HERE'S WHY YOU SHOULD REPLACE THOSE OLD 3A CAPACITORS

[Carlspeak]

Below are a group of screen shots of Woofer Tester 2 (WT2) capacitor measurements. They include recently removed Industrial Condenser Corp. (ICC) wax/paper caps. You know, the large cheese block one with the 3 metal straps which contains the 50 and 150 uF 50 VDC caps. Also the 6 uF cap. Each was measured with the WT2.

The caps I intend to replace the ICC caps with are a 50 and 150 BP 100 VDC NPE cap sold by PE and an Erse Pulse X 6 uF 400 VDC film cap. Those were also measured.

Now, look at he blue curves on each test. You'll see the curves rise upward as frequency increases. The ICC caps show this trend much more than the BP and Erse caps do. A perfect cap would stay at -90deg. for the entire measurement, yielding a flat, blue line. To the right of the plots are columns of data. The second and third columns represent the plots. Where appropriate, the data is in scientific notation. The fourth column is the cap's ESR and the last column is its true uF at each of the 31 frequencies sampled. The second column is the cap's capacitive reactance (i.e. impedance).

What makes the new test data better than the old? Note how the phase and uF data are much more consistent with the replacement caps over the range of frequencies tested. Also note that the 50 and 150 ICC caps are 8% and >10% respectively above target. The 6 uF cap is only 1%. The 50 and 150 BP caps are within 2% and the 6's are about the same. The ICC caps aren't horribly deviated from target but their higher level of variation and nominal test values are consistent with measurement taken on other caps in the past and generally higher than target.

If you want to learn more about how these measurements were taken, pick up the May issue of AudioXpress.

AR3a IND Cond Corp 6 uF.pdf

Erse 6 uF 400 V film.pdf

AR3a IND Cond Corp 150 uF.pdf

BP 150 uF NPE.pdf

AR3a IND Cond Corp 50 uF.pdf

BP 50 uF NPE.pdf

[genek]

Good info, Carl. Is this an all cap-measuring exercise, or does "intend to replace" mean you have a pair of 3a's standing by for recap? If so, any chance you can do before/after measurements of the actual speaker system and do some listening comparisons between recapped and non to assess how caps that measure different translate to differences in system measurements and actual changes in sound?

[Carlspeak]

I have a pair standing by. One is opened and one speaker I haven't opened yet, but know it's in need of a refoam also. This job is tight for time so I won't be able to do what you asked - this go around. Sorry.

[michiganpat]

interesting...I'm surprised the original caps measured so close to spec capacitance wise....

[genek]

That's why I'm interested in full-system measurements before/after and in listening tests. My experience with recaps (much more limited than Carl's) has been that caps that weren't low quality to begin with don't drift nearly as much as most people seem to think they do and unless the old cap is grossly (20% or more) out of spec or had suffered obvious failure (bursting, leaking, etc.) it was awfully hard to be sure there was really any significant difference in the sound before/after. My biggest concern about old caps is usually whether they are still blocking the necessary frequencies to protect drivers from damage.

[rrcrain]

Considering how many capacitor checkers work, this makes a dremendous amount of sense. My digital capacitance meter measures the RC time constant, basically at one frequency. This new method shows a caps performance across a frequency range better depicting the audio circuit it's used in.

[Carlspeak]

It looks like the charts are scaring some people away. The first one has about twice the number of downloads as the second and so forth.

If you study the ESR data in the 4th column from the left you'll see the series wired 6 and 50 uF ICC caps had about 0.20 to 0.25 ohms. Whereas, the newer caps were well below 0.02 and 0.1 ohm respectfully. These differences will result in slightly higher SPL output from the mid and tweeter. However, Gene I wouldn't expect any earth shattering differences in overall response performance. Also worthy of note is the dissipation factor ran about 2 to 2.5% for the vintage caps - not bad at all.

What's not measureble (at least by me), are the more subjective aspects of listening impressions that come out of a recap of a vintage speaker.

Keep in mind, what I presented in the first post was a data set from one sample (i.e. each cap).

Also did some AR2a 4 and 6 uF caps a while back for the AE article. The 4 uF ICC tap was over 5 (+25%), ESR was around 0.3 ohms. the 6 uF ICC tap was around 7.5 (also +25%). ESR was around 0.4 ohms. A Dayton 4 uF MPP cap averaged 3.97 uF and its 31 data point coeficient of variation was only 0.55% vs 4.65% for the ICC cap.

[RoyC]

We sent John O'Hanlon a fairly large number of caps to measure in a similar manner (using different testing equipment) a few years ago.

His results were much like Carl's for the Industrial Condenser block caps, with some showing a much greater degree of variation. The small Callins type black and red caps, found mostly in KLH speakers and AR-11 era models, were consistently poor. The original, large AR can-type Sprague caps used in mid 70's AR models actually measured better than many of the modern electrolytic caps included in the testing. John's results showed some inconsistencies in modern npes, which motivated him to suggest film caps as replacements when budget allows.

Thanks for the data, Carl.

Roy

[Carlspeak]

We sent John O'Hanlon a fairly large number of caps to measure in a similar manner (using different testing equipment) a few years ago.

His results were much like Carl's for the Industrial Condenser block caps, with some showing a much greater degree of variation. The small Callins type black and red caps, found mostly in KLH speakers and AR-11 era models, were consistently poor. The original, large AR can-type Sprague caps used in mid 70's AR models actually measured better than many of the modern electrolytic caps included in the testing. John's results showed some inconsistencies in modern npes, which motivated him to suggest film caps as replacements when budget allows.

Thanks for the data, Carl.

Roy

Roy, it would be nice if John's measurement data, if still available, were compiled into a spreadsheet or some other form for us technical types. Could that be done? I realize my measurements delve into the subject more deeply, but confirming data from separate sources is always handy to study and learn.

[genek]

However, Gene I wouldn't expect any earth shattering differences in overall response performance.

That's what I'd expect, subtle changes if detectable at all by ear. I'm constantly perplexed by the people who recap (especially newer speakers from the 80s and 90s) and wax poetic about veils being lifted and day and night differences in sound. Especially when I listen to their speakers after and think they sound pretty much the same as they did before.

[Carlspeak]

That's what I'd expect, subtle changes if detectable at all by ear. I'm constantly perplexed by the people who recap (especially newer speakers from the 80s and 90s) and wax poetic about veils being lifted and day and night differences in sound. Especially when I listen to their speakers after and think they sound pretty much the same as they did before.

Where applicable, a recap's sometimes big difference is caused by ESR changes. It all depends on how much the old cap's ESR has drifted. uF values also drift in old caps, but I think the shift back to target brought on by a new cap is not as sonically noticeable. From there, the improvements become exponentially more subtle to a level that requires golden eared panelists doing many tests followed by a rigorous statistical analysis to find a difference.

However, as you point out, expectations can run high when someone recaps and the waxing poetic soon follows. Or, after 100's of hours of 'break-in', they reach a nirvana state - at least psycho-acoustically.

[RoyC]

Roy, it would be nice if John's measurement data, if still available, were compiled into a spreadsheet or some other form for us technical types. Could that be done? I realize my measurements delve into the subject more deeply, but confirming data from separate sources is always handy to study and learn.

Carl,

John's measurements were much the same as yours. Maybe he will check in here at some point.

The ESR readings you measured were similar to John's as well, and was the motivation for his recommending the addition of .3+/- ohm series resistors to replacement caps for those who may want to try to mimic the higher ESR of the original caps. (I personally don't ascribe to this practice, as we really don't know what the original ESR was...and the level controls typically show variations of greater than .3 ohms.)

Roy