Direction of Hovland Musicaps

[UncleRobb]

I have two quartets of bundled 3.3uF/100V Hovland Musicaps waiting to replace the 13uF motor run caps in the A/AA xover networks of my La Scalas. The caps have one lead with clear insulation and the other with a clear insulation with blue stripe. What is the recommended wiring direction?

Thanks, Robert

[3dzapper]

The stripped lead is the inner foil and goes toward the load.

Here is a link:

http://www.hovlandcompany.com/pdf/brochures/MusiCap.pdf

Rick

[UncleRobb]

Thanks Rick. Somehow I overlooked this link. I just put the caps into my xover network and I am burning them in.

Robert

[WMcD]

Gosh this is silly.

It goes with directional wire.

I'm waiting for the day of the directional resistor.

Gil

[JohnA]

LOL!

[levir]

I have successfully upraded my Klipsch KG5.5 2-channel x-over with Hovland caps. I think this upgrade gives more value and performance.

[WMcD]

Just so you know, I'm not taking a stand on high quality caps. At least not here.

However, there is no reason I can comprehend that they are polarized in terms of performance (save, electrolydics). When the sine wave goes negative, how can they tell plus from minus?

Gil

[Al Klappenberger]

Gil,

Gee.. You should know that the inside foil has to be the input side or all the stored electrons will fall out all over the floor! You always got to put stuff inside the bucket, not outside!

Al K.

[UncleRobb]

Gil: Film/foil capacitors are of course non-directional, but, depending on the construction, the two connections can still be different. This is the case for the Hovland Musicaps, which are rolled-up, and the leads to the inner and outer parts of the foil are clearly marked. When these caps are connected in a way that both leads are not equivalent, for example in a CR high-pass filter, then it will make a difference how the leads are connected. Whether the sonic difference is large is another question. When caps of limited quality are used, it might even not possible to hear a difference at all. However, when I spend $100+ on high-quality caps, then I feel that I should wire them in the way the manufacturer recommends.

Al: Very funny.... but what can be expected from somebody who puts caps of different values in parallel into the signal path

Let the flame war begin...

Robert

[Al Klappenberger]

UncleRobb,

That's the story they tell about directionality in caps. The only problem is that it's total BS! The purpose of a capactor is to make both inner and outer foils equal at some level of reactance. If they are equal, then they are EQUAL! Inner and outer foils are coupled together by the very action of being a capacitor, therefore, no difference! The fact that each side of a capacitor in an operating circuit may be at different potential (voltage) is not a factor. If it was, resistors would also have directionality since they too operate with different potential on each side.

Putting two in parallel solves the problem of increasing losses at high frequency. As frequency goes up, the ratio of reactance to internal losses (I.E. "Q" or X/R) goes down. The higher the capacity the thinner the dielectric so the higher the losses up high. Solution: parallel a big one with a small one. The smaller one is called a "bypass cap". That's good science and not BS. It is less important these days becasue of the much lower losses of modern caps. It's become traditional to do it. That's really why I do it since I use only polypropylene caps. An inductor, by the way, works the other way around (almost).

Al K.

How's that for the opening salvo of the war?

[WMcD]

Err, yes I do understand how the "plates" of the capacitor are sort of jelly roll. Granted, it is therefore not symetric mechanically.

My problem is that the sine or musical waveform is symetric around zero volts (save for small d.c. offset in some amps without transformers). If one direction of charge in the electrostatic field is good for the plus half of the input, and bad for the negative half . . . well enough. But then the music / sine reverses, it is bad for the positive input and good for the negative. How can that work?

Polarity or directionality in these caps is nonesense. If it makes people sleep better following the advices of the manufacturer, I'll agree.

I must agree with your explanation Al, at least in theory. We can increase the Q by judicial selection of caps. I'm not sure I agree with you on the merits (in overall performance) of capacitor or inductor Q considering the dominant effect of the voice coil resistance and the resistance created by the horn load.

I believe it is overstated. But not inaccurate. Something to consider in careful engineering, which Al does so well. I have a lot of respect for your work and you might make me a convert. All non believers stand on the cliff of perfect faith.

Thanks Al for your explanation. It made me think more deeply.

None the less. Suppose we need 10 uF. It could be made up by a lossy 9 uF together with a non lossy 1 uF. The result would have a better Q than a 10 uF lossy one. But not by much. It seems we'd be better off with ten 1.0 uF in parallel. I don't see anyone doing that, though.

My recall is that the bypass, small cap, came up in the situation of non polarized (equally polarized) electrolydics of large value. Maybe with relatively large standard caps too. People who used the helper, small caps claimed there was a defect in the performance of large value caps at low voltages or current. To my knowledge, no one ever showed this was true.

This was a bit akin to distortion of push-pull amps at around the zero voltage and glitches right there. In big caps when imposed voltage, or resulting current (they're different in caps), flipped around, there was some more pure performance added by the helper which, in this mythology, patched up the problem.

I'd be the first to get on board if there was any hard data.

I do wonder about electolydics. How fast can a more complicated system get repolarized? If this is a problem, can the helper cap really help?

Best,

Gil

[UncleRobb]

Perhaps we can agree on that the effect of directionality in wound non-electrolytic caps is an effect for which only those audiophiles may apply who virtually can hear the grass grow and the fleas cough. The two leads of a cap in a non-symmetric network experience different potentials, V, and currents, I, as a function of time, i.e. V(t) and I(t). Thus when a non-ideal cap that itself is non-symmetric with respect to an exchange of the leads is used in such a network, its behavior will (slightly) depend on the direction how it is wired.

Non-equivalent capacitors, e.g. a smaller one used to bypass a larger one, have likely different time constants (i.e. T=C*R) which, when they are used in the signal pass, to my understanding, can create some smearing of the audio signal.

What do you guys say now?

Robert

[WMcD]

Uncle Rob,

You're very diplomatic. Sometimes I fail to be. I'll say my mea culpa's.

I must hold my ground.

If the cap had different characteristics when the signal shifts from plus to minus, it would cause more distortion, not less.

It is technically incorrect that there is source of signal or a load from the standpoint of the a.c. signal in this setting. The sine wave is always reversing in polarity. Current is being pushed forward and back. It can't be optimized for one and not the other, or both. This is nutzo.

Stuff like this compelled PWK to make BS buttons. It is nonsense and manufacturer's hype.

Honoring the manufacture's sugggestion does not cost us anything. So do it.

However. I just don't like it. If anyone fails to apply rigorous and critical thinking to small items, then you're on a slippery slope. Logic goes downhill.

Gil

[levir]

Here is a picture of my upgraded x-over. If you follow the network, you can figure out the orientation.

http://www.pbase.com/image/23906747.jpg

[Al Klappenberger]

Guys,

I did some research into the issue of bypassing a big cap with a smaller one of lower losses. The "conventional wisdom" does not seem to hold water! Computer analysis shows that the cap with the larger value gets the current no matter what the other one in parallel happens to be or the frequency. This is in keeping with Ohms's law for AC. The same results with the impedance bridge measurements connecting a 3.3 uF nonpolaraized cap in parallel with a 1 uF polypropylene bypass cap. The total Q was simply the combination of the two at all frequencies! So.. Unless I am missing something, that's more BS!

As to polarity of caps, I am convinced that's BS!

Al K.

[UncleRobb]

Al, what exactly did you simulate? Did you take (DC) resistance, inductivity, and dielectric losses of a (real) capacitor into account? What exactly did you measure and how do you define Q, e.g. AC-resistance/DC-resistance? From what you write, I conclude that you e.g. did not account for dielectric losses. In order to measure the problems introduced by different caps in parallel, you might also want to use some more refined setup.

Robert

[Al Klappenberger]

Rob,

I didn't dig to far into the subject. I simply calculated the equivalent sereis resistance represented by the Q (X/R) of a 39 uF cap witha Q=100 and a 1 uF cap with Q = 1000 at 400 Hz and put them into the computer as two parallel shunt caps. I had the computer calculate the current through both at 400 Hz and 20 KHz. I am quite sure that I have not gotten to the bottom this question. A little voice inside my head is telling me I have missed the point. I just am not sure what to test next! The computer treats them a seperate entities, which is what I expected it to do. There is more to this and I expect it will require more detailed computer models that I have of a true capactor.

Al K.

[Deang]

http://www.capacitors.com/catalogs/multicap_1.html

[WMcD]

Hmm, I'm glad we're moving away from directionality. Any hard date in this area would be welcome.

BTW, I have the highest respect for Al. K. We only met once at the first/last prilgremage to Hope. If we have any debate between us, I think he is more careful and precise than necessary with low Q components, etc. It is damning with high praise, not low. Smile.

My guess is that he put the parallel caps on the GR bridge and found that it matched computer simulations. Is that correct, Al?

Uncle Rob. In all kindness, this may be the debate between guys in the trenches. Do you have some data to contradict Al's findings. I love to see hard data and empirical results. It could convert me. Smile again.

Gil

[Al Klappenberger]

Gil,

The respect is mutual! I often read your posts and have never found anything in them that I thought was bad advice. You also have a very good feal for explaining things. I, on the other had, probably then to confuse some people. My wife is a classroom school teacher and tells me that I am a very poor teacher! I tend to agree.

I started to measure a few nonpolorized 3.3 uF caps with and without a 1 uF bypass cap at 20Khz with the impedance bridge. I didn't bother to do any calculations, but it seemed to be acting just like you suggested. The total Q is just the combination of the two. My gut huntch is the total Q is probably equal to Q = sqrt(q1*q1 + q2*q2). This the formula for the total Q of an L/C "tank". It probably apples to two caps as well though I didn't bother to try to prove it.

Al K.