Non-harmonicly related distortion products can not exist, that is for two input frequencies the distortion products are only at A*F1 +- B*F2 where A and B are integers. I am speaking here of amplifiers and preamps, where this behavior is not observed, if you can present even one bona fide observation of this in capacitors hats off to you.
Last edited:
Regarding COG versus Polypropylene and Polystyrene this gives a lot of good information:
http://www.waynekirkwood.com/Images/pdf/Cyril_Bateman/Bateman_Notes_Cap_Sound_4.pdf
It also shows that it is not just the material used that counts. Also the physical implementation is important. And the DC bias applied!
And I think all the notes from 1 to 6 are worth reading.
This also seems to support the thought that COG are excellent capacitors, but Polystyrene are slightly better:
Low distortion oscillator tests measurement circuits | TMW
http://www.waynekirkwood.com/Images/pdf/Cyril_Bateman/Bateman_Notes_Cap_Sound_4.pdf
It also shows that it is not just the material used that counts. Also the physical implementation is important. And the DC bias applied!
And I think all the notes from 1 to 6 are worth reading.
This also seems to support the thought that COG are excellent capacitors, but Polystyrene are slightly better:
Low distortion oscillator tests measurement circuits | TMW
Well that really is just kind of the nature of things.
The same holds true for musical instruments.
Notice what happens with DC added to the ac thd..... increases of 20-30dB are common.... worse in electrolytics. Note too that asymetrical waveforms have a DC component.
I have measured thd with asymetrical waveforms and find with FFT that it was primarily 2H. Thx-RNMarsh
I have measured thd with asymetrical waveforms and find with FFT that it was primarily 2H. Thx-RNMarsh
Dick,
Doug Self maintains (and backs it up with measurements) that if you can keep the ac levels across an electronlytic below 70mV or so, the distortion is negligeable. That means that sometimes you need to use a much larger cap value that you would for the circuit time constants.
What's your take on that?
jan
Doug Self maintains (and backs it up with measurements) that if you can keep the ac levels across an electronlytic below 70mV or so, the distortion is negligeable. That means that sometimes you need to use a much larger cap value that you would for the circuit time constants.
What's your take on that?
jan
I am confused.
According to http://www.janascard.cz/PDF/An ultra low distortion oscillator with THD below -140 dB.pdf they are able to get -150db.
But according to Bateman the capacitors have a distortion closer to -130db.
Any one got ideas?
Btw I have looked and can't find Bateman's website.
His articles were around 2003 in electronic world
I do have them in pdf format that I can share if anyone wants them.
According to http://www.janascard.cz/PDF/An ultra low distortion oscillator with THD below -140 dB.pdf they are able to get -150db.
But according to Bateman the capacitors have a distortion closer to -130db.
Any one got ideas?
Btw I have looked and can't find Bateman's website.
His articles were around 2003 in electronic world
I do have them in pdf format that I can share if anyone wants them.
Some of the Cyril Bateman pdf's can be downloaded from here:
Capacitor Sounds, Speaker Cables and Crossover Inductors.
I have one of his CDROMs if anybody needs anything else.
Capacitor Sounds, Speaker Cables and Crossover Inductors.
I have one of his CDROMs if anybody needs anything else.
Last edited:
RE I am confused -- me too, but some of the issues here are the uncertainties of the measuring gear. These results are all near the residuals of the oscillators and analyzers. As such, the water is cloudy and not assuredly safe to drink.
This is likely all just measurement limitations. The Bateman setup is accurate to probably about -120 dB (it's some time since I've read his articles, so with reservations), so any comparison for lower distortion capacitors is meaningless. The Janásek article is, as pointed out several times already, pretty naive with quoting results at the -150 dB level. It could as well have been that the Polystyrene cap has higher distortion than the C0G, and that the higher distortion just happend to better cancel the contribution of the notch filter.
Samuel
I certaily would agree from the tests i have done recently.
Regarding ac level across caps as they were tested... Yes, keep the voltage drop across them as low as possible. This is easy to do in coupling cap apps. But in tone controls, EQ and filters, it doesnt work out quit so well. -Thx-RNMarsh
Microphones and phono catridges do not pass DC and an AC coupled signal in cluding music and these "asymetrical" waveforms do NOT contain DC. No tests that I have seen have any indication that the capacitor non-linearities can not be modeled in a simple way.
Besides most of even Bateman's results were very low order (in many cases seconds only) and still below -90dB. The audibility of this is debateable, the usual anecdotal evidence aside.
Semantics, i suppose. DC as in Direct Current -No. Average level that is DC (non-zero), yes. Not concerned any more about coupling cap's distortions and causes. We got rid got rid of the electrolytic coupling cap used in high-end audio (except Asian sourced recievers still use a lot of them). Then we got rid of the coupling cap altogether (in ss designs); DC servo took care of the major remaining issue of dc offset and drift issues... if they exist any more in better designs. But caps used for active or passive cross overs, filters, EQ and situations where a significant voltage is developed across them is of more concern now. [for the recording and playback chain]
Caps now need to be thoroughly tested/documented as they are still used... with voltage drops across them and with DC on them etc. Some tests show significant thd under these conditions.
Caps can still be sorted by audiophiles by their D.A for general selection and tests for THD generally hold up for such selection. The reasons for thd are many, but include the materials and construction methods and QA/QC etc. D.A. is just one issue of several for the designer to be aware of. All should be by now. All cap properties and parasitics are modeled as linear properties but the differences in thd are not modeled. So testing is best way for specifics.
Thx-RNMarsh
Last edited:
That's not semantics. Taking the average of a signal is a (crude) form of low-pass filtering. If the (long-term) average of a waveform is non-zero, it has DC content. Asymmetric music waveforms are not a result of DC/non-zero average, but of specific phase relations of the harmonics.
Samuel
Geeez. Look here... Put up a repetitive pulse... zero to + some level for some period. It has a dc level or average. no filtering involved. Put up a pulse with + 1 and neg -.5 for same periods... Average is + or non-zero. Lets not make this more compex than it needs to be. Asymetric test waveforms with a dc/average level that is not zero. Thx -RNM
Last edited:
Yep, geeez here too. Sorry, I don't get your point at all. Why do you want to test with a signal that has nonzero DC content if audio does not..? Perhaps Scott can help out...
Samuel
Samuel
Don't associate these test exercises to music. I disagreed with the examples Walt provided 30 years ago and I still disagree. A trumpet blast looks asymetrical on a scope, it DOES NOT contain a DC component, the time spent above 0 will always time average to 0 with the time spent below 0. You certainly can not add sine waves in any phase relationship to create a DC voltage. Neither of your test waveforms have physical meaning in real world sound propagation, both postulate a permanent offset in local barometric pressure.
No Samuel I can't help I'm afraid we must just move on this falls into the too much to lose category.
Last edited:
Waveform is chosen to bring out issues or imperfections from ideal. Which it does. You can debate it amoungst yourselves as to whether it is relavent.
Many people try to describe what second or third or even vs odd harmonics sounds like. But its best if you hear them for yourself. Same with capacitors 'sound'. Other's described the sound of electrolytics vs films and then film against film. Those descriptions did not seem to be caused by esr, esl or Fr nor THD to me. The test for DA and the way it was done matches exactly the descriptions for the sound and in rank order. But you can best understand this if you heard those caps for yourself. If you dont hear anything with any cap type then there is no issue to explore. My role was to find out what the listeners heard and try to correlate it with a characteristic and then a test.
But-- I'll let you have the last word on this old and tired subject. Good night.
Thx-RNMarsh
Many people try to describe what second or third or even vs odd harmonics sounds like. But its best if you hear them for yourself. Same with capacitors 'sound'. Other's described the sound of electrolytics vs films and then film against film. Those descriptions did not seem to be caused by esr, esl or Fr nor THD to me. The test for DA and the way it was done matches exactly the descriptions for the sound and in rank order. But you can best understand this if you heard those caps for yourself. If you dont hear anything with any cap type then there is no issue to explore. My role was to find out what the listeners heard and try to correlate it with a characteristic and then a test.
But-- I'll let you have the last word on this old and tired subject. Good night.
Thx-RNMarsh
Last edited:
At this point in cap measurement history, it is the voltage drop across the cap that I would now focus on. As I said before, there are many applications which develop a signal voltage across the cap -- such as, EQ, filters and oscillators. Using the lower DA films and better construction caps give lowest measured distortion in such use.
Thx-RNMarsh
Thx-RNMarsh
Last edited:
It is true that acoustically it is impossible for sound waves to produce a DC average.
The pressure wave will always return to zero.
But it is not true of electrical signals.
A series of positive going pulses will produce an average DC signal.
It is just the difference between electronics and acoustics.
The pressure wave will always return to zero.
But it is not true of electrical signals.
A series of positive going pulses will produce an average DC signal.
It is just the difference between electronics and acoustics.