Please explain amplifier output impedance

Sean, well said!

There is no universal correct amplifier output impedance. This is why we and others offer a means to adjust output impedance/damping on some models. Many system designers have known this for years, and damping adjustment controls were found on some amplifiers going back to the 1950s or before. The reason that there is no one correct output impedance is because the speaker is a mechanical system with inertia and its transient response is very much determined by the source impedance driving it. Each speaker design behaves differently, and it is usually desirable to achieve what's called "critical damping". If the mechanical system (speaker) is under damped, the cone will overshoot and ring in response to a step transient. If it is overdamped, it will undershoot and be slow to reach its proper position. Underdamping happens when the output impedance of the amplifier is too high, where as overdamping occurs when the output impedance is too low. In terms of the bass reproduction, underdamping results in an overly warm resonant reproduction, and overdamping results in a overly dry and heavy presentation. Neither of these are correct. When the amplifier output impedance is optimum, the bass will have its most natural reproduction. A damping/feedback control on the amplifier lets the user best tune the speaker, and yes, the room acoustics enter into it as well.

I made a gadget for observing the optimum damping of a speaker cone in response to a step transient driven from various impedances. The device uses a lightweight conductive foil that can be taped to the cone of the driver, and a matching RF based capacitance probe is used to detect cone movement and display it on an oscilloscope. I can confirm various degrees of over and under damping, and have tested speakers that range from wanting to see as high as a 7.5 ohm amplifier output impedance (this was a large driver and quite unexpected) to speakers that are still underdamped with zero output impedance. I find that most high-end speakers fall in the range of working best with 2-4 ohms amplifier output impedance.

Of course, this is the bass driver movement and is easiest to measure. The higher frequency drivers in a particular speaker may want to see something different, and it seems still impossible to be completely scientific in determining what is really best when the whole frequency spectrum is taken into account. This is why it is really best to simply listen with an open mind, and not assume that there is a spec that will really tell you how good or accurate a speaker-amplifier combination is going to be.

Over an over again people get hung up on worrying about the issue of a high output impedance amplifier impedance affecting the frequency response. It is very easy to try to explain the sonic differences by making the argument that the frequency response of the system wobbles all over the place because the amplifier output voltage goes up at frequencies where the speaker impedance rises. Yes, this does happen, but how flat do these people think the frequency response in the listening room is if this were not to happen. Yes, if the output impedance of the amplifier was 50 ohms, and a very underdamped speaker reached a 50 ohm impedance, there would be a sharp peak in the system response. Whereas some speaker drivers can reach 50 ohms, an amplifier with an output impedance of as high as 5-10 ohms will largely tame it (but perhaps not properly damp it). In practice, the system frequency response flatness will likely have the similar degrees of deviation for all amplifiers from zero to 10 ohms output impedance. The particular frequency points of maximum and minimum response can be expected to shift around slightly, but this alone would not account for a significant enhancement or degradation of the sound. They will sound very different though, but because of the damping.

It should be noted that Mr Berning is a manufacturer of audio gear. As far as i know, all of his tubed amplifier designs are of a VERY high output impedance, hence his propensity to defend this type of product.

Those that are interested in such subjects may want to look at some "reasonably unbiased" test results as conducted and published by Stereophile on this very subject. The article titled "Questions of impedance interaction" has several different graphs documenting the linearity of four different amplifiers driving three very different speakers.

Remember, these graphs are of the actual amplitude linearity / loading characteristics / frequency response of the amp going into the speaker, NOT the measured response AFTER the speaker output. In effect, you are seeing how various amps deal with various loads, not the actual output of the speakers as driven by those specific amps.

Given the phenomenally distorted / non-linear frequency response / loading characteristics of some of these amps, the output response of the speakers would be even more distorted / non-linear. After all, the speaker can only try to reproduce what is fed into it, and if the amplifier is feeding a phenomenally non-linear signal into the speaker, one can only expect further distortions after the speaker.

As can be seen in the test results, the two low output impedance solid state amps ( Hafler & Aragon ) remain extremely consistent in amplitude output ( linearity of frequency response ), regardless of the load they are presented with. While these would commonly be referred to as a high damping factor design, one can see the benefits of using a very low source output impedance as compared to a higher ( speaker ) load impedance. That's because the difference in impedance between the source and the loudspeaker load acts as an "impedance buffer".

This "impedance buffer" (or "higher damping factor") reduces the potential for the varying impedance that the speaker presents to "load down" or "modulate" the output of the amplifier. This can be seen as output remains quite consistent, regardless of frequency or the impedance that these amps are presented with.

The Melos tubed amp results are kind of tricky. In the text of the article, Mr Norton states the following: "Since the Melos 400 also had a relatively low output impedance for a tube amplifier (at 0.43 ohms at low and mid frequencies, rising to 1.2 ohms at 20kHz, from its 8 ohm tap), I took that opportunity to run some frequency-response measurements using an actual loudspeaker as the load for the amplifier".

While this would seem to be a pretty stable design as far as tubed amps are concerned, and the results somewhat confirm that other than a sloping high frequency response, the results are somewhat misleading. That is, the Melos might have an output impedance ranging from .43 ohms up to 1.2 ohms as frequency varies, but this is on the 8 ohm tap. If you re-read the text pertaining to the actual testing on this amp, the results that Norton chose to publish were based on the 4 ohm tap of this amp.

This is somewhat deceiving and most people aren't technically sharp enough to catch something like this. One has to question why the tests and wording were published in the manner that they were. This is why i said that the test results were "reasonably unbiased". I don't know the reasons of why Mr Norton did or worded things the way that he did, but the text and test results were not consistent pertaining to this one specific brand of product.

Using the approach that Norton did i.e. choosing a lower impedance tap on the output transformer, the transformer itself and the tubed output stage sees less stress. In doing so, there's more of an " impedance buffer" built into this approach, hence the pretty decent output linearity / loading characteristics presented by this amp under this set of test conditions. As a side note, distortion was probably also drastically reduced due to the reduction in output stage modulation by using the tap that he did.

Now, take a look at the results of the tubed Sonic Frontiers amp. This amp has an output impedance of above 3 ohms, resulting in a damping factor of less than 2.7 into an 8 ohm load. As can clearly be seen, the lack of "impedance buffer" or "damping factor" is extremely evident. The amplitude linearity / frequency response into any of these loads is quite poor, especially when compared to the low output impedance SS designs.

As can be seen, the varying impedance of the speaker directly coincides with the varying loading characteristics of the amp. In some cases, the linearity of this amp is inferior to the frequency response of what many would consider even a "decently flat" loudspeaker. Driving the Martin-Logan's, the actual frequency response of the amp looks more like a scenic shot of hills and valleys than a piece of "reference grade audiophile approved" electronics.

With this set of speakers, the Sonic Frontiers amps is down well over -5 dB's by 20 KHz with a massive depression in the warmth, lower midrange and midrange frequencies. All of this with an amplifier that has an output impedance of slightly over 3 ohms. Obviously, this type of situation is an extreme example of why "system synergy" becomes SO important in certain installations.

Obviously, this type of performance puts the Sonic Frontiers ( and other amps like it ) into the "specialty category". That is, they are FAR from being universal in application. This is due to the inability to drive various loads with any reasonable form of amplitude linearity. As a side note, it should be noted that a lack of amplitude linearity is NOT measured as a part of THD, IMD, etc..., but is a measure of "frequency response tolerance". In effect, it is a distortion all to itself.

For sake of reference and comparison, the test results that i've seen for one of Mr Berning's products showed the amp as having an output impedance of appr 5 - 10 ohms or so ( can't remember the specifics ). This could be why he included the following in his response: "In practice, the system frequency response flatness will likely have the similar degrees of deviation for all amplifiers from zero to 10 ohms output impedance".

All i can say to that is "HA HA HA". If such were the case, all amps would sound pretty similar on any given set of speakers because the SYSTEM frequency response would all be equally flat. As was verified in the Stereophile test results, this is obviously NOT true. If such were the case, there would be very little reason to buy one amp over another, as the only difference would be power ratings and potential spl's. As most all of us frequenting this forum know, such is simply not the case.

On top of that, Mr Berning also stated the following: "Yes, if the output impedance of the amplifier was 50 ohms, and a very underdamped speaker reached a 50 ohm impedance, there would be a sharp peak in the system response". While MANY speakers do have a 50 ohm peak in the low frequency region, especially vented designs with large woofers, most amps do not come anywhere close to having a 50 ohm output impedance. That makes this analogy useless in the real world.

Getting back to the real world and applying a high level of logic to what Mr Berning did state, let's refer back to the Stereophile article i previously referenced. As was seen in that articles test results, we could definitely see that as the output impedance of the amp was brought closer to the input impedance of the speaker i.e. reducing the "damping factor", the response became more peaky / less linear. That's because of the lack of "impedance buffer" that i previously mentioned. In an audio amplifier, the response would get even more ragged looking / non-linear if the output impedance of the amp actually matched the input impedance of the speaker. This would be due to having a near zero damping factor / impedance buffer.

As such, Mr Berning was not exagerating in his 50 ohm - 50 ohm example and Stereophile's test results tend to confirm this. Applying further logic to what Mr Berning stated, he is basically telling us that the response of his own amp, with a measured output impedance of 5 - 10 ohms, will produce very peaky, non-linear response when driving a speaker with a nominal impedance of 5 - 10 ohms. As poor as the Sonic Frontiers amp performed with a "damping factor" of 2.7 into an 8 ohm load, the Berning amp would have a "damping factor" of near ZERO at 8 ohms over part of the audible bandwidth. How linear do you think that would be?

Obviously, we are talking about another "specialty" product here, which i don't have a problem with. People buying such a product should be familiar with what they are buying, so long as the manufacturer and sales people are honest and ethical about the performance potential of the product and what it takes to make it work as they planned.

In that respect, Mr Berning has been honest and forthright about at least one of his products, and we should thank him for contributing to this thread. His contributions allowed us to dig deeper into this subject and actually make some real world comparisons in an apples to apples format, courtesy of the Stereophile test results. Sean
>

One more thing. Sometimes i'm simply debating something based on the technical merits of the subject at hand. I do this from an educational standpoint, and to be honest, because i like to debate. There's nothing wrong with discussing opposing points of view, having some fun and possibly learning something along the way. : )

As i've said many times in the past, i only know what i do because others have taken the time to share their knowledge and try to educate me. I'm trying to do the same thing. Sometimes, my methods may not be the best or most "friendly", but then again, i've never claimed to be "PC" or a "teacher" by trade. I'm simply a big-mouth sharing my point of view, so take it for what it's worth : )

Having said that, my pointing out what "I" consider to be technical flaws in a product does not mean that one can't like / listen to / enjoy a product that i have attributed these specific "negative" connotations too. As i've said before, one should buy / use / listen to / enjoy what brings them closer to the music and their system goals. The owner of the system is the only one that has to be pleased with the end result, negating most any other comment that i or anyone else could ever make about a given product or installation.

As a case in point, i typically tend to like the sound of Cary tubed amps, even though i don't own any of them. I know that they are quite coloured and less than linear into most loudspeaker loads, primarily due to their high output impedances and some other factors. Like some of the products mentioned above, these items fall into the "specialty" products category and i treat them as such. They can require very careful attention to detail in terms of selection of mating components to say the least.

Having said that, the end result of installing one of these amps ( or something similar ) into a system can be a very luxurious and smooth sound with great spatial characteristics. While i know that i'm NOT listening to strictly what was on the record and am hearing quite a bit of "system artifacts", it doesn't make it any less enjoyable.

I just wanted to put things into perspective, as i know that some will be upset with my post above for various reasons. If it upsets you, blow it off and consider the source : ) Sean
>

Well let's see how long this thread lasts. I'm betting on late Sunday afternoon and I have no dog in this fight.