Speaker voicing and capacitor types used

[DaveD]

I have been told by two members of the AudioAsylum Vintage Asylum that classic speakers- which were voiced by the designer with eletrolytic capacitors- will sound brighter and less well balanced if the old capacitors are replaced with metallized polypropolene types of equal value. So, high quality modern electrolytics sould be used. They did not render an opinion on film and foil capacitors or the Danish Jensen mineral oil impregnated film or the Jensen metallized paper capacitors.

I would appreciate any experienced comments on this issue, and recommendations for capacitor type to use.

The speakers in question are KLH 17, but the question would apply to all classic designs with electrolytic capacitors. My goal, aside from protecting the tweeters from damage, is to make the sound as pure as it can be without changing the voicing, or tonal balance, of the speakers.

[rrcrain]

Several of us (perhaps more than several?) have successfully replaced old electrolytics with the new metalized poly caps and have been more than sightly impressed with the improvements. I've no experience with KLH, but it's worked extremely well with the AR90s I redid.

[kkantor]

This is 100% correct. If you want to *preserve* the original sound of the speakers, don't change the cap type, other than to repair obviously broken parts.

If you change the cap type, you will change the crossover response and alter the speakers voicing and balance.

Of course, you might like the new sound better than the original designer's choice. But, if you want to go that route, there are some very cool, inexpensive digital equalizers on the market now that will let you play with great precision, repeatability and reversiblity.

[Guest dogmeninreno]

>This is 100% correct. If you want to *preserve* the original

>sound of the speakers, don't change the cap type, other than

>to repair obviously broken parts.

>

>If you change the cap type, you will change the crossover

>response and alter the speakers voicing and balance.

>

>Of course, you might like the new sound better than the

>original designer's choice. But, if you want to go that

>route, there are some very cool, inexpensive digital

>equalizers on the market now that will let you play with great

>precision, repeatability and reversiblity.

Ken, What equalizer would fit a set of AR9's being bi-amped by 2 GFA555II's vertically? Looking at the woofer sections in perticular. I havn't had an equalizer for 30 years. Thanks, Dale

[tysontom]

>This is 100% correct. If you want to *preserve* the original

>sound of the speakers, don't change the cap type, other than

>to repair obviously broken parts.

>

>If you change the cap type, you will change the crossover

>response and alter the speakers voicing and balance.

>

>Of course, you might like the new sound better than the

>original designer's choice. But, if you want to go that

>route, there are some very cool, inexpensive digital

>equalizers on the market now that will let you play with great

>precision, repeatability and reversiblity.

Thank you Ken for this important and prophetic message! Anyone contemplating a change in crossover components -- other than repair -- should in my opinion read and understand Ken's message.

I think that it is also true that if you feel that you can improve upon an original design, it's your prerogative, and you should not hesitate to do it. Subject testing is vital, of course, to determine if changes have made an improvement. Keep in mind, however, the only way to know for sure that these “improvements” are real or imagined is to accurately and *objectively* measure the response of the finished product. And, unfortunately, the only way to properly do this is with relatively sophisticated test and measurement equipment, and the appropriate knowledge and training to perform these tests, and the ability to interpret the results when finished, and before finishing to be certain that measuring equipment meets measurement-standard calibration, and that the calibration is traceable to US Bureau of Standards, and so forth. No easy task. “A beautiful hypothesis slain by an ugly fact,” as they used to say.

--Tom Tyson

[johnieo]

I sincerely appreciated reading the valuable comments "voiced" by Ken and Tom. Most likely, I have committed an unpardonable "acoustical" sin by replacing my leaky NPE and impregnated paper caps with thin-film caps. However, is there one unique voicing for an AR-3a? One of my AR-3a's was manufactured with impregnated paper (150/50/6 uF) capacitors, whereas the other was, a decade later, factory converted using NPEs; the inductors are the same. Was the voicing of the two speakers identical when each was new? Should I have used new impregnated paper in one and new NPE in the other? Both with NPE? Both with paper? I honestly had no idea and that was part of my dilemma in choosing replacement caps.

Since NPE electrolytics shelf age (oxide thickness changes) and reform after sitting idle, there comes the issue of the voltage and current conditions under which capacitor break-in (reforming) is performed--conditions that must affect capacitance and crossover. If one breaks in new NPE caps under "heavy metal" conditions, will one have distinct capacitance values (and voicing) from one that is broken in on Glen Gould playing The Well Tempered Clavier?

If one consistently enjoys Wagner, Saint Saens, and Beethoven at full orchestral power levels will one's (NPE) speakers be voiced the same as those owned by a person who enjoys, before bedtime, a diet rich in Dietrich Fischer-Dieskau quietly singing Lieder? What would they sound like with impreg paper caps?

I am delighted with this forum, its info, and experienced comments, as well as the help given off-forum by several folk. Like the onion, when I peel one layer ... Cheers,

[Guest Bret]

>If you change the cap type, you will change the crossover response and alter the speakers voicing and balance.<

In what way?

6uF isn't always 6uF?

Bret

[johnieo]

A well written article by Jung and Marsh (1980) describes capacitor characteristics important to audio amplifiers (and presumably to crossovers). It discusses Dielectric Absorption (DA)--also known as recovery voltage -- and other capacitor properties. The authors state that DA is the phenomenon responsible for observed differences in musical detail, dulling, sharpness. They show that DA varies greatly with capacitor type. Likely this is an important metric in describing sound quality and why one should not change the capacitor type (assuming the mfr used only one type!).

http://www.capacitors.com/picking_capacitors/pickcap.htm

[tysontom]

>I sincerely appreciated reading the valuable comments

>"voiced" by Ken and Tom. Most likely, I have committed an

>unpardonable "acoustical" sin by replacing my leaky NPE and

>impregnated paper caps with thin-film caps. However, is there

>one unique voicing for an AR-3a? One of my AR-3a's was

>manufactured with impregnated paper (150/50/6 uF) capacitors,

>whereas the other was, a decade later, factory converted using

>NPEs; the inductors are the same. Was the voicing of the two

>speakers identical when each was new? Should I have used new

>impregnated paper in one and new NPE in the other? Both with

>NPE? Both with paper? I honestly had no idea and that was part

>of my dilemma in choosing replacement caps.

>

>Since NPE electrolytics shelf age (oxide thickness changes)

>and reform after sitting idle, there comes the issue of the

>voltage and current conditions under which capacitor break-in

>(reforming) is performed--conditions that must affect

>capacitance and crossover. If one breaks in new NPE caps under

>"heavy metal" conditions, will one have distinct capacitance

>values (and voicing) from one that is broken in on Glen Gould

>playing The Well Tempered Clavier?

>

>If one consistently enjoys Wagner, Saint Saens, and Beethoven

>at full orchestral power levels will one's (NPE) speakers be

>voiced the same as those owned by a person who enjoys, before

>bedtime, a diet rich in Dietrich Fischer-Dieskau quietly

>singing Lieder? What would they sound like with impreg paper

>caps?

>

>I am delighted with this forum, its info, and experienced

>comments, as well as the help given off-forum by several folk.

> Like the onion, when I peel one layer ... Cheers,

John,

Thank you very much for your enlightened and knowlegeable approach to this forum. Your input is valuable to this forum, and your scientific knowledge is important to everyone here.

You raise an interesting point regarding the capacitor changes in the AR-3a from the early (1967-1972) to the later versions (1973-1975). The first-generation AR-3as did indeed use the Industrial Condenser Corp paper capacitors up to about 1972; the later-generation AR-3as used Callins or Sprague Compulytic electrolytic capacitors (50 to 60 V). The second generation AR-3a (not including, in this instance, the AR-3a "Improved" version which was a European model) was indeed changed in several important respects: (1) the crossover frequencies were stated as 525 and 5000 (vs. 575 and 5000 on the first series); (2) the speakers began using a back-wired driver-mounting technique (vs. the original "hard-wired" front-wired in the early versions); (3) a change in the grill-cloth material and mounting technique and (4) a different-type label on the back that dispensed with the warranty card stapled to the back of the speaker, etc. I always felt there was a very subtle difference in the two versions, but I was never sure, and I did not have the opportunity to compare the two side-by-side -- something you are actually able to do. Retrofitted AR-3s-to-AR-3as would reflect the specific capacitor types, depending on the time period in which they were received at AR.

The important take-away here is the crossover difference, and for a long time I doubted this, but SteveF, a higly knowlegeable AR historian on this forum, has several letters from AR clearly stating the change in the crossover frequency of both the AR-3a and the AR-5 that occurred in the second-generation AR-3as. In the case of your AR-3as, perhaps the best plan would be to chose one type or the other, and make both speakers the same if, and only if, you feel there is a difference in the sound of the two as they now exist.

--Tom Tyson

[Guest Bret]

Thank you very much for pointing me to that page.

That link gives me enough jargon to be able to ask my question better, which hopefully allows me a better understanding of the issues involved.

Bret

[ninohernes]

Tone wise. Paper in oil caps sound different from poly caps. Paper in oil caps tend to sound warmer and more natural than poly caps.

[soundminded]

It depends on what you mean by "voicing." If you are talking about the relative loudness of one driver versus another at a particular frequency, the type of capacitor should not make any difference. The crossover frequencies and attenuation should remain the same. However, the term capacitor is a mathematically conceptual term while the physical devices we call capacitors would have to be correctly modeled with many parasitic elements both linear and non linear which of course will vary from one design to another. The sound of a loudspeaker system does not only depend on steady state frequency response but other factors as well such as transient or impulse response. In this case, some types of capacitors may be a better or at least a different sounding match to loudspeaker drivers than others. In this case, the modified version may not sound exactly like the original. Physical testing as well as double blind AB testing of the type AR once did is likely only performed by the best and most well equipped speaker designers today such as those at the top of the Harman Industries empire. So for the experimenter, it's hit or miss.

One thing examination of the AR9 crossover circuit and Tim Holl's excellent explanation revealed is that when you change any of the values, you are definitely playing Russian Roulette with the design. Also, if you change the corssover frequencies or attenuation rates partcularly of the midrange and tweeters, you risk damaging them due to overexcursion at frequencies they are not intended to handle.

[Guest Bret]

>In this case, some types of capacitors may be a better or at least a different sounding match to loudspeaker drivers than others.<

I think this is exactly the case made on Northcreek's site. (that's an old AR-guy, too, isn't it?)

As an issue, I think we have a very un-dead horse here. We keep leading it to water and it keeps refusing to drink. If you'll go to page 27 of this forum and look at the very earliest postings available - there it is as an issue.

We all know that we can't test this subjectively and make any "carved in stone" pronouncements. What we need is a restored speaker that has been tested for changes in it's original measurements. If the difference isn't measurable, but we can hear it. . . well, I'll just drop that.

It seems to me that when a designer is sitting at the drawing board designing a crossover that he almost surely isn't getting specs for DA and DF of every capacitor into the calculations. Presumably there are better and worse numbers and everything else remaining equal nobody would purposefully spec a capacitor that they knew was going to muddy-up the sound.

Galileo refusing to use better glass because of his original design intent? Less scattered light improves, not the design, but the performance - while staying quite true to the design.

Using "worse" capacitors as a matter of design doesn't even make sense to me.

I can understand not changing values, or components with values that change other things (like the impedance of a choke) would be changing the intention of the designer. But using capacitors known to "ring" or "sludge" the signal passing through them on purpose?

The only place this gets "iffy" is if in final listening tests limitations of the design, or components, were compensated for. (Muddy high-end? Roll it off! Etc)

Bret

[soundminded]

It is not always possible to say which is better. Things usually aren't that simple. Often in life there are tradeoffs. As one parameter gets better, one or more others get worse. Sooner or later you have to make a choice.

"What we need is a restored speaker that has been tested for changes in it's original measurements."

Would a 40 year old AR3 hermetically sealed in a box and unopened after all these years sound the way it would have had it been opened 40 years ago? Would some or all of the components inside change audibly? Probably but nobody knows for sure. What about the ones in the real world, do they sound different than they once did? There's probably no way to find out short of repeating the exact manufacturing process from scratch and duplicating every last step along the way. That's not going to happen.

[Guest Bret]

>Would a 40 year old AR3 hermetically sealed in a box and unopened after all these years sound the way it would have had it been opened 40 years ago?<

That really isn't what I meant.

Instead, a laboratory comparison of current production capacitors in a classic system might yield some interesting results and would probably come very close to producing "an answer" complete with a quantifiable explanation.

I have no idea what would be involved.

Likewise, if I could ever learn enough to be able to do it (doubtful, since I'm feeling older every day), it might be handy to calculate the adjustments needed in the crossovers to compensate for the differences in the currently available replacement drivers.

As far as the capacitors go, I hate the thought of experimenting because my preference tells me nothing about the correctness of my choices. (although I admit that Rich's experience does skew my tendencies)

Okay, and in the interest of full-disclosure, it sounds like a lot of time, money, and work to produce a result I can't defend as "correct." I’d always wonder, “Did I screw-up my speakers?”

Bret

[johnieo]

I’d always wonder, “Did I screw-up my

>speakers?”

Right on! My current fear!

Regarding two important capacitor parameters (dissipation factor and dielectric absorbtion) Jung and Marsh say: "The effects of DF and DA can be perceived differently. DF is primarily a contributor to phase and amplitude modulation and DA reduces or compresses dynamic range."

Paper and NPE caps both degrade seriously after 30-40 years. I did a quick DA check on my obsolete 150 uF paper and NPE units and found DA to be 40% and 15%, respectively. New caps of these same types would have much lower values. Poly films caps would have the lowest values. When comparing the sound of various cap types, we might qualify our results "NPEs sound ... than polyfilm." Are we comparing polyfilms to new or obsolete NPEs? Likely there would also be some change in sound quality when old caps are replaced with the same type.

Perhaps it's the volume compression (taking the edge off sharp transients) that produces the warmer sound? Were they available, would the '60's speaker designer have uses today's polyfilm caps? We can only guess.

Improper thou it may be, I have committed poly caps for my pair; at least the two will be the same. Soon we shall see how they sound.

[kkantor]

No 6uF is NOT 6uF. There are also other things going in the cap, most notably Equivalent Series Resistance, ("ESR"). This looks like a low value resistor in series with the pure capacitance, and it must be taken into account in the ciruit model to achieve highly accurate crossover.

If you then go and change your 6uF cap to a "better" one with a different ESR, the response of the crossover will change in ways you might not expect. The tweeters might get louder. There might be an underdamped bump just before the woofer rolloff, etc.

The Marsh and Jung thing is really full of much silliness, but if you can ignore that, I think you should do a little homework on crossovers. There is plenty of material out there. I even wrote some of it, and I will post it if I find it.

[kkantor]

I personally think this article is pretty silly, for the most part.

[kkantor]

How much to you want to spend, and how much flexibility do you want?

http://www.musiciansfriend.com/srs7/sid=04...ase_pid/184127/

http://www.musiciansfriend.com/srs7/sid=04...ase_pid/183589/

http://www.musiciansfriend.com/srs7/sid=04...ase_pid/182479/

[kkantor]

I don't think cap "types" have any associated sound, per se. There are infinite ways they can effect things, depending on where they are in the circuit, how they are used in relation to the drivers, etc.

Really, you have just two choices:

1- If you like what you got, repair it and get back to listening.

2- If you dream of something different/better, modify it with new caps, coils, eqs, whatever you want. Have fun. Leave a trail of bread crumbs in case you decide to find your way home again.

[soundminded]

I agree that the final "voicing" of loudspeakers is best done with an equalizer. IMO, it should be the last task of installing a high fidelity sound system after all other factors like speaker placement and component selection are optimized. It seems to me that a parametric equalizer is not useful for most audiophiles and a 1/3 octave equalizer is far too daunting. A good 10 or 12 band equalizer is what I prefer.

I have not found built in pink or white noise generators, spectrum analyzers, or calibrated microphones particularly useful in a home setting. I think that the best insturment for adjusting an equalizer is a well trained musical ear which is familiar with the sound of unamplified musical instruments. I feel very fortuante to have been exposed to "live" music most of my life. That being said, it takes me about two years to adjust an equalizer to my satisfaction starting from scratch. It's like adjusting every color of a color television set separately and when you adjust one, the others near it change as well. With no reference such as a black and white picture to help you null your way to the right settings, it's hit or miss so one or two minor adjustments at a time with consideration of whether you have gotten closer or farther away from "accurate" is about all I can do.

IMO, the digital compact disc gives the home audiophile an unprecedented source of wide range undistorted sound never before available to the consumer. Units with 4 way repeat which allow for an arbitrary segment of music to be repeated endlessly automatically, allows for direct comparisons of one setting and another. I have found that even small changes of just one decibel for a single octave are audible.

One thing to keep in mind is that there are no standards for making recordings. Therefore a sound system which sounds accurate playing one recording can sound highly inaccurate playing another. Two nearly identical discs prove the point. Green Hill Music GHD5019 Dixieland Hymns and GHD2020 Dixieland Jazz are the same in every respect other than the musical selections except one and that is that different recording engineers made them. The bass level of 5019 is substantially greater than on 5020. Which one is right? The question is absurd. Right and wrong don't have any meaning in this situation and anyone who thinks that recording engineers won't twiddle the knobs, dials, and sliders on a studio mixing board that looks like the cockpit of a 747 is just kidding themselves. All that power at one's fingertips, who wouldn't use it?

[Guest Bret]

>This looks like a low value resistor in series with the pure capacitance, and it must be taken into account in the ciruit model to achieve highly accurate crossover. <

Thanks Ken, you really answered a much bigger question with this explanation. Since I've never designed a crossover (and don't plan to start) I didn't know that the series resistance was part of any calculation done. It could have just been a by-product of parts availability that was ignored completely at the drawing board.

If that were the case I wouldn't feel the least guilty about changing it.

BUT - you've opened another can of worms in saying that all these values were used and are important:

I can't FIND any new 2500uF Sprague capacitors. How on Earth am I supposed to be able to parallel capacitors to get that value that would have a matching ESR when I don't know what the original ESR was? And there were one or two other values in an AR-90 that simply aren't around anymore.

Please give us your best advise on what to do about those values which don't seem to be available anymore.

Bret

[johnieo]

I appreciate that the numerical value of the dissipation factor (Equivalent Series Resistance) is important in crossover design. It depends on cap type, and in NPEs, on cathode material (Al or Ta). DF also changes with age--age was my reason for scrapping. However, as Tom noted, AR itself changed cap type in the 3a. That's what made me wonder how important cap type was in the 3a. e.g., were initial differences in DF and DA between those types less than the change in either one with time?

[rrcrain]

Pardon the interruption, but the AR90 bass crossover uses a 250uf cap, not a 2500uf cap. I personally replaced the original with a 100 and 150uf Solen placed in parallel and have been quite pleased with the results.

[r_laski]

>Pardon the interruption, but the AR90 bass crossover uses a

250uf cap, not a 2500uf cap. I personally replaced the

original with a 100 and 150uf Solen placed in parallel and

have been quite pleased with the results.<

What he meant to say was the AR-90 has a 350uF cap in the woofer crossover. I paralleled a 200uF, 100uF, and 50uF North Creek cap to get 350uF. This is one mother of a capacitor stack.

There is a 2500uF cap in the AR-9 and the AR-LST. The other big cap is the 470uF in the AR-9.

Rich