You should be able to adjust the positioning of your diode such that you can utilise the original heatsink. This would really clean up the unit's profile.

Looks like this mod has made its way to eBay sales. www.ebay.com/itm/152517809130
The description is a bit sketchy though - superglued the heatsink, missing fan pin, and not confident in strength of solder job.

You have an idea what one could make so that one can operate the pod83 without original cooler has already tried to operate another fan on it but the tacho signal takes the pod83 not at the intel fan seems to make something different.

This is interesting indeed!
I've tried to run my P83 at 100 MHz once, but couldn't get it stable. I didn't think that the VRM would be the limiting factor, rather than the CPU itself.
I guess an overclocked P83 is by far the fastest option for a socket 3 platform, certainly for Quake 😀

That's mine

https://www.youtube.com/watch?v=Lga48-v_azs

Scali wrote:

This is interesting indeed!
I've tried to run my P83 at 100 MHz once, but couldn't get it stable. I didn't think that the VRM would be the limiting factor, rather than the CPU itself.
I guess an overclocked P83 is by far the fastest option for a socket 3 platform, certainly for Quake 😀

From my three POD83 units, one would work at 100 MHz and standard voltage. There are a few other success stories with 100 MHz and standard voltage (3.3 V), so it seemed natural that a few extra 100 mV would bring some of the others into stability range. Thus far, I haven't seen any POD83's which didn't work at 4 V and 100 MHz. I'm sure some will turn up at some point.

Zock wrote:

That's mine

https://www.youtube.com/watch?v=Lga48-v_azs

What motherboard were you using? 31 fps is pretty decent. I think the most I've recorded was 36 fps with a Geforce 2MX PCI and a POD100. It was very driver-dependant to achieve that high score. I don't recall what Quake score the Voodoo2 and Voodoo3 received on the same motherboard.

Why is it necessary to power the fan externally with this mod?

Chadti99 wrote on 2020-07-29, 21:49:

Why is it necessary to power the fan externally with this mod?

It isn't.

feipoa wrote on 2020-07-29, 21:55:

Chadti99 wrote on 2020-07-29, 21:49:

Why is it necessary to power the fan externally with this mod?

It isn't.

Got it, thanks! One more question, should I set my motherboard voltage to 3, 4, or 5 volt for this mod ?

5 V.

Based on your questions, I'm guessing that you didn't entirely read the thread.

feipoa wrote on 2020-07-29, 22:52:

5 V.

Based on your questions, I'm guessing that you didn't entirely read the thread.

To the contrary I’ve re-read this thread numerous times. Quite a bit over my head and some folks trying some different things, it’s not quite clear to me. I appreciate your responses and thank you again.

I see. You'd need to pick one person's design or the others, not both at once. I suppose if you have no electronics background, it might be confusing. This thread assumes you have some pre-elementary level of knowledge concerning voltage, resistors, current, etc.

When I read the following:

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)... 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.

it seemed pretty clear I am using the 5 V setting on the motherboard as there would be no other possible means to achieve 5 V using any other lower voltage jumper setting, that is, unless someone went really nuts and had a step-up transformer of sorts on the POD, which I've never seen on any CPU upgrade.

sliderider wrote on 2015-11-16, 03:37:

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

You would need to decap the pods die but my guess is there is no circuitry.

If all pods had the same die with the Future POD DX4 overdrive In mind it’s possible 4x was hidden in there (albeit completely useless)

Average diode voltage drop is .6 to .7 with current about 20% margin under upper current limit for reliability and commonly not done this way except current flow through was below 100mA or less due to cost. If you are seeing greater than that, the diode will break down quickly, for reliability you need 10A, even 20A for this. Not practical and incorrect way, and impacts the voltage with diode junction noise. That not right way either.

Yet, all voltage input of any electronics to generate reference voltage *requires* voltage stablized and this is done with small value capacitor or capacitor & inductor.
Use the GND pin of the regulator through the diode to GND and parallel the diode with small value of capacitor to stop oscillation and voltage noise to minimum. This is correct electronics design when designed linear regulator circuits. Also people usually use a IC or a small low current circuit driving a small transistor in turn to drive the regulator to set up a voltage output as well. Another is use a choice of grounded zener diode using 5V through a resistor (around 50mA through zener diode) and use the zener-ed voltage to drive the lifted GND pin, again parallel with small capacitor to stabilize the voltage against noise.

Otherwise replace that non-standard regulator on this 83 and use a adjustable linear regulator with this circuit I described above. Even some consumer electronics used high current transistor driven this way as well to get different voltage. This yet but not ideal if current is more than 500mA. Adjustable linear regulator is made for this greater than 1A design is correct way.

This was commonly done in consumer electronics back in the day and that was done in very early days of electronics before adjustable high current linear regulators become common.

Cheers,

If you read back through the post, you'll notice that I tried using the Pentium's VRM with a trim pot, but I don't think it is a LDO regulator because the max voltage I could get out of it was 3.73 V, which wasn't enough to make the chip stable at 100 MHz.

These hobby toys aren't really put in use 24/7. They are turned on for a few hours, then stuffed away for a year, then brought out. Thus I feel using a 6 A diode for this purpose offers the simplest solution without needing to remove the Pentium's heatsink, or having cables running around the chip. If someone is already with removing the heatsink, then just as well remove the Pentium's VRM and replace it with a low drop-out (LDO) regulator and set it for 4.1 V. If you don't want to remove the heatsink, yes you can still have a variable VRM with some components on dangling wires.

Alternately, you can just cut the Pentium's lead to the VRM and connect it straight to the 4 V output on the motherboard's VRM, as another user has done.

I measured the voltage drop across this 6 A diode today and it measures 0.81 V at full load after a 20 minute uptime. The 0.9 V mentioned previously was just an inferance. My Sedra/Smith text mentions 0.6-0.8 V drop is typical. The variance is due to diodes with different current ratings; the higher the current rating, the more current required to get into the 0.6-0.8 V range. This diodes reaches 60 C at full load. According to the datasheet, it can have a max current of 5.4 A at 60 C, or 6 A at 54 C, etc. Under these conditions, I am not concerned about the diode burning up.

I have seen this technique in other areas, most notably on memory modules which use 3.3 V IC's. But to make them 5 V compatible, they use two diodes in series to drop that 5 V down to 3.3 V. These modules were produced in the late 90's and early 2000's.

For your design ideas which do not require replacing the Pentium's VRM or have a dangling VRM, could you please sketch out your circuit? I have several 500 mW zeners ranging from 2.4 V to 18 V. If another design can be realised using discrete components like diodes, resistors, and through-hole capacitors, it might be simple enough for others to implement without needing to remove the heatsink. Thank you for your suggestions!

EDIT: I went looking through my zeners. The two which sort of fit the bill for the reverse breakdown are 3.9 V and 4.3 V. I could hook one of these up in the standard poor man's regulator configuration, e.g. 5V in series with 1K in series with the zener, to GND. Then the CPU's load is placed in parallel with the Zener in this ckt.

I went digging through my old notes and it looks as if I tried something similar already. But instead of using a reverse biased zener, I used several regular diodes in series. It has been way too long and I don't recall the outcome of this experiment except that it seems like I needed a really low resistor value (0.2-ohm?) for this to work. The mess is attached.

Any specific diode types you'd recommend? My hardware dealer of choice has 50 different types (5A/6A) on offer.
Maybe it's worth checking out if there are any suitable TO-220 packaged diodes to pretty up the mod a bit?

Yes, perhaps if it fits nicely. I just went with whatever I had on hand at the time which could handle the high continuous current.

So i.e. 2x 1N5402 would be suitable? Those are the only ones I have lying around without having to order anything.

Yea should be OK. I've done this mode with two parallel 3 A diodes as well. If the previous poster scared you, you can always use 3 or 4 in parallel to split up the current. I have had no issue with my single 6A diode though.