Please explain amplifier output impedance

Sean's post above is right on but I would like to add a couple important technicalities.

First and most importantly, amplifiers are tested by STATIC means. This means two things: The test signal is continuous and repetitve, e.g., a sine wave or square wave. Secondly, the load is a resistance and not an IMPEDANCE. The difference between the two gets butchered all the time but basically, impedance takes capacitive and inductive effects into account whereas resistance does not. Many times, people use the word "impedance" but if you take a close look, they are actually using an "averaged resistance" at best. This is incorrect use of terminology but it runs rampant, especially here. :)

The sum total is that music is a very dynamic signal that is constantly changing. The speaker's impedance, in most cases, is a ridiculous mess of ups and downs. Combine the two and you get drastically different damping factors, reflected waves and varying slew rates at different points in time AND for different frequencies. None of this information is faithfully represented by the manufacturer's specs.

But as we already know, you have to listen to get a feel for how an amp/speaker combo works - and that some who measure great, fall short in reality due to poor handling of dynamics. Listening is the best feedback on performance you can get because only then are all the real variables taken into account.

The issue of negative global feedback is different than that of negative local feedback. The two are, again, not to be confused. Global feedback puts the entire amp in the loop whereas local feedback is only for the active devices. This latter one is always required for very good stability but the former is optional, depending on the quality of component matching, parasitic inductances, capacitive coupling, type of active devices and layout quality.

I have looked at the output impedance curves of a few amplifiers using an impedance analyzer. In the frequency domain, most of them are very flat but have an inductive rise at high frequencies (>50kHz or so).

Arthur

Negative feedback in high levels is quite audible and it doesn't sound good.

Agreed. A badly designed circuit will sound terrible. Most manufacturers try to avoid building unstable circuitry. However, in the pursuit of ridiculously high damping factor specifications (for marketing purposes), it is certainly possible to build a dangerously unstable circuit. Extremely high amplifier gain will lead to instability and oscillations. This occurs when feedback times open loop gain approaches negative one. In this case, the closed loop gain will approach infinity...which is of course not possible and everything becomes oscillatory, distorted and clipped.

In general, typical SS amplifier circuitry (with negative feedback loops), although very linear when operated within tolerance, are not at all forgiving when they are over-driven; typically when over-driven they sound harsh and then damage speakers fairly quickly. Certainly, high levels of negative feedback are likely to lead more quickly to catastrophic behavior.

Since music is very dynamic, it is relatively easy to over-drive equipment. Some SS gear has built in protection circuitry that is designed to detect and protect equipment from damage.

Arthur: Those are very good points and well worth clarifying. Thanks for taking the time to point them out AND explain them.

I have often said that it is the sum of manufacturer spec's that count more than any individual spec by itself. Even then, most manufacturers don't provide the quantity of spec's that one needs to make such information truly useful.

As to your comment about most amps having a linear output impedance up to appr 50 KHz, that is kind of generous in my experience. Many amps exhibit a noticeable increase in output impedance at or slightly above 10 KHz. How severe this is will depend on the design of the amp. By 50 KHz - 80 KHz or so, performance is starting to suffer quite noticeably. This is why many amps round the leading edge of a 10 KHz square wave. That is, the higher output impedance is part of a bigger problem i.e. limited bandwidth due to the amp being too slow to properly respond. Combine the limited bandwidth / lack of speed with the rising output impedance and you end up with that slightly rounded square wave that you see so often in Stereophile test measurements.

If you think that this sounds "bad", there are REALLY slow / limited bandwidth / higher output impedance amps fail the 1 KHz square wave test. When this type of amp encounters a very fast high energy high frequency transient, most of the attack, definition and duration is lost. This translates into a soft sounding blur, which some people like. This is probably more true with digital recordings and playback, which tends to sound hard, bright and glaring in many systems.

Other than that, this thread could go on and on contrasting various designs and goals. Suffice it to say that there are a LOT of variations that come into play with any / every design. When all is said and done though, the end result is a summary of what the designer / engineer thought was most important. Whether or not you like that product will depend on your own personal preferences and how well that specific component blends with the other gear in your system. As far as i know, there are no spec's to quantify personal preference. Sean
>

To bring this discussion back around to my original question, when a reviewer states that a particular loudspeaker would likely work best with a low output impedance amplifier, should one assume (based on comments made earlier in this thread), that a low output impedance measurement is .1 ohm? At what measurement (expressed in ohms) does the threshold cross from low output impedance to marginally high output impedance...1 ohm?

There is no hard and fast threshold but IMO, anything over 1 ohm is definately high output impedance. Many SS amps will be about 10x lower and tube amps are up to 10x higher so 1 ohm threshold seems to a decent rule of thumb.

Sean - My comment about amp impedance rising above 50kHz was based on at least a 3dB rise so it would visibly start to deviate around 20kHz. And you are definately correct in saying that many amps can't even do that well. I read all the Stereophile measurements several times each, no matter the piece of equipment. I am happy they are still around to do the good old fashioned measurements, even if I realize they aren't very applicable to musical reality. Nonetheless, I feel there is a lot to be said for an amplifer with an elegant set of curves.

Arthur