question about Polymodding

[elbrute]

A question about Polymodding. I understand the rule of thumb has been 'half the uF value of the lytic...or thereabouts' when replacing a cap, but do you HAVE to reduce the uF of all the caps your replacing, or is it just certain circuits (VRM ect...). Prolly been asked and answered before, but I haven't found it. Any help would be nice.

[stj]

Re: question about Polymodding

no, the dropping of the value was just because until recently polymers didnt come in very high values.
now i think the largest is 2700uf - i could be wrong.

[Hitto]

Re: question about Polymodding

I agree. Recently polymers are also cheaper and (I hope so) more robust.

I think that as long as the E.S.R. of a polymer is not a problem for the circuit you can preserve the original capacitance. Doing so you can combine the benefits of polymers without a single trade-off.

[elbrute]

Re: question about Polymodding

Good to know. Thanks for the replies.

[halaster79]

Re: question about Polymodding

is it true that as long as the ESR is equal or lower than the ESR of the cap you are replacing, the motherboard should work properly? (obviously with a matching voltage and uf)

[ChaosLegionnaire]

Re: question about Polymodding

no it depends on the vrm design. if it uses a high switching frequency, esr matters more so u must use the same esr or lower. u can get away with a slightly lower capacitance in that case.

if its a low switching frequency, capacitance matters more so u must use the same capacitance or higher. u can get away with a slightly higher esr.

[stj]

Re: question about Polymodding

so in short, try to use the same capacitance!
esr will be lower than an electrolytic anyway.

[Uranium-235]

Re: question about Polymodding

no it depends on the vrm design. if it uses a high switching frequency, esr matters more so u must use the same esr or lower. u can get away with a slightly lower capacitance in that case.

if its a low switching frequency, capacitance matters more so u must use the same capacitance or higher. u can get away with a slightly higher esr.

How would one know if it used either one?

[stj]

Re: question about Polymodding

number of turns on the inductors, or just get the datasheet for the control chip

[momaka]

Re: question about Polymodding

How would one know if it used either one?

number of turns on the inductors, or just get the datasheet for the control chip

There is no precise guide to this really. It's mostly a rough ballpark.

Generally, this "high" frequency vs. "low" frequency VRM thing is mainly meant to rule out much older motherboards. For example, many old ECS motherboard (but also a few other cheapo ones) used a KA7500 or DBL494 or TL494 PWM chip - i.e. the same ancient PWM chip that's been used in ATX power supplies for decades. These PWM chip doesn't really go much above 50 KHz switching frequency, if even that, and the motherboards are 2-phases max. The ECS K7S5A, for example, has 4 coils... but it's only a 2-phase design due to the fact that KA7500/TL494 having only 2 outputs. Either way, motherboards based on these ICs tend to have either bigger inductors or just more larger capacity capacitors... and typically it's the latter. Because of the relatively low switching frequency, ultra-low ESR caps are not needed, but the capacity is important due to the slow frequency at which the output caps are "refilled" from the coils. In other words, it's like having bigger batteries in a device, but charging the device less often.

That's why on many old motherboards with KA7500/TL494/DBL494, you see at least six 2200-3300 uF caps filtering the CPU VRM. The ESR doesn't matter so much, but the capacitance does. And that's why they got away just fine with using barely entry-level low ESR crap caps (like G-Luxon LZ and similar.)

This brings another point: number of VRM phases.
Generally, the more phases a VRM has, the less capacitance is needed on the output. That's because each phase "refills" the output caps when the others aren't. So more phases = lower ripple current per given output capacity. Thus, on some older motherboards with only a single phase VRM (think Pentium 3 era), it's good to keep the capacity similar to the original (or higher, if possible.) As to whether very low ESR caps are needed... that brings us back to switching frequency: inductors/coils with less than 1-2 uH tend to be "higher" frequency IME. Meanwhile, anything with 4-5 turns or more on a big inductor core is typically medium frequency, more or less. These, and the high switching frequency boards will do very well with ultra-low ESR caps (or polymers).

[Uranium-235]

Re: question about Polymodding

wouldn't the reason why older boards are different because they don't use ATX 12V? Didn't older boards use the 3.3 or 5V rail for CPU power? A lot of those higher uf caps are 6.3v. Vs the 16v input -> Mofset & coil -> output cap 2.5poly or 6.3v lytic from ATX 12v

just a thought

[stj]

Re: question about Polymodding

input cap must handle the input, but the output caps only need to handle the cpu voltage - usually 1.2v or less.

i can say this, the vrm is irelevent if you keep the uf the same and dont try to lower it.