Damping Factor and Overall Negative Feedback.

No speaker made needs over 20:1 damping factor. Most of the time the extra damping isn't a problem but some speakers (certain open baffle designs) might need considerably less to sound right.

I heard an example of the extremely low damping factor amps made by Nelson Pass at RMAF 10-15 years ago. The amp is a current source rather than a voltage source so its output impedance is very high- a multiple of that of the speaker itself. But with the right driver in the open baffle design he was showing off, the bass response, especially considering the small size of the flat baffle, was really impressive.

Most amps of 'high' output impedance tend to act more like power sources (mid way between a voltage source and a current source); if you have a tube amp with zero feedback that is how it will try to behave.

Once upon a time all amps were 'power sources'. But to use them properly you have to have controls on the back of the speaker to adjust the speaker to the voltage response of the amp. JBL, Altec, EV, Klipsch, early KLH, with those controls on the back are examples. These days joined by Sound Lab ESLs, Classic Audio Loudspeakers and quite a few others. But in terms of overall speakers made in the world are less than 1% made.

 

Hi, I understand this post is from over 10 years ago, but could you please elaborate further on why, in areas where feedback could be beneficial, a good design calls for minimum or no feedback? It seems like it should be the other way around until the feedback reaches a balanced level, avoiding excess or deficiency, right?

to the recent responder, i’m not a big fan of negative feedback as Al eloquently stated because of the temporal distortions.

@tomic601 @lanx0003 I miss Al too- he was a good friend.

The 'temporal distortion' is usually called 'phase shift' and is more a function of the bandwidth of the amp rather than the feedback itself.

Feedback has a bad rap due to how its been traditionally applied in audio. Lots of feedback (and this coming from a manufacturer who has made zero feedback amps for the last 45 years) isn't bad. But applying lots of feedback poorly is.

Here are the two primary reasons designers get in trouble with feedback, and these are known problems that for whatever reason have been ignored. The first was written about by Norman Crowhurst back in the 1950s, and it was his observation that feedback was being applied to a non-linear point near the input of the amp (or preamp) that was causing the feedback signal to become distorted before it could do its job. An example he gave which is common in tube amps today is the cathode of the input tube. But it could be the base of an input transistor in a solid state amp (as another engineer, Peter Baxandall, pointed out 20 years later) as well. The transistor in question is usually part of a differential pair at the input of the amp. The other transistor of that pair, the input transistor, is typically outside the feedback loop!

If the feedback signal is distorted, higher ordered harmonics are generated on that account, as well as intermodulations forming at the feedback node itself. Not good- the ear interprets higher ordered harmonics as harshness and brightness. So the appearance is that feedback causes that. But its not that simple.

The other problem is the amplifier design might lack something called Gain Bandwidth Product. This is actually quite common; all tube amps have this problem (if they have feedback) and most solid state amps do too.

GBP is defined as the frequency at which the gain of the amp has fallen to unity gain and so is the highest frequency the amp can pass without excess distortion if running open loop (no feedback). Feedback takes away from the overall gain of the amp and the amp needs gain to amplify the signal so it can be driven to full power with a normal preamp. So if the amp is higher power like 200 Watts it might need 30dB of gain; if there is 20dB of feedback the overall gain of the circuit needs to be 50.

The issue becomes one of bandwidth and gain at the same time. If the bandwidth isn't there, what happens is that at some frequency in the audio band the GBP no longer supports the feedback, which thus decreases on a 6dB slope at first and probably faster as frequency is increased. As a result, distortion rises on a complementary curve. If less overall gain is used in the amp this might push this turnover frequency higher. Alternatively using less feedback will have the same effect.

One of the reasons zero feedback amps can 'sound better' is that the distortion vs frequency curve they have is a ruler flat line across the audio band. If the distortion rises, for example starting at 1KHz, at 7KHz (7th harmonic) the distortion can be higher than the THD measurement suggests! The 7th harmonic isn't musical and is interpreted by the ear as harshness and brightness. So you can see that if an amp has this problem (and most amps with feedback do) that brightness is often the result (this being because the ear interprets all forms of distortion as a tonality).

If you are still with me, the solution to adding feedback is to mix it with the audio signal prior to the audio signal reaching the input of the amp, in the same way that its done with opamps. Then the amp also has to have adequate GBP such that distortion does not rise in the audio band. If the design can meet these two criteria, you can have as much feedback as you want and it stands a good chance of being musical. Mess either one up and you have a problem.

The semiconductors needed to fulfill these requirements didn't exist until sometime in the 1990s. You also needed the will and knowledge of the designer to build such an amp. Class D, which has been a rising star in the last 2 decades, offers a nice tool in this regard as its very easy to generate very high GBP values. As a result there are class D amps now that have very high feedback values and yet are smooth and musical like the best tube amps, but lower distortion overall (so are more transparent).

Put simply, feedback is good unless poorly applied.

That results in what is called transient intermodulation distortion (TIM), and enhancement of particularly objectionable odd-order distortion components.

It is this last bit wherein you see so much variance in amplifier design: odd orders as stated are very objectionable, in no small part because the human ear/brain system uses them to determine how loud a sound is. It is arguable that our ears are more sensitive to this than actual frequency response errors- so electronics that distort these harmonics will have a false loudness and brightness to them. Audiophiles use terms like 'hard', 'harsh', 'brittle', 'bright' to describe this phenomena.

This is why of two amps that have flat frequency response, one might sound bright and the other might not.