Maybe that 25% statement isn't about pure tones but subject to masking when complex signals at high SPL are reproduced?
SY said:
I've never measured it either and I'd love to know. I've always been afraid to use electrolytics, but I hate paying the steep prices for the film stuff. I won't buy boutique components at all.
Here is what I would do.
- resistor in series with the cap, drive this with the signal. Measure the voltage across the cap. I would look at very low voltages, as low as you can measure down to without noise and signal distortion. Plot the input spectrum and the output spectrum to be sure that the distortion is from the cap. You could cross correalte the signals as this would help with the noise as well as normalize out the signals distortion. Try different frequencies. The signals have to be keep well below the LP knee of the RC combination.
I would love to see this data for several caps.
adason said:
Your test is not what I am talking about, and yes, I always refer to audibility of real music as sine wave are a different animal.
We tested compression drivers with real music at at 25% distortion the test panel of 30+ listeners could not reliably detect it. Its mostly 2nd order or course.
gedlee said:
All right then, if you say so.
gedlee said:
Oops! I thought you said the difference was "highly audible"? Now I'm really confused. You either know or you don't know.
Which is it?
MJL21193 said:
I said "as found in ..." I never said that I tested caps. I KNOW that .1% distortion of the crossover waveform would be audible no matter what system creates it.
Why are you trying to be difficult? All your criticisms of me are based on your misreading of what I write.
Cyrll Bateman's articles are on the web now:
http://uk.geocities.com/cyrilb2@btinternet.com/
(the site has a bandwidth limit, dowload the last article first for Al electro info)
you can see from his measurements that bipolar electros measure lower distortion
however the best Al bipolar in series combination just equaled PET/Mylar, and PET wins the measured distortion game if you have to block DC
http://uk.geocities.com/cyrilb2@btinternet.com/
(the site has a bandwidth limit, dowload the last article first for Al electro info)
you can see from his measurements that bipolar electros measure lower distortion
however the best Al bipolar in series combination just equaled PET/Mylar, and PET wins the measured distortion game if you have to block DC
SY said:
You don't have anything better to do? 🙂
Jcx, good stuff. No mention of xover distortion though, and I'm disappointed.
SY said:
I don't know if its relevant but in common cathode circuits I remember to have read in an old article that is good to use the bigger the bypass cap possible not just to silly extend the LF beyond anything phase turn acceptable, but the lower the knee the less the distortion. Maybe electrolytics are weaker for distortion down low.
gedlee said:
Sorry, but it IS true. In DBX testing, the 'trained' (ie. they know what to listen for) ear can hear 2nd order distortion at levels on 5% or more (maybe lower).
But I do understand your point - 2nd order is much easier on the ear and can, in fact, be confused for the original sound if the listener doesn't know what he/she is hearing. And it requires 1 or 2 orders of magnitude more of 2nd order to detect it compared to other (more annoying) forms of distortion.
Errrm, what I was meaning was, why not run in a regime where the voltage across the cap is low? Wouldn't the nonlinearity (if the mechanism is what you hypothesize) be greater at lower voltages (i.e. higher frequencies)?
I have a 1u bipolar (Radio Shack special) and 1k resistor in series, driven by a low distortion buffer (BUF03). If my young 'un cooperates and lets me finish this, I'll have some spectra later.
Of great interest to me is the tweeter feed case, a high-pass with the test frequency being above fo and a higher level signal with lower load impedance (like 8 or 10 ohms). That may have to wait for another day.
I have a 1u bipolar (Radio Shack special) and 1k resistor in series, driven by a low distortion buffer (BUF03). If my young 'un cooperates and lets me finish this, I'll have some spectra later.
Of great interest to me is the tweeter feed case, a high-pass with the test frequency being above fo and a higher level signal with lower load impedance (like 8 or 10 ohms). That may have to wait for another day.
Cloth Ears said:
This test is documented and done with music signals? Because I have a documented and peer reviewed published study to the contrary.
How was the "pure" 2nd order distortion created? How do you know that there weren't other orders?
SY said:
There will be a natural LP filter effect that will supress the harmonics unless corrected for. This may or may not be relavent, but iof you are say 1/10 of the LP point then its not an issue.
I suppose you could just as easily us a HP filter with the same setup. Might even be more enlightening.
gedlee said:
Hi Earl,
Is it possible that the test panel just didn't have the, um, special listening abilities (as determined through self analysis and an internet connection) of the the Audiophiles here and elsewhere, who can hear things like differences in wires with the same LCR, things in the music signal that microphones cannot (while listening to music...captured by microphones), etc, etc.?
Was the system used for the test "revealing" enough? Were "good" wires used to connect the compression drivers? Was the amplification "Mid-Fi", non NFB SS or better yet magical SET, with resolving power down to the sub-atomic level? Might any of this affected the 25% result?
Just curious. Thanks.
cheers,
AJ
But surely replacing these components would vary the distortion percentage but by a small fraction only. (eg, 20% - 30%). The fact of the matter is - if the same test conditions were used with only distortion introduced (by whatever means) into the same set of speakers and listeners couldn't tell, then that is significant in and of itself.
Of course I haven't read Earl's papers so I don't know what was setup and what was played.
Now awaiting SY's capacitor measurements.... (thanks SY!)
David.
AJinFLA said:
Or possibly they just didn't know what the piece of music sounded like without the introduced 25% (2nd harmonic) distortion. It's like when people hear a harshness in a (generally) 2-way and attribute it to the tweeter because it's in that range - whereas it's actually the cone-breakup from the mid-bass that hasn't been sufficiently reduced by the (audiophile approved) 1st order crossover.
The test used research grade insert earphones which are renown for there low distortion. The tests all used the same equipment in an AB comparison of non-distorted versus distorted music signals. The equipment would make the test uncertain at very low levels of comparison, but certainly not at the levels that we are talking about.
You can criticize the test all you want, but until there is data to the contrary that is perform with the same degree of care, the fact remains that all current data supports what I have said. And, as I have said my results have been corroborated elsewhere, like JBL.
No test is ever proof or perfect, but its foolish to discard the "best" data because you don't like the results. The best data that we have is all in agreement on these points. There is no scientific data to the contrary.
You can criticize the test all you want, but until there is data to the contrary that is perform with the same degree of care, the fact remains that all current data supports what I have said. And, as I have said my results have been corroborated elsewhere, like JBL.
No test is ever proof or perfect, but its foolish to discard the "best" data because you don't like the results. The best data that we have is all in agreement on these points. There is no scientific data to the contrary.
Earl, Dave, it will be tomorrow. Too many panics this afternoon. Got it set up, but no time to start running and saving spectra. 🙁
- Status
- Not open for further replies.