Damping factor of the Macs is said to be 12.@dentdog with that much feedback, none at all. OTOH, the fact that the amp runs feedback means it will *sound* a little brighter on top due to added trace amounts of higher ordered harmonic distortion. The emphasis is added on account of on the bench no change in frequency response would be seen.
What, if any, difference in frequency response of my speakers should I hear due to the varying impedance curve?
Showing 8 responses by atmasphere
Not in that sentence, surely! I don't think I jumped the gun; the statement is a common myth and I showed why in my post. You don't need a lot of current to 'control' a speaker, what you need is an output impedance that is sufficiently low, and you don't need 'current' to get that.Ralph you jumped the gun mate, read again, I never tied output impedance and current output together in the same sentence.Then to just complicate what I said above, the current output of an amp comes into as well, to keep the said control over the speaker.This is a common myth. Also to add, an amplifier "with low output impedance", but with not much "current ability" like otl's, have no chance driving a pair of Wilson Alexia's and like, which drop to an EPDR of .9ohm around 100hz (the power region) without serious problems. As an OTL manufacturer I might be recognized as having a bias, but here is the simple truth of the matter: The output section of an OTL has to do the same sorts of things that a solid state amp might do- and that is drive a speaker. It has to be able to make current to do that. Our amps don't have a particularly low output impedance owing to little or no feedback, but if they did, driving this particular speaker really isn't that hard. All of the Wilsons have been fairly easy loads for tube amps in general despite their impedance! We've had very good results with the Sophias, which are a similar load. I've heard that combo many times. I've not heard the Alexias yet. FWIW, the sales manager of Wilson had our amps for some years until an outside organization tried to take over Atma-Sphere about 13 years ago. At that point he sold them to be out of the possible politics. To give you an example of how much current is available in one of our amps, while tampering :) with a set of MA-1s many years ago I noticed that if the output section was deprived of bias, all by itself (IOW not including the filament circuit) it could blow a 15Amp fuse without damage to the power tubes! Certainly, with that sort of current, the power tubes would overheat fairly quickly, but for short (no pun intended) periods of time they can survive much larger amounts of current than one would expect. Our driver section has the ability to drive the output tube grids to about 15Volts positive with respect to the cathodes (which means there can be quite a lot of grid current; this is 15V more than the output section would have on the girds if deprived of bias!) and the output section can be linear in this region. IOW we operate the amplifier class A2. IOW, we have the current, but not the low output impedance. Weird, huh? The fact that the amp does not act as a voltage source is simply because to do so requires about 20 db of feedback. We've certainly built amps that way but never got them to sound as good. Proper application of feedback is a lot trickier than most engineers think! You can't just apply it according to the formulas and expect it to work, because the formulas don't cover everything. This article (be sure to read part 2 as well) does a good job of covering the problems and also proper application, which as far as I can tell is not executed by about 95% of **all** amps employing feedback: http://www.normankoren.com/Audio/FeedbackFidelity.html If the principles laid out in this article were applied in audio today, it would advance the art. So far, its been simpler for us to achieve our design goals by being pragmatic and recognizing that our amps won't drive all loads. But if we can get the amp and speaker to work together, the combination has the ability to cross the line between music and hifi. Roger said that though the OTL/QUAD combination was indeed considered by many to be a good synergistic match, he considered it to be the opposite---an OTL being the absolute worst amp design for use with that speaker.Roger and I don't agree on that one- we have a lot of customers with Quads. I've heard his RM9 (an excellent amplifier BTW) against our amps on ESL63s a lot as a friend used (may he rest in peace) to have them. |
Then to just complicate what I said above, the current output of an amp comes into as well, to keep the said control over the speaker.This is a common myth. An amplifier can have a very low output impedance without having much 'current'. I put the word 'current' in quotes because Ohm's Law says without exception that the current flowing in the load is going to be directly related to the power being dissipated by the load, and that current will be the same regardless of the output impedance of the amp. Example: 2 amps, one solid state with an output impedance of 0.1 ohm, and the other with an output impedance of 4 ohms are both making 50 watts into an 8 ohm speaker. How much current is present? The Power formula is a derivation of Ohm's Law, and is current multiplied by voltage (1 watt = 1 Amp x 1 Volt). If you are dealing with an impedance, in this case 8 ohms, the derivation is Power = Amps(squared) x Resistance. So: 50 watts = 8Ohms x Amps squared. Solving for Amperage we get 2.5 Amps. Note that at no point does the output impedance of the amplifier figure into this- its not like an amp with a lower output impedance will somehow cause more current to flow. IOW the current is determined by the resistance of the load and how much power is being made. Now there is the issue of the amp 'controlling' the speaker. The speaker cone moves in relationship to the energy applied to its voice coil and returns to rest when that power is taken away. What we are concerned about here is the issue of overshoot; that is to say any motion that is not having to do with the signal applied. This is where the concept of damping comes in. If the speaker is in motion and further is motion that is entirely overshoot then what happens is it will make a voltage, since there is a voice coil moving in relation to a magnetic field (this is known as 'back EMF'). That voltage has to be absorbed, and that is done by the output impedance of the amplifier, which acts as a short to that voltage. From this you can see that the amount of current available in the amp really does not play a role in this damping effect. The output impedance however plays a huge role! So what we can conclude is that a low current amplifier that can make the power needed will do the job as long as its output impedance is low enough to provide proper damping of the load. FWIW, no speaker is known to need more than about 20:1 for optimal damping, and each speaker in a box (or not) has an optimal damping. For this reason back in the old days there were a number of amplifiers that had a variable damping control, which was a feedback control balancing current feedback and voltage feedback to allow the amp to have an adjustable output impedance over a wide range such that it could work with any speaker. In the last 4 decades though there has been a race to get more and more damping factor out of amplifiers, but the effect has been to cause most loudspeakers to be over-damped. We've all heard 'tight' bass that has thump but not much in the way of definition. Its a thing that a lot of audiophiles like but its not natural- 'tight' bass does not seem to exist in real life. When the speaker is overdamped, the cone cannot make its full excursion before the waveform starts going the other way. So bass notes in particular get truncated- the body of the bass is less prominent although the thump is there. By limiting the damping factor to no more than 20 or 30 to one this effect can be reduced or eliminated with most modern loudspeakers. |
@georgehifi , that's correct! As I pointed out in the article, any time you mix equipment from the two technologies, the result is likely to have a tonal aberration. The example you give is a good one, although 'fry your ears' is likely not what would happen on account of the higher impedance would result in low distortion from the amplifier. It just would not make the bass energy since designer is expecting 3 db more power output out of the amp in the bass region. Another example is solid state driving an ESL- with the reduced impedance at high frequencies and the tendency for the load impedance to vary about 10:1 (IOW, if 10ohms at 50Hz, could be about 1 ohm at 20KHz) over its range, the result will also be too bright with not enough bass. The reason is that the impedance curve of the speaker is based on a capacitor and not a driver in a box, so the impedance curve is not also an efficiency curve. This is quite unlike a driver in a box, where the resonance of the driver is represented by a peak in the impedance curve (IOW the impedance and efficiency curves are the same thing). This is why the equipment matching conversation is still very much with us 5 decades on! Its also why of two amps on the bench that might measure perfectly flat to 100KHz, one might sound bright while the other does not. Its worth pointing out that if the designer of the speaker is expecting the amp to have a higher output impedance that the crossover will be designed differently as well. The fact that the source impedance of the amp can affect how the crossover works means that drivers might be operating outside of the area for which the designer intended! This is one of the reasons that horns got the reputation for being 'honky'. |
What Al said, plus You know, it is a real shame that you take this tack, because I actually think you have a lot of fans, and built good products, but I've gotten really tired of you constantly hammering the superiority of high output impedance amps as a feature based on false and misleading (that's your own phrase) statements you repeat over and over again. What is it that you're saying here? That I build good products, but - ? they aren't real or something? I'm sensing a contradiction here. IMO you've been misreading what I've been writing. I have qualms about **feedback**, not output impedance, as long as the low output impedance does not come with added higher ordered harmonics. And I can back my qualms up easily enough without going into the weeds. If you are not familiar with Norman Crowhurst, he is a well-known authority in the field of amplifier design. His books are likely a bit rare but the important ones are a free download from Pete Millet's site: http://www.tubebooks.org/technical_books_online.htm In fact I would love it if I could make the output impedance of our amps lower. The problem I have with solid state is that many semiconductors have a non-linear aspect about them that causes them to have higher ordered harmonics (at low levels, but as I pointed out earlier, the ear is very sensitive to that sort of thing) and hard clipping. The only devices that I have found that don't are the static induction transistors made by Sony. IMO They had a chance to really set the audio world forward, but in true Sony fashion (which is to come up with an innovation and then shoot themselves in the foot) failed to make a full complement of driver and voltage amplifier devices to go with their rather amazing output devices. With such technology we could have had low distortion, zero feedback and low output impedance all at the same time! For the most part, what I'm really hammering on is the simple fact that you can't seem to get low distortion **and** low output impedance at the same time. Now I do this in the face of the fact that many solid state amps **appear** to have very low distortion, but what research in the last 40 years has shown is that the ear has distortion on a curve of sorts; the lower ordered harmonics being unimportant to the ear and the higher ordered harmonics are really really important. I freely acknowledge that this flies in the face of the test and measurement regime, which I feel is outdated by research of the last 40-50 years. Can you describe the sound of an amplifier by looking at its spec sheet? Most audiophiles can't; so why is there a spec sheet? The fact is most spec sheets are there to make the product look good on paper and have nothing to do with how the ear hears. That's a pretty good example of the Emperor's New Clothes... Now if one is to say that the specs are the final arbiter, great, no worries. Have at it. But I feel that the ear, not the specs, is why we buy audio equipment and that equipment will sound better if the gear follows the rules of human hearing. Did you mention what your speaker brand is? I missed that bit. |
If this had any basis in reality, at all, speaker testing would be based on power vs. frequency. It isn’t. It is input voltage vs. frequency. Hmm. So you’re saying that a solid state amp can make as much power into a 30 ohm load as it can an 8 ohm load? And that with an ESL, the impedance curve is also a graph of its efficiency? I think we both know that neither is the case. SoundStage Magazine online has a review for the Atma-Sphere MA-1 Mk II.2 mono-blocks: The conclusion here is false; starting with ’with an excellent...’. The reason is that the MA-1 is a zero feedback device and is not a ’typical’ tube amp! If it had enough feedback (about 20db), it would behave as a voltage source and the test as shown would have measured flat. Also, this does not debunk the idea that a solid state amp will have troubles driving higher impedances. IOW, this is typical behavior of a **zero feedback** tube amp and not that of one employing feedback. So one has to ask why we would make an amplifier that has no feedback with resulting (apparent) FR errors, and the answer is that feedback adds higher ordered harmonic content that is easily detected by the human ear. We get low distortion without feedback (assuming that the test equipment does not ground one of the speaker terminals which can result in a bogus test) so then we are simply looking to find a speaker with which the amp will have flat response. Dr. Herbert Melcher, a neuro-scientist, has shown that the brain has a variety of tipping points. One of them appears to be that the ear/brain system will favor distortions interpreted as tonality over actual FR errors. So our approach is that if we can minimize the tonalities generated by the amp due to distortions, it can often appear to be more neutral than amps that measure much flatter on a given speaker on the bench.
We don’t try to give this impression! Erik (@erik_squires what is the brand of speaker you make?) is correct otherwise; most speakers are Voltage Paradigm devices. But not **all**; some are based on the prior Power Paradigm, as are zero feedback tube amplifiers (see link at the end of this post). If you’ve ever seen an older speaker with midrange and tweeter level controls, the reason they are there is not to adjust the speaker to the room but instead to match to the voltage response of the amplifier. These controls are not seen on Voltage Paradigm loudspeakers. Long ago we discovered that our amps sounded much better without feedback (smoother, more detailed, wider and deeper soundstage). It took a while to find out why! In the meantime though, we have always been very careful to ascertain which speakers would be a good match. This is important (at least to us), because this is the dividing line between what makes for a good hifi as opposed to what sounds like real music. That we are still around 40 years later suggests that there are good number of speakers that work with our amps. This is not to say that a Voltage Paradigm speaker can’t sound like real music; it is saying that if that speaker won’t have flat response without an amplifier employing a lot of feedback, then that speaker will never sound like real music on account of the amp. The **reason** is that loop negative feedback ignores a fundamental law of human hearing which is how we detect sound pressure. Rather than detecting the volume of a fundamental tone, our brains sort out sound pressure by detecting the strength of higher ordered harmonics associated with that sound (which are increased if feedback is employed, although in ’trace’ amounts, but our ears are literally tuned to detect those traces). If the harmonics are louder than they should be, then the fundamental(s) will be interpreted to be louder too. That would not be such a bad thing but the extra harmonics are also interpreted as brightness and hardness by the ear /brain system and that is why they impart the sound of a hifi instead of real music. BTW, this is easy to prove using simple test equipment. So we abandoned feedback and as a result the Voltage Paradigm as it has that fundamental flaw (although otherwise works great) and went back to the Power Paradigm (which was what was around prior). We’re not the only ones; SETs, horns, ESLs and a number of other speakers (which are admittedly limited to high end audio) are also built around Power Paradigm principles; we’re not the first or only ones that have sorted out that loop feedback isn’t perfect (see the writings of Norman Crowhurst; these problems have been known for over 50 years). The Soundstage test assumes that all amps are based on the Voltage Paradigm and the simple fact is they are not. If your amp’s impedance is high enough, it will become an ideal "current source" where the output current has nothing to do with the load, and therefore the voltage at the speaker will vary wildly with the speker’s impedance. Atma-sphere has a unique perspective on this issue, and seems to constantly be suggesting a current source is ideal. He stands very much alone in this area.This statement is mostly false. Why: We don’t hold this viewpoint, and Also a current source amp will have an output that in fact varies according to the load, but in an inverse way. Nelson Pass has built some amps based on this principle, but he never made any such amp for production, and I’m not aware of anyone who has. Current Source amps IOW never developed any sort of following. Put another way, if the impedance of the amp is low, it is likely to act as a voltage source. If the output impedance is medium (a few ohms, like SETs and our amps) then the amp will act as a Power source (constant power into all impedances rather than constant voltage). If the output impedance of the amp is very high, multiples of the speaker impedance, then it will be a current source. That being said, the question of sound quality and amp/speaker matching is purely subjective, and you should listen for yourself. I’m pretty sure my favorite tube amps of all time, the CJ Premiere 8s were quite high in output impedance, but I’m not going to try to sell new physics to convince anyone to buy them.Listening of course is what this is all about :) The cj has a fairly low output impedance owing to the loop feedback employed and behaves as a voltage source. It has a classic tube sound as you might encounter with any excellent hifi. For more on the Voltage and Power Paradigms see http://www.atma-sphere.com/Resources/Paradigms_in_Amplifier_Design.php |
They need about 15 Watts of the best amplificationThe above quote is taken from the link Erik provided. I've serviced the 15-watt tube amps that Quad made to go with the '57s; in a nutshell driving a 'difficult load' is something they can't/don't do. But they drive the Quads just fine. When 15 watts is all that's needed, its not a hard load and no mistake :) As Erik points out, the impedance curve is very typical of an ESL. I suspect that what Shadorne calls 'difficult' (and for that matter the author at the link Erik provided) is the simple fact that solid state amps (even ones known for driving difficult loads) can't make as much power into the relatively high impedance that the Quad presents at low frequencies. So you might need a 60-watt solid state amp to make the sound pressure on the Quad that a 15-watt tube amp can. But it will still play it, and if one puts one's hand on the heatsinks of said solid state amp, it will be found that the heatsinks won't get all that warm, despite hours of operation (unless the amp is class A)! IOW, the solid state amp isn't working that hard. Why? Its not a difficult load; the amp is never asked to make anywhere near the power of which its capable. The 'difficulty' might be that the Quad will easily reveal the shortcomings of any amp which is asked to play it; so to get the best out of this speaker, you need a good amp. But that amp does not need a lot of current capability, which is traditionally associated with driving 'difficult' loads. We have lots of customers with Quads and they usually use our smaller amps. Folks, 'small OTL' and 'the ability to drive difficult loads' are two concepts not normally found together :) The simple fact is, this speaker can be driven by a variety of amps (including SETs) that are not known for being able to drive 'difficult' loads. I would not use an SET in this case simply because I don't know of 15-watt SETs that can make the bandwidth that the Quads can, but if you have one, it will work fine. This is an excellent example of how not all amps are the solution for all speakers! |
This is a reason why I strive to make my speakers easy to drive. Easy to drive means they will sound great across a variety of amplifiers and electronics.+1 Not surprised by your finding as Quad ESL can be an extremely tough load. Differences in amplifier damping (output impedance) would be quite audible. This only proves one of the major design issues with Quad.This passage is false. The 'original Quad' (as bdp24 put it) is an easy load to drive. Differences in damping between amps is one of the few things that **isn't** audible on the speaker, owing to the fact that in the bass range the impedance is rather high. This reduces the difference heard between an amp with say 10:1 damping factor as opposed to one with 100:1. OTLs are traditionally thought to not be able to drive 'tough loads' but OTLs can drive the Quads with ease. The real issue here is that the Quad is pretty transparent and does not need a lot of power, so its that 'first watt' that becomes so vitally important. Many traditional solid state designs are not so good at that first watt. Put one on a 'scope sometime and look at how they behave. Usually they have much higher distortion in the first watt than they do at higher powers until they approach clipping. |