Hi Bob! I have been reading and enjoying your stuff for many years! In this case (see "What's All This C-R Stuff, Anyhow?" Sept. 1, 2006, p. 18), I would say that capacitance has varied, not with frequency, but instead with applied potential. Various materials, especially ceramics, vary capacitance with applied voltage. (Yeah, but only a few percent... /rap) If we give cap time to charge up some, its capacitance can vary dynamically. (There may be some materials for which that is true, but most useful capacitors are fairly linear. I have taken some of these caps, with poor response and poor soakage, and charged them up to a lower voltage, and the shapes are just the same, scaled down. No kidding. Can I run them up to higher voltages? I'll try. I charged up a paper-and-oil capacitor to 10 V, shorted it out for a short time, and let go. The voltage went up to 2 V. There must be a lot of dipoles twisting around in that oil! If I charged it up to +100 V, could it charge to more than 20 V? It couldn't go much higher, or it would face some laws of conservation of energy. How about 1000 V? Can you name a material for which the errors scale differently? Most ceramics have minimal voltage coefficients. /rap) Above the R-C rolloff, the voltage across the cap does not vary, because the signal is being conducted across without much charging. (We agree. /rap) And way below rolloff, the signal does not get through at all, so we only see the change at rolloff. But we should be able to characterize it by measuring capacitance (perhaps with tiny sine waves) while varying a much larger dc bias. (Many capacitors will not change much with dc bias, but maybe some will. I'll have to check it. Thanks for raising the question. /rap)
Pease: I just tested every capacitor in my lab by putting a 20-V p-p square wave across them at 120 Hz. Even the ceramics were silent! Every one was silent. But there was an "Audiocap" that emitted some noise. You'd think that a $20 capacitor wouldn't have a flaw like that.
Hello Bob: Your comments regarding two audio circuits not sounding the same despite measurements (see "What's All This C-R Stuff, Anyhow?" again) means there are in fact no measurements, the measurements are wrong, or the published measurements are deliberately misleading. (Or maybe the guy doing the measurements doesn't know where to look. I mean, if you just measure "distortion" but you don't check to see what kind of distortion, that number doesn't mean much. /rap) If in real estate it's location location location, in electronic circuit performance, it's layout layout layout. (We sure agree on that. /rap) Locate sensitive components away from noisy components. Keep digital the hell away from analog inputs. Locate components to keep lead lengths short. Loops will emit and receive electromagnetic interference. Minimize their areas. Poorly laid-out grounds and supply leads have lots of impedance that causes all kinds of problems like noise, distortion, and instability. In one power amplifier design, I found that varying the spot on the chassis where I attached my pc board's ground varied my output noise by over 30 dB! A 90-dB signal to noise sounds good on a power amp, but 120 dB sounds better. Not many commercial designs take this kind of care. I remember the RCA studio tube amplifiers. You could get the same exact schematic on one chassis or on two separate chassis (one for the power supply, the other for the amplifier). The two-chassis version had a significantly better noise spec. What I've learned about audio almost 40 years ago is very straightforward, though radical to the average audiophile. (I always move my power supply as far away as I can from the amplifier—at least from the front end. /rap) Those fancy, expensive speaker cables are a waste of money. Another engineers' lie! 20 feet of standard speaker cable feeding a conventional speaker gave me a ?3-dB point of 8 MHz measured! (The speaker must have been very inductive, because normally, the inductance of 20 feet of lamp cord is excessive compared to 8 Ω, even at 300 kHz. /rap) This one is very audible on a good signal source. The speaker cables' price doesn't directly relate to performance, but the performance is readily measured. Save your money by buying inexpensive lamp cord, and spend the money on good speakers that sound good to you.
John M. Cook
Pease: Thanks for the comments.
Dear Bob: I thought the guys with the golden ears couldn't stand to listen to any amplifier other than a single-ended vacuum triode with a non-thoriated direct-heated filament cathode, preferably with no capacitors or transformers, and no feedback, playing vinyl discs—and don't forget the $1500 oxygen-free water-shielded litz speaker wires.
Pease: Actually, my friends in the know say that electric heating of the cathodes is obsolete. Coal-fired pentodes take a little more time to warm up, but they sound warmer. And, do you know how many 6AS7GAs it takes, to just drive the grids of all the 6AS7GAs you need, to drive from their cathodes into the 4-Ω speaker? Lots and lots!