Cap Value Accuracy

[Pete B]

This supports what I said about measuring around 100 Hz for caps over 10 uF:

http://product.tdk.com/capacitor/mlcc/en/faq/faq00021.html

http://www.digikey.com/Web%20Export/Supplier%20Content/Kemet_399/PDF/kemet-measure-capacitance-of-class-ii.pdf?redirected=1

http://www.cde.com/resources/catalogs/AEappGUIDE.pdf

Just checked my Fluke 87, an excellent meter, only measures up to 5 uF directly.

[Carlspeak]

Electrolytics are often measured around 100 Hz because their value is somewhat frequency dependent.

The tolerance is from the rated value however they are measured not over the frequency range.

It would be good Carl to measure a big NPE at both low and mid frequency on your meter. I have some

decent meters but I don't think any of them measure capacitance.

I use LAUD to measure capacitors and inductors.

Not all DMMs work like this, but it is one reference:

http://en-us.fluke.com/training/training-library/test-tools/digital-multimeters/how-to-measure-capacitance-with-a-digital-multimeter.html

I use WT2 to meas. caps and inductors.

Ran a few scans today on a 100 uF Bennic 100V (madisound), and a 100 uF Jantzen MKP 400V

The bennic meas. 101.57uF @ 109 hz and 97.33 uF @ 1188 hz.

The Jantzen meas 97.05 uF @ 109 hz and 97.1 uF @ 1188 hz.

At these hi uf values, I don't think the NPE meas. variance (although larger than the film) is significant.

Also checked a 50 uF 100V ERSE NPE and a vintage Sprague compulytic 50 uF 50V NPE

The ERSE meas 50.47 uF @ 109 hz and 47.1 uF @ 1188 hz.

The Sprague meas. 55.94 uF @ 109 hz and 54.9 uF @ 1188 hz.

PM me with your email address if you want full screen shots of those tests.

[RoyC]

Yesterday I sent some Parts Express and Erse npe's (8, 12, 50 and 100uf) off to John O'Hanlon for him to measure at the lab. They all measure 10%+/- to the high side on several inexpensve cap meters I have here. I should have his measurements in a week or so, and will report back.

Roy

[Pete B]

I use WT2 to meas. caps and inductors.

Ran a few scans today on a 100 uF Bennic 100V (madisound), and a 100 uF Jantzen MKP 400V

The bennic meas. 101.57uF @ 109 hz and 97.33 uF @ 1188 hz.

The Jantzen meas 97.05 uF @ 109 hz and 97.1 uF @ 1188 hz.

At these hi uf values, I don't think the NPE meas. variance (although larger than the film) is significant.

Also checked a 50 uF 100V ERSE NPE and a vintage Sprague compulytic 50 uF 50V NPE

The ERSE meas 50.47 uF @ 109 hz and 47.1 uF @ 1188 hz.

The Sprague meas. 55.94 uF @ 109 hz and 54.9 uF @ 1188 hz.

PM me with your email address if you want full screen shots of those tests.

You mentioned that you had a meter that measured at 120 Hz and 1 KHz that is what I was talking

about because I don't have a meter like that.

I use LAUD that provides similar curves to what you described above and you are not seeing what

people here are talking about, they seem to be within spec which is what I've also found over the

years.

[Carlspeak]

Pete, I went to the WT2 because I think my LCR meter displays capacitive reactance instead of true capacitance at the 1 kHz and 120 hz levels.

[Pete B]

You're meter has a range that is in agreement with how the manufacture's actually test and rate

their components, the question is if you get a different reading at the two frequencies. I expect

that you will.

The big capacitor companies know what they are doing and have far better test equipment than

we do for these measurements. I seriously doubt that they are getting it wrong. In fact, I think I

looked up their test procedures about 10 years ago when I looked at some Bennics in LAUD.

They were off at 1 KHz and that is what made me dig into it, they were within spec at 100 Hz.

[DavidDru]

I was hoping to have the guys at Erse test them and see what they came up with but apparently that person has been out for the last week or so. We shall see what they come up with.

[RoyC]

You're meter has a range that is in agreement with how the manufacture's actually test and rate

their components, the question is if you get a different reading at the two frequencies. I expect

that you will.

The big capacitor companies know what they are doing and have far better test equipment than

we do for these measurements. I seriously doubt that they are getting it wrong. In fact, I think I

looked up their test procedures about 10 years ago when I looked at some Bennics in LAUD.

They were off at 1 KHz and that is what made me dig into it, they were within spec at 100 Hz.

Hi Pete,

See below...Bennic specs state +/- 10% at 1KHz. I agree it seems to be a meter/npe measurement method issue, but perhaps it has more to do with meter quality (?). Carl's WT2 measurements are more like what I would expect to see. The interesting thing is that film caps do not appear to present a measurement problem, regardless of the meter used.

MODEL BENNIC BI-POLAR ELECTROLYTIC CAPACITOR (B05A) LEADS AXIAL LEADS, 40MM LEAD LENGTH TOL. RANGE : ± 10% (K) @1KHZ

Roy

[Pete B]

Hi Roy, most test around 1K for 10 uF and below and the 100 ish above, but that is possible for Bennic.

Audio companies do not always follow industry standards. Perhaps I flipped the frequencies from when

I made those Bennic measurements but I do remember looking up the test conditions for a reason.

Films are so much more consistent and closer to ideal that it is not surprising that they are easier to test.

[RoyC]

Motivated by Pete's insights, Carl's WT2 measurements, and emails with John O'Hanlon, I decided to dig for some specs for common capacitance testing meters, and upgrade my own WT2 program for capacitor measurement. This is what I found:

-Non polar electrolytic caps (npe's) measurement results are significantly more frequency dependent than even the least expensive fim cap (mylar or poly).

-Simply put, as frequency increases measured capacitance goes down. As frequency decreases it goes up. This is much more pronounced with npe's. The changes are insignificant with film caps.

-In light of the above, I found that the current crop of handheld cap meters automatically vary frequency with the range setting! The typical meter is measuring caps below 2uf at 800hz, 2 to 20uf at 80hz, and above 20uf at 8hz. This will always result in higher readings for npe's rated at 1000hz, such as the Bennics discussed above.

-See the technical specs of this popular meter...which are nearly identical to a more costly B&K meter used at a shop for which I do speaker repairs.

http://www.newark.com/tenma/72-8150/capacitance-meter-with-resistance/dp/41M0998

-All of the above was able to be verfied with WT2 measurements of npe's and film caps.

The bottom line is that npe's are likely to measure higher than stated value with typical handheld cap meters.

Roy

[Carlspeak]

The question remains: Is the actual meas. value within 10% at the freq. normally used for that value cap? I suspect what Roy found is well known in the cap industry and they 'should' be building them to that standard.

For example, a 6 uF NPE cap (common tweeter cap) should measure 6 uF + - 10% in say the 2-6 kHz range?

[RoyC]

The question remains: Is the actual meas. value within 10% at the freq. normally used for that value cap? I suspect what Roy found is well known in the cap industry and they 'should' be building them to that standard.

For example, a 6 uF NPE cap (common tweeter cap) should measure 6 uF + - 10% in say the 2-6 kHz range?

Well, if Bennic is the norm, and the only published spec is +/- 10% at 1000hz regardless of stated value, who knows...and meters telling us what they measure only at very low frequencies are not providing much useful information.

Npe behavior relative to frequency compared to that of film caps seems to support Ken Kantor's recommendation to replace npe's with npe's if the goal is to stay as close as possible to the original sonic character of a speaker system designed around this type of cap.

Btw, I received an order of Bennic npe's today, and they measure no better than their Erse and Parts Express counterparts.

Roy

[owlsplace]

Well, if Bennic is the norm, and the only published spec is +/- 10% at 1000hz regardless of stated value, who knows...and meters telling us what they measure only at very low frequencies are not providing much useful information.

Npe behavior relative to frequency compared to that of film caps seems to support Ken Kantor's recommendation to replace npe's with npe's if the goal is to stay as close as possible to the original sonic character of a speaker system designed around this type of cap.

Btw, I received an order of Bennic npe's today, and they measure no better than their Erse and Parts Express counterparts.

Roy

Interesting conclusion ... I'm too lazy to do the math to get an idea of how much frequency dependent difference there would be in a crossover compared to the normal degradation of a fifty year old driver. Are we picking nits or is it worth serious consideration.

Then again, why bother to replace the old npe's if you are unable to determine their condition or the condition of the replacements?

Roger

[Carlspeak]

..........

Then again, why bother to replace the old npe's if you are unable to determine their condition or the condition of the replacements?

Roger

NPE's are made with an electrolytic chemical paste that degrades over a timeframe similar to that of foam surrounds. As NPE's degrade, their key performance properties (uF and ESR) drift over time out of spec. This WILL affect the sound.

So, the precautionary thing to do is assume old NPE's have drifted and take advantage of replacing them when doing a refoam job.

With regard to KK's recommendation to replace NPE's with NPE's will indeed keep the sound as indended by the designer, but the drifting will still occur years into the future. I have found mylar film caps have a key property (ESR) similar to NPE's and won't drift. So, if your doing a recap and want the performance to not shift 10-15 years out, mylar film caps are the way to go. They are the yellow MET type sold by ERSE and also the Carli caps sold by Madisound.

Lastly, I believe Bennic also private labels their caps for Parts Express. So, both sould be of the same quality.

[RoyC]

Carl,

I've been a big fan of the mylar/polyester variety of crossover capacitor for a long time, and it continues to be a go-to cap for me...but it is still a "film" cap with frequency behavior and esr more like a polypropylene than an electrolytic cap. Many old speakers have simple crossovers with gradual crossover slopes, and a single cap may influence alot of territory.

Between esr and crossover slope differences resulting from frequency response characterisitcs, it is not hard to see why using film caps in place of npe's can result in more pronounced high frequencies and upper midrange harshness. I have sometimes found this to be the case in later speakers dating from the 80's onward, and attribute this to healthy drivers and flatter initial frequency response. Otoh, I believe the behavior of film caps can serve to brighten up the downward sloping high frequency response of older speakers, resuting in a perceived improvement...and using mylar caps maintains some lost series resistance to smooth things out a bit.

At this point my opinion on using new npe's has improved considerably, and I no longer believe the "high" measurement results to be one of questionable npe quality. I strongly suspect better npe's last much longer than they are given credit for in speaker forums. (I've measured the original "Culver" npe's in a boatload of 80's Jensen era Advents, and I have not found one to measure poorly....even though I have replaced many of them as insurance.)

Roy

[DavidDru]

I took the ERSE NPE's that started this thread to my tech so he could use his higher end test equipment. Same results as my meter. 12uf caps measured at 14.1. He said they leaked a bit too much too. Went through various steps of volts.

He also tested the yellow MET film caps and said they were great.

I will try to get hold of the folks at Erse again tomorrow and see what they think. Hopefully that particular guy will be in.

[owlsplace]

I have two pair of 6's that need to be done and if I remember correctly at least one of them had npe's so maybe I'll do a little a/b cap comparison.

Roger

Added: It shouldn't be too difficult to outboard the caps on these speakers ...

[RoyC]

David,

The point of the last batch of posts in this thread is that these npe's cannot be measured accurately with a meter unless it has multiple user-adjustable test frequencies (which is not likely). I'm certain your tech's meter is not unlike those in the shops I work for. These meters only measure at certain frequencies (nothing to do with "volts"). Your MET cap measured well because it is a film cap, and as such, is not affected much by the frequency at which it is tested. Every new 12uf npe I have here (Erse, Bennic, and Parts Express) have measured between 13.5 and 14.1uf on 5 meters. I doubt Erse will be of much assistance other than to refund the few bucks these caps cost you.

Roy

[Jeff_C]

I came across this lifetime calculator for various types of caps. My maths is not what it used to be 40 years ago, but believe I can see from this formula for electrolytic caps that the expected lifetime can be doubled by increasing the cap's voltage spec. When I did my recap I replaced 50V NPE caps with 100V. Am I right in thinking that the operating lifetime has been doubled. If so this would seem a sensible thing to do when replacing NPE caps with new NPEs

http://www.illinoiscapacitor.com/tech-center/life-calculators.aspx

[Carlspeak]

It's clear from the formula that temp (X exponent) has the most impact on life. So, I would surmise from that, that folks with NPE's in their speakers who play their speakers a lot and live in a generally warm environment will experience more drift in their NPE's properties over time.

[iso]

20UF 100V hi power Alcap do have 0,1 DF, 20UF 50V low loss Alcap do have 0,05 DF... so 50V rated item will have 50% lower losses and heat build up if voltage rating is good enough. Original low cost 50V Alcap do have same 0,1 DF as 100V rated high power one. So... higher voltage rating do not necessarily mean lower operating temperature.

http://www.falconacoustics.co.uk/alcap-claritycap-solen-audio-capacitors/capacitors-audio-crossover-nonpolarised-electrolytic-polypropylene-polycarbonate-polythene-1200uf-3900uf.html?cat=103

Best Regards

Kimmo

[Jeff_C]

Kimmo you will have to enlighten me about DF. What is that a measurement of?

My reading of the formula I linked to is that a higher voltage rating of the cap will increase the life expectancy by a factor Vr / Vo where Vr is the maximum voltage rating of the capacitor and Vo is the operating voltage of the capacitor. I may be wrong.

Carl has noted that the X exponent (derived from temperature) will have a much greater impact on life expectancy than the Vr / Vo part of the formula. I would have to see realistic figures inserted into the formula in order to grasp how it affects the overall life expenctancy.

[iso]

Jeff... glad you asked this

DF is basically ratio of capacitor ESR (equivalent series resistance) to impedance (AC resistance) of capacitor at given frequency. Falcon has given DF figures at 1khz.

If we look at 24 uF AR5 mid range capacitor at middle of pass band. AC resistance of 24 uF capacitor is 4,4 ohms at 1500 hz. This means that of ESR of 0,1 DF capacitor is 0,44 ohms and ESR of 0,05 DF capacitor is 0,22 ohms. So... 10% of energy that will go trough 0,1DF capacitor will generate heat in the capacitor and 0,05 DF capacitor will generate only half as much heat. Good electrolytic´s can usually dissipate this heat in domestic installations, as average power used is usually only few watts. However in PRO installations where hundreds of watts of compressed music is amplified film caps are only way to go, as their DF figures are usually less than 0,001.

DF is not constant figure as it will change when frequency will change for two reasons. ESR of capacitor is not constant at all frequencies but more importantly impedance of capacitor will halve when frequency will double. At this site there are few easy to use calculators that are handy when figuring what happens in electronic devices. http://www.pronine.ca/links.htm

Quite often low losses have been suggested to be reason for using film caps. However, if you look what low losses do mean in AR5 crossover. In series to 8 ohm mid driver at middle of pass band (1500 hz) there are the following elements 24uf cap (4,4 ohm impedance), 1,3 ohm resistor, 0,16mH inductor (1,6ohm impedance) and maybe 4 ohms of pot wiper totaling 11,3 ohms without capacitor ESR. So... ESR of 0,1 DF capacitor will attenuate signal approx 0,3 dB compared to 11,3 ohms in series to drive unit, low loss Alcap will attenuate only half as much. However actual loss due series cap ESR is actually somewhat smaller, as there is also drive unit in parallel to 12 ohms of pot wiper and 2,4mH inductor in series to 24uF capacitor.

I do think that limited life span of NPE:s is the only real handicap they present engineering wise. In woofer circuits it is quite usual that there is small 1,5 ohm resistor in series to cap... so slightly higher ESR should not be any problem here either. However one may think that sound of film cap is better... but we are now entering to territory where absolute phase, phasing of mains plugs and exotic cabling should be also considered as facts of life.

Best Regards

Kimmo

[donquixote99]

Quite often low losses have been suggested to be reason for using film caps. However, if you look what low losses do mean in AR5 crossover. In series to 8 ohm mid driver at middle of pass band (1500 hz) there are the following elements 24uf cap (4,4 ohm impedance), 1,3 ohm resistor, 0,16mH inductor (1,6ohm impedance) and maybe 4 ohms of pot wiper totaling 11,3 ohms without capacitor ESR. So... ESR of 0,1 DF capacitor will attenuate signal approx 0,3 dB compared to 11,3 ohms in series to drive unit, low loss Alcap will attenuate only half as much. However actual loss due series cap ESR is actually somewhat smaller, as there is also drive unit in parallel to 12 ohms of pot wiper and 2,4mH inductor in series to 24uF capacitor.

I read this paragraph and am left with questions about 'what low losses mean in AR5 crossover.' What is the DF of the OEM cap? What happens to the sound if this changes? I'm getting that it will be attenuated less, but significantly less? Will there be any other audible changes in crossover performance?

[Carlspeak]

I think the whole DF diversion has complicated the the discussion. DF is directly related to ESR (ESR is easier to understand because the units are ohms) and like ESR varies with fequency.

Read more about it here....

http://en.wikipedia.org/wiki/Dissipation_factor