Power supply filter caps?

[savage]

I just came across a bunch of high voltage metallized polypropolene caps, ranging from 30uF to 7uF 400vdc. Any reason I wouldn't want to use them as power supply filter caps?

[R ski]

I never seen values that large, however, If they are that value, I am sure they will work. Less leakage, than electrolytics, they would work fine.

[J Martin]

They make much larger ones than that. I use the 630 volt Solen caps. They get huge as the values go up. Much better than electrolytics unless you like the slower response which is just a part of some traditional amp designs. Try them with different circuits to find were they sound good to you.

[R.G.]

I've been preaching about using metalized polypro caps for tube amp filters for a long time, years in fact. It's part of my "Immortal Amplifier" idea.

The only real downside to them are cost, size and weight. They are otherwise more capable than electros and will literally last forever if you don't manage to burn them up or physically destroy them.

A simple and cheap(er) way to get polypro caps in suitable values is to buy the oval metal case motor-run caps at an air conditioning supplier store. They come in up to 50uF and up to 630Vac. 37Vac is a standard value, and that's good for 533Vdc. These caps are intended for high current use, something that not all signal caps are.

If you have a cheap source for polypro film caps and can stuff them into your amp chassis, go for it.

If you get nay-sayers telling you that they won't sound good, tell them that you can 'dirty down' the polypros with external resistance and inductance to be as bad as any electrolytic cap.

[bob p]

there's one thing about the use of polpro caps in the PS that has always confused me -- those caps are non-polar, aren't they? to perform PS filtration, don't you have to have a polarized cap so that AC gets blocked and a DC potential of the appropriate polarity is established? i tend to think of nonpolar polypro caps as being DC blockers that are great for passing AC (great for coupling caps).

Maybe I'm suffering from a misconception, so if anyone could help to clear up the confusion, I'd appreciate it.

[Matt T.]

bob,
They're blocking DC and allowing AC (down to pretty low frequencies) to pass...just like any cap. The fact that they are polar or non-polar doesn't have any effect on this (other than maybe ESR or other different properties of that cap type). The direction of the rectifier determines the 'polarity' of the voltage...that's why the bias filter caps and bias rectifier are connected 'backwards'.

[bob p]

thanks Matt. yes, I misspoke. i understand that the polarity of the diodes makes all the difference in establishing whether the voltages delivered to the caps are positive or negative. i guess that at this wee hour i just can't explain what the point is in using polarized caps in a power supply (and making a point of orienting them properly) if you can just go ahead and use non-polarized caps instead.

[Matt T.]

Size, weight, and cost were the original reasons and I think people tend to stick with what works...especially when it's smaller, lighter, and more cost effective (short term anyway). These days you can get non-polarized caps at much larger values in much smaller packages and perhaps even a good price (I've never looked into it but I always wondered about those run caps for motors that RG mentioned)...so it's an option that for all practical purposes didn't exist way back when.

[bob p]

yes, today's electrolytic caps are a lot smaller than they used to be. the old paper in oil caps had a low capacitance per unit volume, and the caps were HUGE. today's aluminum oxide / alkaline electrolyte caps have a much higher capacitance per unit volume and are much, much smaller.

one thing that i wonder about when proposing to replace electrolytics with the poly / film caps is the frequency response. electrolytic caps don't function at high frequencies, while the poly caps have extended/relatively unlimited high frequency response. doesn't the diminished high frequency response of the 'lytic caps make them superior in some respects for PS filtration?

essentially, when we're talking about a PSU we want the caps in the filter network to act as an EXTREME low pass filter where all of the "high" frequencies are completely attenuated and we're left with nothing but pure ripple-free DC. isn't the inherent ability of the filter to attenuate high frequences limited somewhat when you replace electrolytic filter caps that have impaired high frequency response with poly caps that have extended high frequency response?

i'm wondering if when we propose to use poly caps in the PSU in the name of enhanced reliability, we're actually trading off one demon for another. i'm hoping that somebody who has a better handle on the theory of this stuff than i do can chime in on this.

[Rob Mercure]

Bob,

No, actually it's the opposite. Remember the filters work in two ways - first they serve as a energy storage unit which tends to "fill in the holes" between pulses of rectified AC. But they also function as a frequency dependent "bypass," "shunt" or, if you will, "short" from the + side of the PS to ground (at least in the most common tube amp). Thus any alternating current in the PS is "shorted out" to ground - this why we have the troubles with "motorboating" from dried up filters - the signals from one stage are coupling to another stage in phase - nothing is "shorting them out." So the increasing "impedance" of common aluminum electro to higher frequencies is a problem cuz you aren't "shorting" them effectively and you wind up keeping much more funk in the supply rail. It's common practice in HiFi design - and I do it (when I can remember to <grin>) to bypass the electrolytic filters with a "small" (0.01uf) film or other capacitor to ensure that the higher frequencies are effectively bypassed to ground - and it's been noted again and again on this forum how much RF "hash" is generated in the PS's silicon rectifiers.

I think sometimes we get the "coupling" role of capacitors mixed up with the "bypass" role and that creates some confusion.

Hope this helps

Rob

[bob p]

No, actually it's the opposite.
...
I think sometimes we get the "coupling" role of capacitors mixed up with the "bypass" role and that creates some confusion.

Thanks, Rob. that does help to clear things up for me.

[R.G.]

Rob is, as usual, correct.

Polypros are a more nearly perfect capacitor than an equal-capacitance electro, in that they have the same capacitance, but:

- lower Equivalent Series Inductance, which ruins a filter's effectiveness at high frequencies
- lower Equivalent Series Resistance, which ruins a filter cap's effectiveness at all frequencies to a certain extent and contributes to internal heating of the cap
- lower leakage, which looks like a paralleled high value resistor leaking DC around the cap.

Hifi and tweako stereo nuts were fond of using polypros to "bypass" electro caps to help clean them up a bit.

This is what I alluded to in my earlier post. If the amp should sound too different with only electro caps changed for polypros, you can insert real resistors and inductors to approximate an electrolytic by "dumbing down" the polypros.

But you can't insert negative resistors or negative inductors with electrolytics to make them look as good as polypros. Mother Nature doesn't let us do that.


As for the polarized/non-polarized issue, think of it this way: non-polarized caps can block DC from either direction. Polarized caps can only block it from one direction. So the polarized caps are only half the cap a non-polarized cap is.

Seriously, if you could get non-polarized caps of a similar size, weight, and price to electrolytics, electros would vanish overnight. It's only because electros are cheaper and smaller per capacitance-volume product than nonpolars that electros exist at all.

[Arthur B.]

Rob is, as usual, correct.
But you can't insert negative resistors or negative inductors with electrolytics to make them look as good as polypros. Mother Nature doesn't let us do that.

Actually, there is a way of creating Negative Resistance.

[R.G.]

Actually, there is a way of creating Negative Resistance.

Well, more correctly, there are multiple ways of emulating a genuine negative resistance. What the wiki shows is one of those ways, there are others. That negative resistance circuit is a negative impedance converter set up to synthesize a negative resistance. And the article mentions the caveats
- "If the operational amplifier is ideal": there aren't any of these;
- "it is possible to let a real generator behave (almost) like an ideal generator": the "almost" is important;

Notice that for the task of adding a negative resistance to a power supply filter capacitor, you would need an "opamp" capable of supplying tens if not hundreds of amperes of current in pulses to the real capacitor which you're trying to change by adding a negative resistance. The resulting amplifier's needs are significantly more complex than the circuit you're supplying with power - and more importantly, the amplifier you'd use to do the emulation need its own power supply to be able to do the emulation.

That being said, there are genuine negative resistors; at least, there are two-terminal devices which act like negative resistors over a small portion of their operating range. Some of the stranger variants of diodes do this. But they are not capable of being soldered in series with a capacitor to make it look more perfect, as I was referring to.

[Amp Kat]

I've seen those motor caps at Granger before and often wondered if they would work in amplifier applications but was under the impression they would't filter DC but now I can experiment. I've also seen them in some power supplies where they take the motor caps and run them on AC off of a winding going to no output supply then the other windings or used for an 100vac telephone ring voltage and +/- 24 & 48vdc. I would think the motor caps are to smooth the AC voltage out as it's about 600 volts at that cap. I have some 100uf and down to 22uf Solen fast caps and they are super large especially the 100uf 630 and have though of mounting them on the side of the cabinet or even building a whole seperate power supply like Mr. E.J. does. I have noticed and can certainly relate to Hi-Fi applications for these caps as they really don't seem to have the grind E-caps do but as RG says I haven't done the things to make them grind so I'm sure theres other ways to add to it. The London Power Studio has a load of Solens in them and this amp is very versatile with Power Scaling and several tone changes but it's drawback is exactly what RG is saying everyone else is saying about it being slightly lifeless but after hearing replies from this Forum Kevin made some serious changes to the Studio of course with a price increase to justify it but I'd like to know if he did some of the things RG has mentioned because it seems IMO that it needs it to give them a more grainier tone. I sure would like to try it because the idea of never changing filter caps in your own personal amp is a sweet thing not to worry about.