Tony Gee has spent probably more time than anyone evaluating caps and sharing his questionable results. A few years back I noticed how he seemed to be rating the higher voltage (e.g. also are typically more expensive) caps higher than lower voltage units and ran a simple correlation. I was surprised to see that indeed, with a couple of outliers removed from the analysis, there was good correlation between his subjective ratings and voltage rating. I wrote a paper on the subject that is now up to over 1100 downloads here
Capacitor voltage rating vs performance - The Classic Speaker Pages Discussion Forums
The paper included some communication between me and him about what I found, which initially, seemed to surprise him. I got him to agree to a special test for which I waited some months for the results. I requested at least a SBT be done, but he ignored that; much to my dismay.
Secondly, has anyone tried hooking a cap up to a relatively high power signal source up to 20 AC volts with a high wattage 8 ohm dummy load in place of a speaker? With some caps, at about the 10 volt level, they literally start to 'sing' if you listen closely. Other caps don't. I'm not sure if it's a microphonics effect or something else. Give it a try and see for yourself.
Check out at this link starting at post #185.
Burning in hi-pass capacitors - Page 10 - Techtalk Speaker Building, Audio, Video, and Electronics Customer Discussion Forum From Parts-Express.com
We're not getting near answering the Mundorf question here, but I can relate my experiments with three capacitors in a simple 2.2uF + 47K resistor highpass crossover to a metal dome tweeter. The lowpass was just an inductor on simple little Mordaunt Short compact reflex speakers on stands. Here's a ca. 1997 photo to prove I don't make this stuff up!
This was a very good front end of high-end Rotel preamp and power amp, my current RCD-965BX but iced with an expensive Rotel DAC. The DAC brought a stunning bass register and a silky top end. Wish I still had it actually.
The music was Karolina Kruger's "Se pa meg", IIRC. A very good and delicate female vocal.
What I noticed was a sibilance on her voice, but interestingly, it was not appearing in the centre of the stereo image, but rather out of phase on either side of her vocal. Sibilance, or a hissing on "S" sounds really, is about 7kHz, as it goes. The crossover must have operated about 3 kHz.
Opening up the speakers, I found 2.2uF non-polar electrolytics. I swapped them some smallish Solen (250V?) MKPs I had lying about. This made a very noticeable difference, and the sibilance disappeared. Being interested by now, I bought some very fat 630V 2.2 uF Audio Grade Polypropylene Capacitors at Maplin. Swapping these in and out, I decided that the big fatties from Maplin had a lot more top end than the Solen's. A lot brighter. Well, that's what I know.
An externally hosted image should be here but it was not working when we last tested it.
This was a very good front end of high-end Rotel preamp and power amp, my current RCD-965BX but iced with an expensive Rotel DAC. The DAC brought a stunning bass register and a silky top end. Wish I still had it actually.
The music was Karolina Kruger's "Se pa meg", IIRC. A very good and delicate female vocal.
What I noticed was a sibilance on her voice, but interestingly, it was not appearing in the centre of the stereo image, but rather out of phase on either side of her vocal. Sibilance, or a hissing on "S" sounds really, is about 7kHz, as it goes. The crossover must have operated about 3 kHz.
Opening up the speakers, I found 2.2uF non-polar electrolytics. I swapped them some smallish Solen (250V?) MKPs I had lying about. This made a very noticeable difference, and the sibilance disappeared. Being interested by now, I bought some very fat 630V 2.2 uF Audio Grade Polypropylene Capacitors at Maplin. Swapping these in and out, I decided that the big fatties from Maplin had a lot more top end than the Solen's. A lot brighter. Well, that's what I know.
Attachments
Last edited:
We're still not getting any nearer to answering the Mundorf question here; and we won't either. It's too subjective a subject and SBT and DBT testing requires great rigor to do correctly.
Your experience with the Solens replacing NPE's is interesting in that much chat I've read on the web is just the opposite. Solens used to recap are much more sibilant than many other film caps and, particularly NPE's.
The caps you ended up using at 2+ lbs sterling are a long way from Mundorf's. I recently upgraded the xover in a pair of Swiss-made Ensemble Reference speakers. The 6.8 uF 1000 V caps ran about $150 U.S. each!
Your experience with the Solens replacing NPE's is interesting in that much chat I've read on the web is just the opposite. Solens used to recap are much more sibilant than many other film caps and, particularly NPE's.
The caps you ended up using at 2+ lbs sterling are a long way from Mundorf's. I recently upgraded the xover in a pair of Swiss-made Ensemble Reference speakers. The 6.8 uF 1000 V caps ran about $150 U.S. each!
I've really no idea which particular model from the vast Solen range they were. But cheapish.
I do suspect that Solen, Mundorf and Jantzen follow each other quite closely in what is a fairly established technology of metallised film polypropylene. FWIW, at Solen, you can go Teflon too, which has a higher dielectric constant.
Those 630V Maplin's jobbies, to my engineering mind, are pretty good if unpretentious. I'd suspect they are really designed for mains power applications.
Should I mention that replacing electrolytics with compact MKT polyesters in the output of cheapish CD players is a good upgrade too? Much nicer top end. Oh, I just did!
I do suspect that Solen, Mundorf and Jantzen follow each other quite closely in what is a fairly established technology of metallised film polypropylene. FWIW, at Solen, you can go Teflon too, which has a higher dielectric constant.
Those 630V Maplin's jobbies, to my engineering mind, are pretty good if unpretentious. I'd suspect they are really designed for mains power applications.
Should I mention that replacing electrolytics with compact MKT polyesters in the output of cheapish CD players is a good upgrade too? Much nicer top end. Oh, I just did!
Wow , way to cop out. I guess you had nothing to show us from the beginning. "Real science", you sound like a "know it all" that likes to spout his opinion then back it up with.........nothing.
So doing a simple SBT is above anyone's ability on here. Fine. Subjective opinion is just that, opinion and no science involved. Have fun discussing "what you heard"without any controls set in place to verify it.
Last edited:
I think you guys are getting too bogged down in this SBT and DBT business. You all have ears, hopefully not cloth-ears, so I think you should try some stuff yourselves.
There's a great deal of subtlety and engineering good practise to apply to crossovers. Putting a capacitor near an inductor, or placing inductors so they interact and crosstalk clumsily will dramatically alter tonal response. This is the simple physics of an inductor near a conducting plate or coil. Quality of components will affect Q as well, and ferrite cores have considerable non-linearities, especially when there is a DC component, as will be noticed when mains transformers buzz due to DC pickup from the mains.
Look at this typical SEAS metal dome tweeter:
Some considerable supersonic resonances which a poor amplifier will make a mess of handling due to RF decoupling of the output. I must be wrong in recalling that HF tweeter shunt was 47K. 47R makes more sense and would make it better behaved.
We have established that capacitors "Sing" and coils "Buzz", and that crossover components interact and the effect is audible. The trick is to reduce it to a minimum with good layout and a balanced set of high quality components.
There's a great deal of subtlety and engineering good practise to apply to crossovers. Putting a capacitor near an inductor, or placing inductors so they interact and crosstalk clumsily will dramatically alter tonal response. This is the simple physics of an inductor near a conducting plate or coil. Quality of components will affect Q as well, and ferrite cores have considerable non-linearities, especially when there is a DC component, as will be noticed when mains transformers buzz due to DC pickup from the mains.
Look at this typical SEAS metal dome tweeter:
An externally hosted image should be here but it was not working when we last tested it.
Some considerable supersonic resonances which a poor amplifier will make a mess of handling due to RF decoupling of the output. I must be wrong in recalling that HF tweeter shunt was 47K. 47R makes more sense and would make it better behaved.
We have established that capacitors "Sing" and coils "Buzz", and that crossover components interact and the effect is audible. The trick is to reduce it to a minimum with good layout and a balanced set of high quality components.
A "simple SBT" won´t give you any useful data.
Something to read:
http://www.emea.europa.eu/docs/en_GB/document_library/Scientific_guideline/2009/09/WC500002928.pdf
http://apps.who.int/rhl/LANCET_966-970.pdf
Those documents are for setting up experiments to answer a very different question. The question here is, "can the poster hear a difference between A and B?" not, "what percentage of the population can hear a difference between A and B?"
Setting up a controlled listening test to answer the former question is not that difficult. See Linear Audio Volume 2 for ways that can be easily accomplished. Of course, before doing that, it's also easy to check for electrical differences that would be in the range of clear audibility.
Setting up a controlled listening test to answer the former question is not that difficult. See Linear Audio Volume 2 for ways that can be easily accomplished. Of course, before doing that, it's also easy to check for electrical differences that would be in the range of clear audibility.
I think you guys are getting too bogged down in this SBT and DBT business. You all have ears, hopefully not cloth-ears, so I think you should try some stuff yourselves.
There's a great deal of subtlety and engineering good practise to apply to crossovers. Putting a capacitor near an inductor, or placing inductors so they interact and crosstalk clumsily will dramatically alter tonal response. This is the simple physics of an inductor near a conducting plate or coil. Quality of components will affect Q as well, and ferrite cores have considerable non-linearities, especially when there is a DC component, as will be noticed when mains transformers buzz due to DC pickup from the mains.
Look at this typical SEAS metal dome tweeter:
An externally hosted image should be here but it was not working when we last tested it.
Some considerable supersonic resonances which a poor amplifier will make a mess of handling due to RF decoupling of the output. I must be wrong in recalling that HF tweeter shunt was 47K. 47R makes more sense and would make it better behaved.
We have established that capacitors "Sing" and coils "Buzz", and that crossover components interact and the effect is audible. The trick is to reduce it to a minimum with good layout and a balanced set of high quality components.
All well and good IF people could put aside their expectation bias. "Oh look at that pretty expensive cap, it MUST sound a LOT better than that ordinary one" etc etc....
Caps do "sing" and coils "buzz" BUT does that influence the MEASURABLE response? So far no one has bothered to see.
Like SY just said, that's not related to the problem here. Good Lord if you or anyone can't set up a simple SBT and TRY then you are hopeless.Would you rather discuss "why" SBT and DBT are "useless" with audio testing or would you rather get some real results?
Any acoustic resonance shows up on an impedance curve. Google "Heyser plot." Whenever a cap rattles, you get a pigtail at the resonant frequency- that's what cured me of ever using "audiophile" caps again.
See, for example, Duncan, Williams, and Dodds, AES Convention Paper 7682 (2009), and from the same authors, IOA Reproduced sound 22, Vol 28.Pt 8, November 2006.
See, for example, Duncan, Williams, and Dodds, AES Convention Paper 7682 (2009), and from the same authors, IOA Reproduced sound 22, Vol 28.Pt 8, November 2006.
Rather than the acoustic radiation from the surface of the capacitor, it's about the behaviour of the capacitor varying, and that being modulated onto the signal.
This shouldn't be about the last few percent being the most significant because once the quality of a cap is ninety-something percent there and your speaker still sounds bad, you won't fix it with a perfect capacitor.
This shouldn't be about the last few percent being the most significant because once the quality of a cap is ninety-something percent there and your speaker still sounds bad, you won't fix it with a perfect capacitor.
Not only did that paper show objective measurable evidence of capacitor resonance issues, but equally important, showed the researchers how difficult it was to get a panel of subjective listeners to hear differences they were trying to show. They ended up using a 'golden eared' panel instead of random listeners and using a specific test protocol.
Here's the link for the Clarity Cap research, if you want to read it:
ClarityCap Capacitor Information Page
Very interesting, IMO. There's something going on around 10-20kHz.
ClarityCap Capacitor Information Page
Very interesting, IMO. There's something going on around 10-20kHz.
I understand your frustration BUT! did you read Tony Gee's ratings of these caps at humblehomemadehifi? This is probably the BEST subjective information you can get. He's done many, many cap tests and spends most of his days involved with speaker building and assessing capacitors in loudspeakers.
I would much rather base a significant purchasing decision on his opinion than a random poster here (if you're lucky to get one) who's done exactly what you ask). Who's opinion would you trust?
Gosh then those caps must be heard only by Golden Ears and only by using the specific test protocol. If it was difficult to get the "subjective listeners" to hear a difference then just maybe there wasn't a difference to hear? Just asking since I don't have the paper in question to read. If you have to resort to hand picking a panel and using only a specific test, then that's pretty much admitting there isn't much to hear to begin with.
Here's the link for the Clarity Cap research, if you want to read it:
ClarityCap Capacitor Information Page
Very interesting, IMO. There's something going on around 10-20kHz.
I've seen that before from Clarity and wasn't much impressed. It's their own "test" using their own product. Lets see some 2nd or 3rd party tests.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.