Jump to content
The Klipsch Audio Community


  • Content Count

  • Joined

  • Last visited

Community Reputation

33 Excellent

About Backfire

  • Rank
    Advanced Member

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. Another point is that what is commonly called the "damping factor" is a rather misleading, not to say almost meaningless, concept. I sketched a couple of diagrams in a pdf, to show what I mean. Its not a new point I'm making, and is perfectly well understood by those who analyse equivalent circuits, but it is one that does not seem to be widely appreciated. The calculations I've included are a bit rough and ready, but they illustrate the key idea. The main point is that the speaker resistance (literally the resistive component of the speaker's impedance) should be being added to the output impedance of the amplifier, and it should appear in the denominator in the calculation of a meaningful damping factor. (Not added into the numerator, where it is commonly put!) Because the speaker resistance is generally a major contributor to the overall impedance of the speaker, it means the true damping factor will never really get much larger than about 1. damping-factor.pdf
  2. It probably depends what you mean by "for testing purposes." The pinout is the same as 12AX7 or 12AU7, but the 12DW7 is a funny hybrid, with two unequal triode sections; one has a lot more gain than the other. You won't get fireworks if you try 12AX7 or 12AU7, but it may only give a rough idea of how it's working.
  3. Yes, I was speaking a bit loosely. People sometimes speak of a class AB amplifier as "operating in class A up to X watts," where X is the maximum power it can put out before the tubes essentially stop conducting at the bottom half of the audio cycle. It can give a useful characterisation of how much power it takes before the tubes go into cutoff.
  4. I hunted around with Google, but I couldn't track down a schematic for the Joule-Electra. However, as far as I can make out, it is a circlotron. They can be very good indeed. I have one that I built a few years ago, using EL509 tubes in the output stage. I am a bit sceptical about the Joule-Electra claim that is is biased for full Class A at 100 watts into 8 ohms, though. That would mean 5 amps maximum current. With three tubes per side in the circlotron, that means a maximum of about 1.7 amps per tube. That is certainly not a problem for the 6C33C, as a maximum current. However, if it were truly running in class A up to 100 watts, that would mean the quiescent current in each tube would need to be about 0.83 amps. With a plate supply voltage of probably about 150 volts, that would be about 104 watts tube dissipation, which is well in excess of the rated 60 watts maximum for the 6C33C. It's not a problem to have the plate dissipation exceed 60W instantaneously, but it is a problem to have 104 watts steady dissipation. Usually in an OTL, 6C33C tubes are biased for about 200 mA quiescent current. All of which is a long-winded way of saying that I think it is almost certainly running in class AB, and not class A. At low powers, up to a couple of watts, it will effectively be running in class A, but transitioning to AB at higher powers. Nothing at all wrong with that, though! It's just a little bit naughty, I think, when manufacturers make overstated claims about class A operation. (Joule-Electra would not be the only one!) I really like the circlotron design; a truly symmetrical type of output stage. I'm sure the Joule-Electras must sound superb!
  5. I completely agree with those saying that capacitors are not a big problem. And another point is that if one has any doubts as to whether a capacitor might have any deficiencies, one can easily check by using an oscilloscope to look at the signal across it. The ideal would be that a capacitor in the audio path (e.g. a cathode decoupling capacitor, or the final capacitor in a power supply) would have zero audio signal across it. Of course in practice, being of only finite capacitance, it will have some non-zero AC reactance, and so there will be a small audio signal across it (less and less as the audio frequency increases). The magnitude of the expected audio signal as a function of frequency is of course easily calculable. If the capacitor were causing any non-linear distortion of the audio signal, then this would show up in the signal one would see across the capacitor using the oscilloscope. If the distortion were enough to be audible, then it would easily be visible and measurable. By measuring the distortion, if any, in the signal across the capacitor one could easily estimate the distortion it would cause in the audio output from the amplifier. Almost certainly, unless one has made a really unsuitable choice of capacitor or it has some serious fault, the distortion it would cause would be completely negligible. But in any case, if the capacitor is causing any problems of this kind it is easily measurable and understandable using basic physical principles. I would be much more inclined to believe what the instruments were saying than the anecdotal accounts of a human who probably hears what he wants to hear.
  6. Yes, I think you are right; most OTL amps these days do not have a capacitor between the output and the speaker. I have four different home-built OTL amps myself, and none of them has an output capacitor as such. Same is true, I believe, of commercial designs like Atmasphere or Bruce Rozenblit's Transcendant Audio amps. There are still capacitors in the output path, because the audio signal passes through the power supplies. The same is also true, of course, for an SET or push-pull amplifier. It is not true that OTLs require paralleled output tubes. Two of my OTLs, for example, each use just a pair of 6C33C tubes in the output stage (totem pole), so no paralleling of tubes at all. They both give 25W into 8 ohms. In any case the alleged "choir effect" of paralleled tubes is a non-existent phenomenon, in my opinion, that is not recognised by any reliable authority.
  7. Yes, the 6C33C OTL is about 25W into 8 ohms. I'm just using the single power supply for the two channels, if that's what you mean. The big toroid on top is 117-0-117 secondary, for the main positive and negative HT supplies for the output tubes, and also, with voltage doubling, for the input and stages driver. There are a couple of smaller toroids under the chassis, for the heaters. The main HT transformer is a bit of an overkill in terms of power handling, I think. I don't have any problems with it. The design of the OTL is based on one by Tim Mellow; it appeared in Audio Express, February 2010. The power transformer for that other OTL is a bit bigger, and sits in my front garden...
  8. The first is my everyday amplifier, a conventional type of OTL amplifier using 6C33C output tubes. I quite like the second one too; again OTL, using 6082 output tubes. The novelty with this one is that there are no transformers at all; it runs directly from the mains supply. It's based on one of the earliest OTL designs, by Dickie and Macovski in 1954. Regretfully, my constructions tend to be rather utilitarian in comparison to some of the beautiful ones built by others!
  9. Fully agree on everything you said. And having now looked over some of Swensen's writings in that document, I agree he seems to be fairly firmly grounded in physical principles. I have the impression he is rather better than Hasquin in this respect. I agree also, especially given the sort of power supply he is discussing, that taking proper account of the entire system as a whole, power supply and amplifier, is necessary. It surprises me somewhat that he bases his analysis so much on PSUD2, since although not bad for what it can do, it is very much restricted in its capabilities when compared with LTSpice. Having the ability to turn on a step load in a power supply is all very well, but not much help for doing a proper simulation of the interaction between the power supply and the audio signal in the amplifier. This would not be so crucial if one had a nice beefy final capacitor in the power supply to keep the supply voltage steady. But in a case like he is discussing, I think being able to do decent simulations of how the amplifier audio is interacting with the power supply is rather important. One thing in Swenson that caught my eye and puzzled me is his statement (p. 25) that "Many traditional LCLC and CLC type designs have radically varying impedance at different frequencies. I have found this does not sound good. Its impossible to achieve completely flat impedance response, so I try and keep it to a range of 3 to 1 or better." I presume he is speaking of the variation over the audio range. With a 50uF final capacitor in the power supply, that would mean about 0.16 ohms impedance at 20 kHz. He cannot possibly mean that he aims for an impedance of less than about 0.5 ohms over the entire lower-frequency audio range! So what, I wonder, does he really mean when he says he aims for less than a 3 to 1 variation?
  10. I haven't looked at the Swenson document, but I did look at some parts of the Hasquin document, and some of what I saw there seems to be, to say the least, misleading, and in fact wrong. One of his points that he emphasises is that large capacitors in a power supply "hog all the current," and prevent the amplifier getting what it needs. He says, for example, "Once the capacitor has used its low impedance to source current, it then uses that same low impedance to hog any and all available current to regain its charge." In another e-mail, he says "The thing that most of you fail to understand is not only is the capacitor a low impedance source of current, it is also a low impedance current sink. As soon as there is a slight change in voltage across the capacitor it begins an attempt to sink current to recover its lost charge. Since it is a low impedance device, it competes directly with the output tube for current resources. Considering a 50uF capacitor has an impedance of only 159 ohms at 20Hz, how well do you think a tube that has an Rp of 800 ohms (2A3) is going fair against the capacitor? The answer is not very well. The lower impedance device will always win a larger share of the resources. This is the reason why there are so many people that believe a lot of capacitance sounds bad in a tube amplifier; THAT’S BECAUSE IT’S TRUE!" He is inviting the reader to draw the conclusion that the tube with an Rp of 800 ohms will be coming off badly when compared with the 159 ohms impedance of the 50uF capacitor. Presumably, by the same token, he would be saying that if the load were a 10 megohm impedance instead of 800 ohms it would be coming off really, really badly! Of course, that is not true at all. Likewise, if the power supply capacitor were 500uF instead of 50uF, he would be inviting one to conclude that the amplifier with Rp = 800ohm will be coming off really badly compared to the 15.9 ohms impedance of the capacitor. Again, his conclusion is incorrect. One thing that he is not taking into account is that the voltage drop across the capacitor when it supplies the increased load current will actually be a lot less if the capacitance is greater. And all that the amplifier cares about (or rather, that the listener cares about) is that it should be getting the required voltage from the power supply. As long as that voltage doesn't sag too much in the face of the increased load, that is what matters. It becomes rather clear what is going on if one plays around with some simulations in LTSpice. (PSUD doesn't seem to be so useful for this because, as far as I understand it, never having used PSUD, about the only way one can try to incorporate the effects of a time-dependent load on the power supply is to turn on a step-function load at a given time.) In LTSpice it is easy to simulate the power supply and a periodic time-dependent current load, such as the sinusoidally-varying current demanded by a class A SET amplifier when it is reproducing a loud sinusoidal signal. It than becomes evident that the output capacitor in the power supply is not "hogging all the current" in the way Hasquin says. In fact, if one is using sensible inductance values in a CLCLC or LCLC power supply (like critical inductance rather than the tiny little inductances in their so-called "LSES" supplies), then the current through the final capacitor is a sinusoid of just about the same magnitude as the sinusoidal current passing through the output stage of the amplifier, but about 180 degrees out of phase. In other words, as one would expect, the current passing through the second inductor in the power supply is very nearly constant. (To put that another way, during the portion of the audio cycle when the output stage is drawing more than the average current, the excess is supplied from the capacitor. Then, during the portion of the audio cycle when the output stage is drawing less than the average current, the capacitor is recharged again. The capacitor is drawing an increased current during the portion of the audio cycle when the amplifier doesn't need it. And, of course, contrary to what Hasquin is implying, the large the value of the final capacitor in the power supply, the more nearly constant is the output voltage from the power supply. If he is saying that having a constant voltage supply for the output stage of the amplifier sounds bad, then he is really saying that he want to have a power supply that injects some added sound effects, by having a supply voltage that bounces around with the musical signal. Maybe that is indeed what he and some other people want. But they should be more honest, I think, about saying that that is what they personally prefer to listen to, and they should not be prosetylising to the whole DIY community that everyone is is making their power supplies wrongly, and that their way is the only proper way to do it.
  11. Taking this, and some of your other assertions, at face value, you must surely admit that by many people's standards they are extraordinary claims. It is therefore reasonable to ask for proof of extraordinary solidity. Knowing, as we do, that the human ear and mind, as with other senses, can easily be deceived, the surest way of being certain that the claimed effect is genuine is to carry out double blind listening tests. You spoke, for example, of hearing a "disgusting skew" if L1 and L2 in an L1/C1/L2/C2 power supply were unmatched. That sounds like something that it would therefore be very easy for you to demonstrate as audible (to you), if you were to take part in properly controlled double-blind listening tests. You are not talking about a tiny, subtle effect; you are talking about something that "sounds disgusting" to you if there is a mismatch. It should be incredibly easy for you to demonstrate that you really are hearing that, and that it is not a result of some expectation bias on your part. When you think of all the years you have probably spent arguing with the many people who don't believe your claims, and the fact that it seems to be very important to you for people to believe the claims you make, it would be relatively easy for you to produce quite compelling supporting evidence by taking part in a rigorously-controlled double-blind listening test. Maybe consider doing something like that?
  12. Honestly, it simply isn't worth the expenditure of time and effort sifting through 25 pages of Hasquin and 37 pages of Swenson to see what might be in there amongst the blah blah blah. I can already see the gist of the thing; it is not that hard to understand what happens if one uses varying amounts of inductance and capacitance in a power supply. It is all standard EE understanding. If they are really saying that they just happen to like the various colourations that result from having a bit of hum and a bit of sagging, or whatever, then that is fine. To each his own...
  13. You speak as if these guys were like Moses bringing down some sacred scripts from the mountain! They are just a couple of guys with their own understandings, and maybe sometimes misunderstandings and prejudices too. Those documents you linked to are a bit of an undigested dog's breakfast and I really don't have the patience to sift through all the blah blah blah. One thing did catch my eye, where Mr Hasquin is talking about the merits of using smaller rather than larger capacitors in the power supply. He says that "In all our engineering books we are taught that a high capacitance power supply is good. The authors imply that the low impedance of the capacitor will be able to deliver current to the output tube quickly and prevent voltage sag. On the surface this looks good on paper. However, it could be more wrong. Once the capacitor has used its low impedance to source current, it then uses that same low impedance to hog any and all available current to regain its charge." (I think he probably meant to say "couldn't be more wrong" when he said "could be more wrong.") He goes on to say that "the capacitor will always rob the output tube for available current in order to maintain its charge." It sound like he is probably a bit confused here. The larger the capacitor, the more steady the supply voltage available to the output tube will be. In the limiting case where the power supply capacity goes to infinity, the voltage supplied to the output tube will be absolutely rock steady. (A bit like having a B+ supply coming from a set of car batteries connected in series.) He goes on to say that "This is why over stuffed power supplies always sound slow. The over stuffed power supply prevents the output tube from reproducing dynamics." Maybe he is trying to say that he considers it desirable, from an audio point of view, to have a supply voltage that is wobbling around in correlation with the demands placed upon the power supply by the amplifier. I think I have heard it said sometimes that the kind of "audio expander" effect that one gets with a saggy power supply actually sounds quite lively and dynamic. If that is what he is really talking about, and he is really saying that the sagginess of a poorly-regulated power supply is what is desirable, then that is absolutely fine, but it is good to be clear about why he is after the properties whose virtues he is extolling. As always, of course, the discussion of such matters is very different for a class A amplifier and a class B amplifier. For class B, like a typical push-pull amplifier, the net current demand (averaged over the audio frequency timescales) gets much higher when the music gets loud. By contrast, in class A (like an SET amplifier), the net current demand (averaged over the audio-frequency timescales) is more or less constant, regardless of whether the music is playing loudly or softly.
  14. My reading of Maynard's reply to you is that he has already read the documents that you allude to, and that he disagrees with them. By the way, one should not make the logical fallacy of of concluding from Schopenhauer's words that all ideas that are ridiculed are necessarily the truth.
  • Create New...