------------------------------ Date: Thu, 10 Jun 93 11:00:11 EDT From: jas@proteon.com (John A. Shriver) Subject: "Bypass" Capacitors Capacitors generally have an impedance versus frequency curve that goes down for a while, hits a minimum, and then goes back up again quite fast. Inductance, leakage, dielectric absorbtion, etc., catch up as the frequency increases. This information is on any properly detailed capacitor data sheet. (They can be as long as an IC data sheet.) For an electrolytic capacitor, that knee can well be at an audible frequency. For a film capacitor, that knee will be at a much higher frequency. By paralleling the two, you get a compound unit whose impedance remains low over a larger frequency range. It behaves more like an ideal capacitor. There are, however, some problems with it, it's not an ideal capacitor, and the two halves can resonate between each other. (This is mostly a problem with with bypassing electrolytic capacitors, which are very crude approximations of ideal capacitors in the first place.) As for 10 uFD capacitors, even a film one is going to be less than ideal. To keep the bulk down, it will be one of the more volumetrically efficient dielectrics, which are not as good. It will also be metallized film, which again is not as good. Thus, one does see bypassing of 10 uFD film capacitors. ------------------------------ From: u89cdw@robots.ox.ac.uk (Chas Winder) Subject: Re: "Bypass" Capacitors Date: Thu, 10 Jun 1993 15:39:59 GMT In article <1v7b4hINNdbp@uwm.edu> "James A. Cobbs" writes: # As I understand it: if you are using a 10 mfd (or larger) you can add a # small value (.01 to .10 mfd) cap in PARALLEL to act as a 'bypass.' Since # nothing (except leakage) goes thru a capacitor what does a 'bypass' do. The problem with `largeish' capacitors, particularly electrolytic types, is that they often exhibit a reasonable amount of series inductance. Hence at high frequency such a capacitor presents a noticable impedance to the signal, i.e. not bypassing it very well, whereas at moderate frequencies the impedance presented to the signal is less, so there isn't a problem then. `Small' capactors (polyester, mica and mylar types, for example) have much less series inductance, and so still manage to pass high frequency signals effectively. Their impedance at lower frequencies is still quite large, but this doesn't matter as the the `largeish' capactor can deal with the lower frequencies. All you need to do is connect the two frequency-dependant impedances together: wire the capacitors in parallel. Different types of capacitors affect the signal in different ways: more expensive types (like silver mica or polystyrene) will probably sound nicer than cheapie ceramic ones. -- +-----------------------------+-----------------------------------------------+ | Chas Winder, | Dept. Engineering Science, Oxford University, | | u89cdw@robots.ox.ac.uk | Parks Road, Oxford, UK. OX1 3PJ. | +-----------------------------+-----------------------------------------------+ ------------------------------ From: spaaejg@ucl.ac.uk (Eric Glover) Subject: Re: "Bypass" Capacitors Date: Thu, 10 Jun 1993 16:21:15 GMT "James A. Cobbs" writes: > OK here is one for you techie gurus (I'm not an EE, or I wouldn't >have to ask this question, so keep it simple): I see in high-end speaker >reviews, and I've had recommended to me, the use of 'bypass' capacitors. The general answer is simple: large capacitors (i.e.electrolytics) tend to have inductance owing to their foil construction. This attenuates hf, so bypassing with a smaller low inductance cap will shunt this stuff away, thus turning the arrangement into something more nearly resembling a perfect capacitor. Eric Glover ------------------------------ From: paul@gaitlab1.uwaterloo.ca (Paul Guy) Subject: Re: "Bypass" Capacitors Date: Sun, 13 Jun 1993 01:41:11 GMT In article <1v7b4hINNdbp@uwm.edu> "James A. Cobbs" writes: > OK here is one for you techie gurus (I'm not an EE, or I wouldn't >have to ask this question, so keep it simple): I see in high-end speaker >reviews, and I've had recommended to me, the use of 'bypass' capacitors. > >As I understand it: if you are using a 10 mfd (or larger) you can add a >small value (.01 to .10 mfd) cap in PARALLEL to act as a 'bypass.' Since >nothing (except leakage) goes thru a capacitor what does a 'bypass' do. >The only thing that enters my feeble brain is that it could act as a >aid in maintaining the main capacitor during load changes etc.; but this >whole thing could be a conspiricy by cap manufacturers to unload their >surplus inventory of small value caps. > As much as I might like to subscribe to the 'conspiracy' theory, it's a legitimate technique to bypass this way. Imagine you have a two terminal device, a black box if you wish, that could be a power supply, another capacitor, resistor or whatever. In many cases you want that two terminal 'thingy' to have a very low impedance at some range of frequencies. Unfortunately, the inner workings of this device sometimes won't do that. So, you add a capacitor in parallel with this 'thingy', and you choose values that guarantee low impedance over the range you want. It's much like the way a traffic expressway bypasses the slower downtown areas. Some large capacitors (especially ones that were not designed for your application) may show rising impedances at high frequencies due to inductive effects of the concentric foil wrap, or may have significant resistance due to the thinness of the foil. Suppose your capacitor (the big one) shows an impedance which drops with frequency, but at a certain point, turns around, and the impedance then goes UP with frequency. Not a proper behaviour! At this point, the capacitor is not really behaving as such. So, you put another one in parallel, to take over the job at high frequencies. This one you choose for high frequency performance. Usually the big one is behaving so poorly at the turnaround, that a value of 1% to 10% will take over its function. Ultimately, the second one will start to misbehave at much higher frequencies, and then you can repeat the trick. Electrolytic capacitors have a broad turn-around frequency that's at about 20kHz to 100kHz. Film type capacitors are ok to about 1-5 MHz, and then the size of them causes too much inductance. At that point, you use ceramic or mica disk capacitors, that can be good up several hundred MHz. Beyond that, it gets real weird, as so many factors come into play, that you need a lot of experience. It may seem silly to design an amp, with circuitry that goes WAY beyond the audible range, but some of the sub-components in that amp, especially amplifying devices, may respond up to several hundred MHz. With unbypassed components, it MIGHT oscillate, or do some perverse behaviour that has a large secondary effect on the audio performance. I remember an early power amp that I built, that seemed to run real hot, with a little more distortion than it should. Turned out, that it was REALLY a 200 watt 500kHz oscillator, the audio amplification was just a side effect. A well designed amp, or any device should not need tweaking, if they put put reasonable effort into its manufacture. The sad thing is that many reputable pieces of equipment can be significantly improved by fiddling ( at least if you do it in an intelligent manner). -Paul -- ----------------------------------------------------------------------------- Paul J Guy work phone:519-885-1211 ext 6371 paul@gaitlab1.waterloo.edu home/FAX/message:519-576-3090 pguy@healthy.waterloo.edu ..remember...bullshit baffles brains... ------------------------------ ------------------------------ From: rutgers@prl.philips.nl (Herbert Rutgers) Subject: Re: "Bypass" Capacitors Date: Mon, 14 Jun 1993 13:15:22 GMT >effectively. Their impedance at lower frequencies is still quite large, but >this doesn't matter as the the `largeish' capactor can deal with the >lower frequencies. All you need to do is connect the two frequency-dependant >impedances together: wire the capacitors in parallel. Be careful, the series inductance of the `largeish' (as you call it) can resonate with the smaller capacitor. You get a parallel resonator. The impedance of a parallel resonator has a nearly *infinite* impedance at the resonance frequency, and a lot of phase shift as well (already far from the resonance frequency). Herbert. ------------------------------ Date: Tue, 15 Jun 1993 15:50:15 -0400 (EDT) From: Richard D Pierce Subject: re: "Bypass" Capacitors "Herbert" writes: > >>effectively. Their impedance at lower frequencies is still quite large, but >>this doesn't matter as the the `largeish' capactor can deal with the >>lower frequencies. All you need to do is connect the two >frequency-dependant impedances together: wire the capacitors in parallel. > >Be careful, the series inductance of the `largeish' (as you call it) can >resonate with the smaller capacitor. You get a parallel resonator. The >impedance of a parallel resonator has a nearly *infinite* impedance at the >resonance frequency, and a lot of phase shift as well (already far from the >resonance frequency). Well, no, not quite correct, Herbert. First, this inductance is extremely lossy. Secondly, there's more than enough loss both series and parallel in tha caps. Altogether, these would quite adequately conspire to prevent the Q of the circuit at resonance from ever reaching infinity. Also, the MAXIMUM phase shift such a circuit could have is not "lots", but 180 degrees, plus or minus. And guess how much phase shift there would be at resonance? Exactly 0 degrees. In actual practice, if such a resonance exists, it is of such low Q because of the losses that it is swamped by other effects to the point of being rendered insiginifcant as far as audio applications are concerned. | Dick Pierce | | Loudspeaker and Software Consulting | | 17 Sartelle Street Pepperell, MA 01463 | | (508) 433-9183 (Voice and FAX) | ------------------------------ From: rutgers@prl.philips.nl (Herbert Rutgers) Subject: Re: "Bypass" Capacitors Date: Thu, 17 Jun 1993 12:28:08 GMT Richard, >Also, the MAXIMUM phase shift such a circuit could have is not "lots", but >180 degrees, plus or minus. And guess how much phase shift there would be >at resonance? Exactly 0 degrees. I do not have to guess for that... I simply know. Possibly "lots of" is the wrong term in English, sorry, but 180 degrees is a lot, I think.... >In actual practice, if such a resonance exists, it is of such low Q >because of the losses that it is swamped by other effects to the point of >being rendered insiginifcant as far as audio applications are concerned. I have other experiances at frequencies of about 1 MHz (indeed). Let me say it in another way: You do have the chance (so you have to measure!!) that the impedance of the capacitor in parallel with an elco, shows a (much) larger impedance than only the elco (at certain frequencies). And that is not what you should expect in advance. A small resistor in series with the 'small' capacitor (to lower the Q) sometimes gives better results. Cascading, as said in an answer, is always bad. The Q of an slco is rather bad indeed, but the smaller capacitors do have a quite high Q. Herbert. ------------------------------ Date: Fri, 18 Jun 1993 12:43:26 -0400 (EDT) From: Richard D Pierce Subject: Phase shift hooey, was re: Bypass Capacitors Herbert continues: >>Also, the MAXIMUM phase shift such a circuit could have is not "lots", but >>180 degrees, plus or minus. And guess how much phase shift there would be >>at resonance? Exactly 0 degrees. > >I do not have to guess for that... I simply know. Possibly "lots of" is the >wrong term in English, sorry, but 180 degrees is a lot, I think.... > So what. Let me give you an example of a "lot" of phase shift that has no demonstrable audible consequences: the phase shift at 10 kHz of most speakers under typicial measuring situations is in the realm of 10,000 degrees. Now, before you read further, I should ask you to answer the following to questions: 1. is 10,000 degrees of phase shift "a lot". 2. is 10,000 degrees of phase shift at 10 kHz important or not? Now the answer to question 1 is, yes, it's a lot. The answer to question two, which is the same answer to your comment about 180 degrees of phase shift being "a lot" is the same: any answer is pretty meaningless without come context. The 10,000 degrees of phase shift here is a direct consequency of the 2.94 mSec acoustical delay that exists becuase under standard measurement condition, the measuring microphone is placed 1 meter from the transducer. So in this context, we have a phase shift which increases monotonically with frequency at a rate which is soley dependent upon the distance between the speaker and the microphone (our our ears) and the propogation velocity of sound. Thus, in THIS context, such rodiculously huge amounts of phase are unimportant. Your statement about 180 degrees of phase being "a lot", stated without context and without any further quilification or explanatation, is essentially meaningless. That's my point >>In actual practice, if such a resonance exists, it is of such low Q >>because of the losses that it is swamped by other effects to the point of >>being rendered insiginifcant as far as audio applications are concerned. > >I have other experiances at frequencies of about 1 MHz (indeed). Kindly refer to my last sentence before your comment: "insignificant as far as audio applications are concerned." When does 1 MHz constitute audio? There is no denying the effects you describe might exist to some extent at RF frequencies (but I challenge you to show us on of those "infinite impedance" parallel resonances you talked about). At audio frequencies, I have never seen such effects from the self inductance of capacitors (nor the distributed self capacitance of inductros, for that matter). Sit down and model the equivalent circuit at audio frequencies correctly and you'll find that a: such effects do not exist as you describe and b: the model quite accurately predicts the measured characteristics of real components. I do it all the time, saves a hell of a lot of my clients money by keeping me away from the soldering iron. >Let me >say it in another way: You do have the chance (so you have to measure!!) >that the impedance of the capacitor in parallel with an elco, shows a >(much) larger impedance than only the elco (at certain frequencies). And >that is not what you should expect in advance. A small resistor in series >with the 'small' capacitor (to lower the Q) sometimes gives better >results. Show us real results measured with real components, and show us results done with a reasonable model. Otherwise, your theory and data conflict fairly strongly with practice. Remember, you claimed a resonance with an impedance at resonance of infinity. I will accept a resonance of only 1000 times that of that predicted by a capacitor-only model as equivalent to "infinite". My prediction (based on my own measurement of these effects) is that you'll be lucky to hit 10 times. | Dick Pierce | | Loudspeaker and Software Consulting | | 17 Sartelle Street Pepperell, MA 01463 | | (508) 433-9183 (Voice and FAX) | ------------------------------ Date: Wed, 2 Feb 1994 13:24:00 +0000 From: "henry (h.) pasternack" Subject: Bypass capacitors. >What are the best capacitors that you have used for shunting, dc >blocking, power supply use? There are types of caps that may be >better in different situations. I'd like to put in the good word for NPO and C0G ceramic caps. These are low-drift dielectrics that are reportedly pretty good for audio, albeit unfashionable (Erno Borbeley). I wouldn't use them for signal, but they are superb for decoupling solid-state power supply rails. A ceramic cap has much lower inductance than a poly type. Digi-Key has a decent selection of NPO and C0G ceramics at voltages up to 100V. I'm pretty skeptical about a lot of audiophile capacitor claims. If you look at the IF strip of an FM receiver, you're not going to see polypropylene RF bypass capacitors. If you compute the reactance of, say, an ideal 0.1uF capacitor at 20kHz, you'll discover it's about 80 Ohms. Is it really going to help the audio-band performance of an electrolytic to bypass it with a small poly cap? If the benefit is at RF frequencies, why not use a proper RF cap? Can someone provide REAL DATA on the costs and benefits of using parallel combinations of "audiophile" capacitors? I'd really like to see a good debunking of this subject. -Henry ------------------------------ From: brad.sanders@circellar.com Date: Fri, 04 Feb 94 06:37:10 -0500 [EST] Subject: STUFF To: info-high-audio@introl.introl.com response to: "henry (h.) pasternack" Subject: Bypass capacitors. | | I'm pretty skeptical about a lot of audiophile capacitor claims. |If you look at the IF strip of an FM receiver, you're not going to |see polypropylene RF bypass capacitors. If you compute the reactance |of, say, an ideal 0.1uF capacitor at 20kHz, you'll discover it's about |80 Ohms. Is it really going to help the audio-band performance of an |electrolytic to bypass it with a small poly cap? If the benefit is |at RF frequencies, why not use a proper RF cap? | | Can someone provide REAL DATA on the costs and benefits of using |parallel combinations of "audiophile" capacitors? I'd really like |to see a good debunking of this subject. |____________________________________________________________________ As a tube nut, I used to think the mess about different wires was also just so much fashionable dinner conversation. See the next message from me about how I feel now.... However, Walt Jung convinced me, in very scientific fashion, about the differences in caps back in the 70's. Go check out his series in AUDIO magazine, back around '76-'77. He also had a very good series on op-amps in the early '80s. The reason you don't see poly caps in IF sections is because they are not hi-Q devices at RF frequencies (too much inductance) - just like an electro at audio frequencies. There's no black magic - it all has a very real basis in electronics. Now, so far as "Wonder Caps"? This is like Vishay resistors - at the prices these folks charge, they have a LOT of proving to do on my part. I can get Caddocks for a couple bucks, and VERY high quality caps from a number of sources at REASONABLE prices.... nuke ___ Blue Wave/QWK v2.11 ------------------------------ Date: Fri, 4 Feb 94 09:28:41 -0500 From: Scott Dorsey Subject: Bypass capacitors > I'd like to put in the good word for NPO and C0G ceramic caps. > These are low-drift dielectrics that are reportedly pretty good > for audio, albeit unfashionable (Erno Borbeley). I wouldn't use > them for signal, but they are superb for decoupling solid-state > power supply rails. A ceramic cap has much lower inductance than > a poly type. Digi-Key has a decent selection of NPO and C0G ceramics > at voltages up to 100V. The ceramic caps have a significant change in value with temperature, even the NPO types. This can be a real problem in a tube amplifier which is going to be running significantly above room temperature. On the other hand, for supply decoupling, this isn't a big deal. The ceramic caps also have some piezoelectric distortion problems... the dielectric deforms given a short pulse, and when it returns to its normal shape, it produces an opposing polarity (but much smaller) pulse on the leads. I don't know how significant this really is, and for a supply decoupling application, it's probably no big deal anyway. I strongly recommend glass capacitors for use in the audio signal path where small values with a good temperature coefficient are needed. For supply decoupling, though, they'd be complete overkill. The problem with ceramics on the supply is that they aren't big enough... I use a 1 MFD film plus a .1 MFD film to bypass all the bigger supply caps. Then again, I get the film caps cheap in bulk, so it costs me less than the ceramics would. It couldn't hurt to add some ceramics too. --scott ------------------------------ Date: Fri, 4 Feb 1994 13:06:00 +0000 From: "henry (h.) pasternack" Subject: Bypass capacitors. Brad Sanders writes: :As a tube nut, I used to think the mess about different wires :was also just so much fashionable dinner conversation. See the :next message from me about how I feel now.... Of course different cables sound different, because different cables have different physical and electrical parameters. :However, Walt Jung convinced me, in very scientific fashion, about :the differences in caps back in the 70's. Go check out his series :in AUDIO magazine, back around '76-'77. He also had a very good :series on op-amps in the early '80s. I read the series. Good example of HARD DATA. :The reason you don't see poly caps in IF sections is because they :are not hi-Q devices at RF frequencies (too much inductance) - just :like an electro at audio frequencies. There's no black magic - it :all has a very real basis in electronics. Parasitic inductance does not lower the Q of a capacitor. It increases the reactance and makes it ring. However, my point was essentially as you have stated. A 0.1uF poly capacitor, commonly recommended as a power supply filter bypass, has too much inductance to be an effective RF bypass, and too little capacitance to make a difference at audio frequencies. But it does have high Q. Put a few of these caps in parallel, and you run the risk of inducing ringing on the power supply rails. A liberal arts major whose major understanding of electronics comes from Horowitz and Hill is likely oblivious to these effects, which are not predicted by zeroeth-order models. I don't have the equipment to measure the performance of parallel combinations of audiophile caps. That's why I'd like some cerdible HARD DATA to help me decide whether these parts are worth using and how best to apply them. -Henry ------------------------------ Date: Tue, 8 Feb 1994 14:03:00 +0000 From: "henry (h.) pasternack" Subject: Bypass capacitors. Brad Sanders writes: : I present your original message here in quotes... : : "h> I wasn't referring to you when I made the comments about : "h> audiophiles pissing on engineers... the note I sent to you : "h> is also being posted to the bulletin board. : : Don't recall reading the comment about audiophiles pissing on : engineers. The comment about audiophiles pissing on engineers is something I wrote, but later deleted, in a posting that ultimately appeared Monday morning on 'rec.audio.high-end' and which I also emailed to you. I couldn't remember if I changed it or not. The text you quote now is part of a second personal email message that I never sent to the bulletin board. Happily for me, I didn't say anything so stupid that I am ashamed to see it in public. But I might have been a bit more careful if I knew my message would ultimately end up on the bulletin board. : "h> Does parasitic inductance really lower capacitor Q? I was kind of beat when I wrote to you. Thinking it over, Q in an LCR network is defined by reactance at resonance divided by resistance. Adding inductance lowers the resonant frequency. At resonance, inductive and capacitive reactance are equal. Since the reactance of a capacitor goes up at lower frequencies, adding inductance to an LCR circuit raises the Q. The other way to see it is that adding inductance increases the energy storage without increasing the losses, hence, higher Q. Just wanted to get that straightened out for the record. : You're not looking at the whole story. You said you had read the Jung : articles in AUDIO. If you then have these handy (or even if not), I suggest : you go through them again. Pay particular attention to the properties : of different dielectrics. This is often _more_ important than you seem to : think. And yes, there are measurable differences, two of which are DF and : DA - which Jung discusses in detail. This is why electrolytics can be heard. : I suggest you prove it to yourself - try it! I read the Jung articles at the Terman Engineering Library at Stanford. Wish I were there now. Since it's been about ten years, I don't remember all of the details. I agree that the difference in dielectrics is the key issue, else there wouldn't be much to argue about, except parasitic inductance due to construction. What I wonder is whether the addition of a small amount of quality capacitance can compensate for an overwhelming amount of non-quality capacitance. My intuition says the improvement offered by a bypass capacitor is a function of the relative reactances of the two caps in questions. If the electrolytic looks like, say, one Ohm at 20kHz and the film cap looks like 150 Ohms, I can't see that the superior DA and DF of the film cap will have a significant effect. I do, though, have an open mind. At this point, it would be nice if someone could make some before/after DA and DF measurements of bypassed and unbypassed capacitors. If a significant difference is found, we have good engineering evidence that bypassing is effective. If not, we have to fall back on subjective arguments, with all of the associated uncertainty and bias. : NPO's do not work well in analog circuits for the same reason films do not : work well in digital circuits - they are simply designed for two different : things! Again, to make this very clear, I think there are places where an NPO cap is a good choice in an analog circuit. Specifically, power supply rails of solid-state equipment where low RF impedance is critical. In this application, I would argue, the benefit of a high-quality RF bypass out- weighs the small penalty of dielectric nonlinearity. : In my tube amp, changing wire types makes a great difference - even on wires : suspended in space by the solder connection at each end - but I can't say the : differences would be so apparent in transistor designs, where the signal path : impedances are several orders of magnitude lower. I kind of wish Dick Pierce would step in now, because he is committed to challenging these kinds of statements. I'll try to be less severe than Dick might be in my response. Let me say that I am uncomfortable with this assertion because there exists absolutely no technical framework in which to explain the claimed difference. On the other hand, there is a considerable probability of subjective bias. : [Defending opinions as facts is] a big business actually. See, I "suggest" : to you that in fact there IS a HUGE difference between wire A and wire B. : So, you go "upgrade" all your stuff to wire B. Now, whether in fact one DOES : hear a difference or not will be biased not only by ones desire to hear a : difference (in my opinion, nothing wrong with this), but also by the NEED to : hear a difference, planted there by someone with a purely economic interest : (I have a lot of problems with THIS!) Pretty much exactly my feelings. However, I think the desire to hear a difference is a big problem because it strongly biases the listener. Not always, mind you, but certainly sometimes. : You may be appalled by this, but talk to any of the better analog engineers : - the old hands, the respected genius' (geni?) - and they too will tell you, : analog design is not always "by the book." Again, my sentiment exactly. And, no, I'm not in the least appalled. But the difference between a a good technician and a good engineer is that the technican can tell you HOW, but the engineer can tell you HOW and WHY. I'd like to think that the old-timers, the really experienced engineers, can go back to the textbooks and show you what first-order assumptions need to be extended to accommodate "audiophile" phenomena. Their intuition has a solid, factual, technical foundation, something that cannot be said for the audiophile "magicians." : This is one of the laments of the modern electronics industry, it's why the : truly excellent old hands are getting higher and higher salaries....because : with all the reliance on SPICE, DSP, VERILOG and other junk of that ilk, : there are no "new, young, promising talents..." The only way the younger : engineers know to design is based upon pure canned knowledge. Not much room : to "grow" from that, is there? You'll have no objection from me. I think SPICE is great, but I don't like to use simulators unless I write them myself. How can I claim credit for a design that was cranked out by a computer? I have a major beef with the morons on 'sci.electronics' who argue "bits is bits" and are not even dimly aware of what they don't know. But let's not go too far in the other direction and so romanticize the art of electronics that we forget the strong technical basis of circuit design. : "h> How to express my point? The fact that most engineers are idiots does : "h> not mean that a superior engineer cannot design superior equipment. : "h> Audiophiles, by virtue of not being engineers, are not qualified to : "h> decide what the limits of engineering really are. This anti-engineering : "h> sentiment is really a load of bunk, in my opinion. : : You're being pretty hard on BOTH camps here, aren't you? It's kinda hard : to decipher your point on this. Frankly, this sounds like the point of : view of an establishment dogmatist. Well, if my faith in engineering makes me an establishment dogmatist, so be it. As an "advanced intermediate" amateur, I am continually amazed to discover how much "audiophile wisdom" is just repackaged technical knowledge that has been with us for decades. : Lemme tell you a story. I'll TRY real hard to be brief.... : : [Truncated story deleted.] Just a bit too brief, I guess. I'd like to hear the end of your story. -Henry ------------------------------ From: caudle@hanszen.rice.edu (Chris Aaron Caudle) Subject: Re: Bypass capacitors. Date: Wed, 9 Feb 1994 07:06:52 GMT henry (h.) pasternack (hap@bnr.ca) wrote: : >What are the best capacitors that you have used for shunting, dc : >blocking, power supply use? There are types of caps that may be : >better in different situations. [stuff deleted] : I'm pretty skeptical about a lot of audiophile capacitor claims. : If you look at the IF strip of an FM receiver, you're not going to : see polypropylene RF bypass capacitors. If you compute the reactance : of, say, an ideal 0.1uF capacitor at 20kHz, you'll discover it's about : 80 Ohms. Is it really going to help the audio-band performance of an : electrolytic to bypass it with a small poly cap? If the benefit is : at RF frequencies, why not use a proper RF cap? : Can someone provide REAL DATA on the costs and benefits of using : parallel combinations of "audiophile" capacitors? I'd really like : to see a good debunking of this subject. : -Henry Read Jung's article in 1980 Audio. A very good treatment of capacitors, as well as a couple of test setups to measure the effects discussed. Chris Caudle caudle@owlnet.rice.edu ------------------------------ From: jhas@sh.alcbel.be (J. Haspeslagh SH11 4220) Subject: Re: Bypass capacitors. Date: 9 Feb 94 16:37:00 GMT In article <2jas8c$1v8a@introl.introl.com>, hap@bnr.ca (henry (h.) pasternack) writes: |> |>: This is one of the laments of the modern electronics industry, it's why the |>: truly excellent old hands are getting higher and higher salaries....because |>: with all the reliance on SPICE, DSP, VERILOG and other junk of that ilk, |>: there are no "new, young, promising talents..." The only way the younger |>: engineers know to design is based upon pure canned knowledge. Not much room |>: to "grow" from that, is there? |> |> You'll have no objection from me. I think SPICE is great, but I don't |>like to use simulators unless I write them myself. How can I claim credit |>for a design that was cranked out by a computer? I have a major beef with |>the morons on 'sci.electronics' who argue "bits is bits" and are not even |>dimly aware of what they don't know. But let's not go too far in the other |>direction and so romanticize the art of electronics that we forget the strong |>technical basis of circuit design. "A design cranked out by a computer". This sounds as ridiculous as many of the "audiophile talks" about golden cd's, green pens ... In the good old days designs were thought, made on breadboards and tested ... Now designs are thought, simulated, processed and tested. There is no simulator which creates a piece of hardware. It is what the designer puts in, what comes out. Therefore only GOOD designers give good outputs. And good designers do need the same knowlegde as the "elder" ones. I would like to see the design process of CD-player, DAT components, A/D converters done without simulators. I can be done but no one would be able to afford a CD because of the man-years of design. To be audiophile and engineer is not exclusive, as long as if you know what you 're talking about. But some people tend more to BELIEVE then to investigate. If someone tells I experienced a dozen of times that a DAT recording of LP sounds the same as the LP and that no LP recording of a DAT sounds the same, then still people stand up and tell LP is the best. Why because it sounds better, OK then buy a midfi system with a DSP mode "LP-sound" and you're OK. But if this feature would be available on the market, no one would like it because it is labelled as MIDFI and would be much cheaper then a high-end turntable. BTW I listen to LP's because I listen to the music and the reproduction is OK, meaning I do not perceive any disturbing effects. I know also that CD's give a more faithfull reproduction of the original performance. The only question which arises, if people tend to match a more natural perception to the LP then to the CD, that maybe the recording process like it is now (flat frequency, miking techniques) is not the optimal way of recording. The question is, are the normal miking techniques sufficient to catch all (enough) effects of the performance ? That is also the reason why there are so many discussion about front firing speakers, bose speakers, electrostatic speakers... But none on the net discusses or (maybe knows) the reasons behind the different perceptions all these speakers give. Do f.e. electrostatic speakers sound the same as cone speakers with the same amplitude an phase characteristics in an anechoic room ? If someone answers this questions then this would I think help us a lot instead of the endless discussion now. Johan -- --------------------------------------------------------------- Johan Haspeslagh - Dept SH11 - Alcatel Bell Telephone, Antwerp - tel 32/3/ 240 4220 -fax 32/3/2409947 ------------------------------ From: rpgroschen@mmm.com Subject: Bypass Capacitors Date: Wed, 09 Feb 94 12:10:10 PST :> I'd like to put in the good word for NPO and C0G ceramic caps. :> These are low-drift dielectrics that are reportedly pretty good :> for audio, albeit unfashionable (Erno Borbeley). I wouldn't use :> them for signal, but they are superb for decoupling solid-state :> power supply rails. A ceramic cap has much lower inductance than :> a poly type. Digi-Key has a decent selection of NPO and C0G ceramics :> at voltages up to 100V. Scott Dorsey :The ceramic caps have a significant change in value with temperature, :even the NPO types. This can be a real problem in a tube amplifier :which is going to be running significantly above room temperature. On :the other hand, for supply decoupling, this isn't a big deal. I doubt it, that's why they're called NP0. One important point that seems to be getting lost in this discussion on ceramic caps is that there are *two* construction techniques used: The single layer ceramic and the multi-layer. The MLC's are the one that exhibit most of the 'evil' characteristics that people complain about, and for good reason; these caps are the most mis-used and mis-applied tool since the screwdriver! The single layer caps, while exhibiting a small amount of piezo effect, are much better for bypass, and are used extensively as coupling caps in RF work, just look in your old transistor radio or your TV tuner. Speaking of tuners, they're also stable enough to use in the tuned circuits in those critters, too, and that tube-type gear! The big "problem" (if you want to call it that) with single layer ceramics, is their size. Because there is only two sides to the cap that are silvered, you can't get much capacitance without getting BIG. I've seen these things get to be the size of a quarter to get .1 uf at 500 WVDC. :The ceramic caps also have some piezoelectric distortion problems... the :dielectric deforms given a short pulse, and when it returns to its normal :shape, it produces an opposing polarity (but much smaller) pulse on the :leads. I don't know how significant this really is, and for a supply :decoupling application, it's probably no big deal anyway. Remember, its the MLCs that are the pesky ones! :I strongly recommend glass capacitors for use in the audio signal path :where small values with a good temperature coefficient are needed. For :supply decoupling, though, they'd be complete overkill. Glass is good, but not the only choise here. There's lots of good stuff out there, just peel open your eyeballs........... :The problem with ceramics on the supply is that they aren't big enough... :I use a 1 MFD film plus a .1 MFD film to bypass all the bigger supply :caps. Then again, I get the film caps cheap in bulk, so it costs me less :than the ceramics would. It couldn't hurt to add some ceramics too. :--scott I've got 1 uf, 5 uf, and even 10 uf MLC's in the parts drawer right now. They're not big, (not cheap either!) and work fine for bypass, but my heart is on the tantalums............. Standard disclaimers apply here.................. ------------------------------ From: brad.sanders@circellar.com Subject: Bypass capacitors. Date: Wed, 09 Feb 94 17:16:25 -0500 [EST] Brad Sanders (me) wrote: : I present your original message here in quotes... : Don't recall reading the comment about audiophiles pissing on : engineers. To wit: : The comment about audiophiles pissing on engineers is something I :wrote, but later deleted, in a posting that ultimately appeared Monday :morning on 'rec.audio.high-end' and which I also emailed to you. I :couldn't remember if I changed it or not. The text you quote now is :part of a second personal email message that I never sent to the :bulletin board. Happily for me, I didn't say anything so stupid that :I am ashamed to see it in public. But I might have been a bit more :careful if I knew my message would ultimately end up on the bulletin :board. And I'm not sending you this personally either, Henry - because I think the apology should be public as well..... Sorry! I misread your message, and (mis)understood this to indicate I was reading a CC: of something you had sent to the BBS! Again, I'm terribly sorry if I caused you any undue stress or emberrassment. I would _never_ post a personal letter to myself in any public forum (knowingly, I mean..) Would it be any consolation if I mentioned I too was quite fatigued when I sent this reply? Now, to the meat of the matter.... The parasitic inductance/Q thing is beaten to death. May it RIP. : I read the Jung articles at the Terman Engineering Library at Stanford. :Wish I were there now. Since it's been about ten years, I don't remember :all of the details. I agree that the difference in dielectrics is the :key issue, else there wouldn't be much to argue about, except parasitic :inductance due to construction. What I wonder is whether the addition of :a small amount of quality capacitance can compensate for an overwhelming :amount of non-quality capacitance. My intuition says the improvement :offered by a bypass capacitor is a function of the relative reactances :of the two caps in questions. If the electrolytic looks like, say, one :Ohm at 20kHz and the film cap looks like 150 Ohms, I can't see that the :superior DA and DF of the film cap will have a significant effect. I do, :though, have an open mind. At this point, it would be nice if someone :could make some before/after DA and DF measurements of bypassed and :unbypassed capacitors. If a significant difference is found, we have :good engineering evidence that bypassing is effective. If not, we have :to fall back on subjective arguments, with all of the associated uncertainty :and bias. There are many factors known to afffect a capacitors "sound." Most of these factors have test and reporting standards, so can be assessed in a very scientific way. Rather than rewrite a compendium on capacitor behavior, I will leave a couple of references for any interested parties to consult: (Of course) Walter P. Jung's excellent series printed in AUDIO magazine - sometime in the eighties. Don't have the info at hand, but ANY library should... Al Cooper @ Illinois Capacitor. Sales brochures, catalogs, and technical information are available. When you call Al, tell him you read this in the rec.audio.high-end board on internet, and mention me (Brad Sanders). Illinois Capacitor Inc. 3757 W Touhy Ave. Lincolnwood IL, 60645 (708) 675 1760 Music Interface Technologies Call and ask for the "White Papers" on the MIT MultiCap. CVTL Inc. 3037 Grass Valley Hwy Auburn Ca. 95603 (916) 823 1186 : [Defending opinions as facts is] a big business actually. See, I "suggest" : to you that in fact there IS a HUGE difference between wire A and wire B. : So, you go "upgrade" all your stuff to wire B. Now, whether in fact one DOES : hear a difference or not will be biased not only by ones desire to hear a : difference (in my opinion, nothing wrong with this), but also by the NEED to : hear a difference, planted there by someone with a purely economic interest : (I have a lot of problems with THIS!) : Pretty much exactly my feelings. However, I think the desire to hear a :difference is a big problem because it strongly biases the listener. Not :always, mind you, but certainly sometimes. But I suggest, at least in this case, "the customer is always right!" If it is my system, and _I_ perceive a difference in sound, who are you to disagree? This is rather like chrome on a car, eh? Doesn't improve speed or handling, but _I_ like it (or don't) and it's my money.... To quell your now rising blood pressure , i'll remind you of MY quote above. This is the reason I have a problem with the TAS concept of objectifying subjective realities. You really can't argue with someone stating "I like this better." Whether or not YOU in fact can tell a difference, THEY may be able to do so. It may be insanity, but it is their personal insanity, and it harms neither themselves nor society. Seems like a much more productive habit than drugs! On the other hand, to say "I can hear a difference, and if you can't you're an ignorant fool," either directly or by implication, is the stuff wars are made of. I wonder what would happen if R&T decided oneday to start reporting things like painting steering wheels orange improved handling and putting white shoe polish on tires improved traction? On the other hand, what if they reported "Joe X racing team has found these things to help...." Who's reputation would then suffer? See my point? Happy Listening! nuke ___ Blue Wave/QWK v2.11 ------------------------------ Date: Tue, 15 Feb 94 15:28:04 -0500 From: Scott Dorsey Subject: caps ::The ceramic caps have a significant change in value with temperature, ::even the NPO types. This can be a real problem in a tube amplifier ::which is going to be running significantly above room temperature. On ::the other hand, for supply decoupling, this isn't a big deal. : :I doubt it, that's why they're called NP0. One important point that seems to :be getting lost in this discussion on ceramic caps is that there are *two* :construction techniques used: The single layer ceramic and the multi-layer. :The MLC's are the one that exhibit most of the 'evil' characteristics that :people complain about, and for good reason; these caps are the most mis-used :and mis-applied tool since the screwdriver! Admittedly true. The NPOs, though, were very NPO with respect to the dielectrics that they replaced, but they aren't very NPO compared with some of the film caps available today (or, even better, glass caps). I strongly recommend that anyone interested in the subject call up AVX and ask for their catalogues which contain more information on ceramic caps than you probably care to know. :The single layer caps, while exhibiting a small amount of piezo effect, are :much better for bypass, and are used extensively as coupling caps in RF work, :just look in your old transistor radio or your TV tuner. Speaking of tuners, :they're also stable enough to use in the tuned circuits in those critters, too, :and that tube-type gear! They're a big improvement, that's true, but still not up to the quality of some of the modern film caps, and as you point out, they are impractical for larger values. --scott ------------------------------ Date: Mon, 28 Mar 94 13:42:56 PST From: john@pelican.hac.com (John Drab) Subject: polyester caps & table Q's A'd ;Date: Thu, 24 Mar 1994 13:55:39 -0800 (PST) ;From: skanan@otl.scu.edu (Suttipong "Nick" Kanakakorn) ;Subject: Re: DIY: Capacitors ; ;In my DigiKey catalog, there is Panasonic 10uf 100V polyester ;cost $3.68 each. Get 5 of them. The part# is E1106 and ;the phone# is 800-344-4539. You can by pass them with other ;small cap too. ; ;-Nick If you have decent quality electrolytic capacitors on the input now, it is doubtful that the Panasonic polyester capacitors will result in any improvement. In fact, it is likely that your amp will sound worse. The Panasonic polyester caps show a significant amount of hysteresis when measured with low amplitude sin waves and without dc bias. Unless there is a significant DC bias across the cap (10v or more) this is the wrong cap to use ! Recomended commercial caps for this application include polycarbonate capacitors from WYMA or RIFA, Polystyrene caps from CDE, TRW, or Component research. and teflon from Tex cap. Tubular construction is prefered, but normal wound construction should outperform all but the very best elecrolytics. all of these brands are pretty widely availiable. -jd ------------------------------ From: peterca@ento.csiro.au Date: Tue, 29 Mar 94 09:41:33 EST Subject: Re: DIY: Capacitors >>> A question about my B&K ST-140: it has 47uF input caps, electrolytics(!). It >>> seems that there are no "good" caps available with this value (all are much >>> smaller). What would happen if I replace these with the largest, say, MIT cap? >>> Would it change the circuit or would it just filter at a different frequency? >>There has been advice already about the possible replacements. If you are >>absolutely confident that all your sources have DC blocking caps or DC >>servos to prevent DC appearing on their output an alternative is to get rid >>of the cap altogether. IE remove it and replace it with a piece of wire. > >that might be a bad idea, especially if the cap is used to block dc from >*LEAVING* the amp - ie, it's to isolate the internal biasing stuff from the >outside world. put a shorting plug on the input in question and check for >voltage on the amp side of the cap before just removing it. the designers >usually have a clue as to what they're doing :) Good advice. (Mine was the >> bit BTW) The cap may have the function described in the > bit in which case it certainly should stay. My point however remains that the designers MAY have been simply over-cautious to prevent DC entering and you will have to remember that it is removed if you change preamps. Peter Campbell (peterca@ento.csiro.au) C/O Div. Entomology, C.S.I.R.O. Box 1700 Canberra A.C.T. 2601 Ph.61-6-2464158 (w), 61-6-2516213 (h) Australia 61-6-2464173 (fax) ------------------------------ From: torgd@tolstoy.lerc.nasa.gov (Russell DeAnna) Subject: DIY: Capacitors Date: 14 Apr 1994 16:15:41 GMT This letter was sent by Barry Jacobson in reply to my question about using Mallory capacitors rated at 650V in a passive crossover. --------------------------------------------------------------------- No, you are certainly NOT doing damage to the low-level signal by using a capacitor with a higher voltage rating...in fact, if anything, it will likely be better than a Mallory .1uf polypropylene cap of the same style at 200 V or 400 V or whatever < 630!! (It will also be bigger! and move expensive!) For over 15 years, several friends and I have done extensive component comparisons, none more extensive than with capacitors. We have done both non-blind and single-blind testing with repeated trials separated by days or weeks. The number of caps sampled in these shoot-outs is substantial (and the amount of money spent on caps that did not win these shoot-outs has been even more substantial :-). We have come up with several generalizations: 1. If a cap from Manufacturer A, of Style/Model B, of value X mfd, rated at Y1 volts (which I will denote A:B:X:Y1) sounds good, then a cap from the same manufacturer of the same style and value, but rated at Y2 volts where Y2 is higher than Y1 (denoted A:B:X:Y2 (y2>y1) ) will almost always sound better! 2. If a shoot-out is done to compare caps at X mfd and Y(or greater) volts and polypropylenes and/or polystyrenes are included in the tested population, the shoot-out winner will be either a polypropylene or polystyrene (no, not oil, but not teflon either!) 3. Using the denotational nomenclature above, if A:B:X1:Y sounds great, there is *NOT* a particularly higher level of confidence that A:B:X2:Y (x2<>x1) is also as good. This has been *VERY* frustrating!!!! (It sure would be nice to say that a *line* of capacitors is the best choice!) One note: Mallory would not be my choice at .1 ufd. At that value, the best we have heard actually *IS* the MIT RTX (polystyrene), but not either the film & foil or metallized film polypropylenes. And the 600V is clearly better than the 400V (although the 400V .1uf MIT RTX is good, and better than most other MIT caps of whatever constructions and whatever voltages!) BTW, the other really good MIT is the .01 uf / 600 V RTX...our favorite bypass cap to date. Barry Jacobson ------------------------------------------------------------------------ -- - Russell DeAnna