 V output.

Alternately, you might try following the original approach, which cuts the 5V line from the POD's VRM, but adding one or two diodes in series prior to bridging to the "3.3V" pin on the VRM. I might try this with some 6 A diodes I located in my bin.

It makes sense that the POD needs more Amps rather than Volts, many modern motherboards are made with high output components to give better overclocking stability at a lower voltage. Feipoa I am glad you took a interest in this because I have limited knowledge on circuitry

tried a 50/50mhz boot like I had done before by pulling the fan but the fan signal seems to be null with all the on chip VRM circuitry gone it will boot at 2.5x multiplier with or without the fan.

I'm having difficulty understand this. Are you saying that with the surface components removed, the POD now defaults to 2.5X without the need for a fan signal?

SMD components are required for voltage stability. In fact, without the SMD components, I only receive 2.6 V from the VRM. Putting the SMD components back, I get 3.5 V.

Anonymous Coward wrote:

tried a 50/50mhz boot like I had done before by pulling the fan but the fan signal seems to be null with all the on chip VRM circuitry gone it will boot at 2.5x multiplier with or without the fan.

I'm having difficulty understand this. Are you saying that with the surface components removed, the POD now defaults to 2.5X without the need for a fan signal?

Yes it does not need the fan signal now, I don't know what else would have made it go to 2.5x it happened after I removed the components

I added a 6 ampere diode as the bridge between the "5.0 V" node on the POD's ceramic surface and the "3.3 V" node on the POD's ceramic surface. Also, I cut the VRM's lead which connects to the "5.0 V" node. The voltage drop from the diode allows the POD to run at 4.00 V internally instead of 3.50 V (the default value). Note that "5.0 V" and "3.3 V" are just labels I used in my diagram to define certain nodes on the POD. Coming into the POD, I measure ~4.90 V. This means the diode is dropping ~0.9V at whatever current is going through it, probably 3.5 - 4.0 A.

6 ampere diodes are quite large, having a package size of 9 mm diameter and 9 mm long. For someone else who trys this, you might want to look for 4 or 5 A diodes. I have some 3 A diodes, which are quite a bit smaller, however they are only rated for 3 A continuous and might burn up.

The POD spec sheet mentions that at 25 MHz, the POD requires 1.9 A max. At 33 Mhz, it requires 2.8 A max. At 40 MHz, perhaps ~3.7 A max?

I have run GLQuake in autoplay mode for 25 minutes without any crashes. A good experiment would be to see how low of voltage is needed for stability.

I decided to add a resistor in series with the diode to trop the voltage from 4.0 V to something between 3.7 and 4.0. I found that adding a 0.1-ohm, 2W resistor dropped the CPU's internal voltage to 3.72 V (under load) and 3.82 V (DOS idle). Unfortunately, GLQuake crashes after about 2 minutes. A 2-watt, 0.05-ohm resistor in series would probably do the trick for the minimum stable voltage, which I suspect is 3.80 V under load. Unfortunately, 0.1-ohm is my smallest resistor, other than 0-ohm, that is.

For now, I think I will just leave the single 6-ampere diode in place and run the chip at 4 V. It seems stable.

I also have a POD83 which runs without modification at 100 MHz. I wonder if it runs so well because of the CPU dye itself (a good quality fab job), or could it be that the particular VRM on that POD can handle the extra current better?

You could try putting two 0.1 ohm resistors in parallel. That would get you the equivalent of a 4W 0.05 ohm resistor.

I would have tried that but I only have exactly one 0.1-ohm resistor. It was a desoldered components from something long ago. The next lowest resistor I have is two pieces of 0.4-ohms.

If you have enough space, 1x 0.1ohm + 2x 0.4ohm is about 0,067ohm, which might just cut it. You'd have to be very creative with making room for all that, though, so I guess that might not be possible.

I could try that, but instead I might work on a different discrete diode-based voltage regulator design. When using diodes as voltage regulators it is advantageous to run diodes in parallel with your load (the CPU). This is because the forward voltage drop across a diode remains relatively constant even though the current through it can vary by large amounts (refer to IV curve for any generic rectifier diode).

Or I might do nothing at all. Thus far, I have run GLQuake with a single series 6 ampere diode, as shown in the latest photo above, for over 2 hours in Windows NT 4.0. I am using a GF2 card on my socket 3. There are no glitches or abnormalities in the display. The CPU is running stable at 4.0 V. I'll let it run for another hour.

As it stands, this is a very simple mod that most people can do on their own. It is not necessary to remove the POD's heatsink, nor the 4 capacitors, nor the onboard VRM. Anyone else want to see if their POD can run at 100 MHz and 4.0 V? If you only have 1A or 2A diodes, you can run a few in parallel with each other to reduce the current going through each. So if you only have 1 A didoes and want them to handle a max of 4 A, run 4 diodes in parallel. EDIT: However, circuit simulations seem to indicate that running 2 diodes in parallel will increase the voltage to the CPU by 0.13 V, if 4 diodes, then by 0.31 V. I haven't tested this, so probably best to limit the number of parallel diodes if possible.

feipoa wrote:

OK. This is very simple. […]
Show full quote

OK. This is very simple.

1) short the 5V and 3.3V pins on the POD's voltage regulator
2) insulate the fan's 5V pin from making contact with the gold pad (heatsink side), for example, with a piece of electrical tape.
3) run a 5V wire from the PSU to the fan's 5V connector, perhaps by soldering it on

Any reason you removed the voltage regulator IC? Will it work with the VRM in place? Does leaving the VRM in place, with input/output shorted cause much voltage drop? Perhaps it is sufficient to snip these 2 leads.

You will get an extra 1-2 fps from DOS Quake if you run the POD in WB-mode. If you want to view the numeric benchmark result in Quake: Load Quake; select single player mode; once in the game, press this key: ` to bring down the command console; in the command console type: timedemo demo1; press enter; quickly press the ` key again to hide the command console; the benchmark should start running; wait for it to complete to view the fps result.

But if the fan isn't drawing power from the POD anymore, doesn't that kick it down to 1x multiplier automatically? I thought it assumed the fan is not in operation if it couldn't detect it and it detects it through the power connection.

Also, has anyone managed a 66 x 1.5 multiplier run yet?

sliderider, that method was the previous poster's design. In that design, you need to connect the fan's power lead to the 5 V PSU directly. With the new design, you do not need to connect the fan's power lead to the PSU's 5 V lead. But to answer your question directly, the fan has a tacometer, and it seems that the POD drops down to a 1x multiplier if it does not receive a signal from the tacometer pin. So it does not matter where the fan gets its power from as long as the fan is turning, it will run at 2.5x.

How would you run the POD at 1.5x? I thought it was 1x (no fan) or 2.5x (fan)?

Modified a POD using feipoas method with 5 A diode I pulled of a trashed socket 7 mobo ran quake demo for about an hour with no problems. Don't think I will run it anymore till I get a multimeter and test the voltage, I noticed the diode gets quite warm running quake at 100mhz. In my other post I mentioned that the POD runs at 2.5x with or without the fan when on chip circuitry is removed I tried this in several other motherboards It only happens on the V3.3 M919 tried 3 PODs all run at 2.5x without fan and cpu check says that the CPU is not running in real mode when the fan is pulled

When I used a 3 A diode, it felt really hot. The CPU's voltage was about 3.95 V. With my huge 6 A diode, the voltage was 4.0 V, but the diode was just on the warm side. The 6 A diode I am using has about 3x the surface area as your 5 A diode, so it can dissipate heat a lot easier.

I implemented a modificaiton which allowed for 3.80 V to the CPU, however it was not stable. 3.95 V is stable, however.

I suspect most socket 3, Pentium Overdrive CPUs should run at 100 MHz with this simple single-diode implementation. I tried for 125 MHz, but that was asking for too much.

I just performed this modification on another POD83 CPU. This particular CPU couldn't even boot DOS at 100 MHz at stock voltage, however after the mod it ran everything beautifully - Win95, WinNT4, and GLQuake. I did not have another 6 A diode, so I connected two 3 A diodes in parallel which allows for a smaller profile compared to using a single 6 A diode. I was also able to perform the soldering without having to remove the heatsink, which saves a lot of hassle.

Hi would like to briefly with you thank reading for some time here in the forum and love.
Enjoyed my POD83 modded thanks to you and now he is running without problems.
Here are some pictures he has not become beautiful but it runs very stable. Write soon something more there but is very late 😀

Sorry for my English.

I still have a POD is then REV. 2.0 and will look better.

https://youtu.be/BNFbWrGq6p0

https://youtu.be/Lga48-v_azs