The Indestructable Netvista - Full poly-mod

[c_hegge]

It may be nearly obsolete, but at least it's indestructable It's a P4 2.66GHz s478.

When I first got it, it had 330uF 25v GSCs as the small caps scattered throughout it (and remarkably, only was was bloated) with some rubycons on the VRM.

I decided that 25v was overkill, since you don't get higher than 12v on a motherboard, but then I thought hey, it I'm gonna use 330uF 16v caps, why not go poly, and that I did.

I used 330uF 16v sanyo SEP polys to replace the GSCs. There was one 680uF 6.3v cap there, but it had a blank spot right next to it (which was in parallel with it), so I just filled up the blank spot.

The CPU VRM high got 470uF 16V Sanyo SEPC and the CPU VRM low got 820uF 2.5v chemi-con PSA.

There were 3 1500uF 6.3v ruby MBZs on the other side of the CPU socket (between it and the DIMM slots). According to my DMM, they were only getting 0.6v, so I decided to use the same Chemi-con PSAs as I did on the CPU VRM low side.

EDIT: there was 1 3300uF 6.3v MBZ near the northbridge. It got replaced with a 1500uF 6.3v Chemi-con PSC



Ahh, I love looking at a motherboard and seeing nothing but polies. Thanks Topcat!!


Back in the case (and yes, that is a pentium 3 stock cooler on the graphics card. I stuck it there with some Arctic silver epoxy after the original fan failed)


Stress-testing away. It's been going strong for about 24 hours as I type this.

[shovenose]

Re: The Indestructable Netvista - Full poly-mod

I find one flaw: get my a hammer and an airplane and we'll see if it's truly indestructible :P

JK, that's some industrial-grade computing you can do now :P
Polies in the PSU too?

[c_hegge]

Re: The Indestructable Netvista - Full poly-mod

You can't poly-mod a PSU. Doing so will send the ripple through the roof, since the ESR will be WAY too low. PSUs that come with polies work because they are designed around polies. I'm not too worried about the caps in the PSU, though. It's a delta and I did check it out. It has a few LTECs and I think the rest were japanese (ruby and NCC), but even LTECs seem to hold up very well in delta units.

[shovenose]

Re: The Indestructable Netvista - Full poly-mod

I'm sure it's possible.
Maybe stick a resistor in series with the capacitor?

[ratdude747]

Re: The Indestructable Netvista - Full poly-mod

I'm sure it's possible.
Maybe stick a resistor in series with the capacitor?

no.

ESR and R cannot be mixmatched like that... Poly-modding is one of the many times "what is good for the goose is good for the gander" does not apply.

[shovenose]

Re: The Indestructable Netvista - Full poly-mod

Here were my thoughts:
Voltage: the max. amount of volts it can handle
Capacitance: How much it can hold
Ripple: how much AC and cap can take before it goes bye-bye
ESR: resistance

Wrong?

[ratdude747]

Re: The Indestructable Netvista - Full poly-mod

Here were my thoughts:
Voltage: the max. amount of volts it can handle
Capacitance: How much it can hold
Ripple: how much AC and cap can take before it goes bye-bye
ESR: resistance

Wrong?

yes.

ESR only occurs in AC conditions. The way it behaves is different than R.

It sounds good on paper but in reality it os a bad idea.

Even if ESR and R were the same, the amount of resistance needed would be so low that getting leaded (non-smd) resistors in that value would be difficult. not to mention that even a 1% tolerance (4 band) resistor could still be hard to come by at those values.

[c_hegge]

Re: The Indestructable Netvista - Full poly-mod

The problem is that a PSU's filtering circuits are tuned to a certain ESR and possibly capacitance, which is where it will work at its best. Any change (up or down) will throw out the 'tuning' and cripple the ripple suppression, especially if you're going from only a low-ish (like Panny FC or FK, or Ruby ZL) ESR to a poly, which will have an even lower ESR than a nichicon HZ. The only exception can be cheap gutless PSUs where the manufacturer didn't bother trying to finetune the ripple suppression.

DC-DC conversion like in a CPU VRM is different, and there is little if any tuning, so lower ESR nearly always helps with the ripple suppression.

I may at some point try experimenting with polies in a cheap-ish PSU when I have my new pet stingray and a lot of spare time and polies.

shovenose was right about everything but the ESR, it's more of a resistance to ripple than to current. Remember, caps do not conduct electricity accross them (other than a very small amount of leakage current, which is usually mentioned in the datasheet).

[shovenose]

Re: The Indestructable Netvista - Full poly-mod

The problem is that a PSU's filtering circuits are tuned to a certain ESR and possibly capacitance, which is where it will work at its best. Any change (up or down) will throw out the 'tuning' and cripple the ripple suppression, especially if you're going from only a low-ish (like Panny FC or FK, or Ruby ZL) ESR to a poly, which will have an even lower ESR than a nichicon HZ. The only exception can be cheap gutless PSUs where the manufacturer didn't bother trying to finetune the ripple suppression.

DC-DC conversion like in a CPU VRM is different, and there is little if any tuning, so lower ESR nearly always helps with the ripple suppression.

I may at some point try experimenting with polies in a cheap-ish PSU when I have my new pet stingray and a lot of spare time and polies.

shovenose was right about everything but the ESR, it's more of a resistance to ripple than to current. Remember, caps do not conduct electricity accross them (other than a very small amount of leakage current, which is usually mentioned in the datasheet).

it all makes sense now

[Th3_uN1Qu3]

Re: The Indestructable Netvista - Full poly-mod

The problem is that a PSU's filtering circuits are tuned to a certain ESR and possibly capacitance, which is where it will work at its best. Any change (up or down) will throw out the 'tuning' and cripple the ripple suppression

Hasn't this been beaten to death already? There is nothing inherently tuned about the output LC. On the other hand, the compensation of the error amplifier in the PWM controller is tuned around the filter, and drastically altering the ESR could cause it to oscillate.

[c_hegge]

Re: The Indestructable Netvista - Full poly-mod

Thanks for the correction.

btw, I only found 1 post here where you actually meantioned that (which I had largely forgotten about), so it hasn't exactly been beaten to death.

[kc8adu]

Re: The Indestructable Netvista - Full poly-mod

i have polymodded psu's.
all they did was get cleaner output.was it enough to be worth it?not unless you have the parts coming out your ears like i do.
btw if there is to be another app that screams for polys besides dell sff this is it!these can cook mbz!

[severach]

Re: The Indestructable Netvista - Full poly-mod

ESL is what shows up only with AC current. ESR is resistance just like R. The reason it takes AC to measure ESR is that there's a capacitor in the way which blocks the DC coming from a standard resistance meter.

Testing at 100Khz adds some resistance from ESL and the electrolyte.

http://en.wikipedia.org/wiki/Equival...ies_resistance