Listener fatigue: what does it really mean?

Listener fatigue is nearly always a consequence of distortion.

Two distortion sources are IM distortion and odd-ordered harmonic distortion. Both are known to be irritating to the human ear- and this has been known since the early 1950s.

What is less understood is some of the studies of the human ear that have occurred since then that relates directly to listener fatigue.

One of those things is that the ear/brain system uses odd ordered harmonics to sort out how loud a sound is. To do this it is very sensitive to the presence of odd ordered harmonics. Audiophiles have terms for the presence of trace amounts (less than 0.01%): hard, harsh, bright, etc.

This is why two stereo pieces can measure flat on the test bench but one might sound bright while the other does not.

In addition it is useful to know that the ear is tuned to birdsong frequencies. Knowing that makes its easier to understand how the ear can be that sensitive to the presence of the 5th, 7th or 9th harmonics.

The ear translates distortion as tonality. A 2nd harmonic is interpreted as warmth, a 7th as brightness.

IM (Intermodulation Distortion) is a form of harmonic distortion that the ear usually translates as brightness. This is where two more more tones can interact in the stereo system to produce other frequencies, the sums and differences of the tones involved. It is caused by non-linearities in the system, any place where the tones can interact, such as a feedback node in an amplifier, poor power supply design or a breakup in a speaker cone.

There is a special form of IM distortion called inharmonic distortion that only occurs in digital products (good luck looking for the specs on that- most manufacturers don't publish the numbers). These are intermodulations between a fundamental frequency and the scan frequency of the recording rather than harmonics based on the fundamental tone as is encountered in analog recordings. The ear treats these as brightness as well.

There is a tipping point in the human hearing perceptual system where the ear will favor distortion as tonality over actual frequency response errors. As a result is is often more important to have low distortion rather than flat frequency response. For this reason its usually best to reduce the types of distortion that the ear finds most objectionable and noticeable before going after frequency response errors.

Csontos, you are spot on.

The specs an amplifier has on paper don't reflect the whole story though. An amplifier with very low distortion with steady state signals (sine waves, used to test for distortion) can act quite differently when asked to reproduce a waveform that is constantly changing.

The feedback used in the amplifier can have a huge bearing on this phenomena. So it does not follow that the lowest distortion on paper will also be the least fatiguing.

Generally speaking, the use of feedback in an audio circuit will reduce most forms of distortion but will leave audible amounts of odd ordered harmonic distortion caused by the ringing of the signal in the circuit due to propagation delays in that circuit. In a nutshell, the feedback always arrives back at the input of the circuit slightly too late to do its job right.

The higher the frequency, the more pronounced this problem becomes. Since our ears use the odd orders to sort out how loud a sound is, essentially the use of negative feedback in an audio circuit violates one of the most fundamental rules of human hearing. To avoid this you have to avoid the use a negative feedback.

Such amps and preamps that do so will seem to have higher distortion on paper, but quite often will have less listener fatigue on this account.

Csontos, you put your finger on the leading edge of what is possible with transistors.

Here is one of the difficulties you are dealing with. Transistors have a non-linear capacitive aspect inherent in the junctions of the devices. In fact some semiconductors take advantage of this capacitance, as in the example of varactor diodes that are used in the tuning of modern FM radios.

This non-linear capacitance is magnified by the amount of current through the junction. It results in non-linearities in the amplifier that employs such devices. Now if you were reading between the lines, one way to reduce this problem is have the amp drive a higher load impedance. This will reduce the current in the driver transistors and output devices.

There are other advantages to a higher load impedance besides this one, but that is one effective means of reducing harshness in transistor amps. Of course you will get less power, but in high end audio usually we are more concerned about sound quality rather than raw sound pressure.

Now in tubes there is also an inter-element capacitance; the difference is that it is constant and unchanging regardless of signal level rather than non-linear and changing. This makes it a lot easier to build a circuit that is low in odd ordered harmonics.

This seems to me to be one of the things that has to be dealt with in a transistor design before overcoming the apparent advantages of tubes insofar as listener fatigue is concerned. One patented method to get around this problems involves heating the power transistors to a high (+100c) temperature- I have heard one of those amps and its quite impressive- but also very expensive- over $100,000 for a 100-watt/channel amplifier. Plus it made all the heat of a class A tube amp of the same power.

So it seems to me that in more practical terms tubes do have the upper hand in this regard, as its fairly easy to build a zero-feedback tube amplifier. That is hard to do with transistors; but Ayre and the Pass First Watt amps are examples, and IMO these are some of the best transistor amps made today.

Csontos, the 2nd harmonic is considered musical to the human ear (contributes to the coloration known as 'warmth' or 'richness'), and is not used as a loudness cue. Its the 5th, 7th and 9th that are, and they are likely to be in slightly higher numbers in a bridged transistor amp.

Petepappp, people often associate brightness with more detail. They are not the same. Detail is that which tells you more about the music. Brightness can allow you to do that, but at a price of listener fatigue. What works better is when you get more detail coupled with a relaxed presentation.

BTW, you might be interested to know that in most audiophile conversations, the word 'dynamics' can be safely replaced by 'distortion' without changing the meaning of the sentence. IOW, many audiophiles don't pick up on the fact that distortion can masquerade as 'dynamics'.

Csontos, I've seen the amp before. Its nice and simple, and I don't mind being called a purist when I say that I don't like transformers. But with the right transformers this amp might perform quite well, although it might not double power as you cut impedance of the load in half. However IMO that particular trait in an amplifier is way overrated.

Charles1dad, to your point the problem we are having with semantics is that distortion often masquerades as dynamics. This is so common with audiophiles that I no longer use the term- I use impact instead, when referring to actual changes in dynamic contrast.

An excellent example in recorded from is side one band 2 of the Soria series (RCA) Verdi Requiem. Here you can experience the dynamic range from triple pianissimo (ppp) to triple forte (fff), IOW from very quiet to very loud, such that to play it properly will bring most systems to their knees in just a few seconds.

Now, with respect to the comment of lower powered amps on high efficiency speakers by Learsfool:

SETs have a particular distortion character that is exactly what I am talking about (although other systems can do similar things). In an SET, to really hear what the amp does you need to have a high efficiency speaker. This is because the distortion of the amp vanishes at lower power levels. When there is distortion it obscures detail. This is why SETs always get good marks for low level detail.

At power levels up to about 25%, the main distortion is the 2nd harmonic, followed in much lower amounts by the 3rd and 4th harmonics. These harmonics contribute to the lush character that SETs are known for. I have often thought that SETs have become more popular with digital products, as they seem to add harmonic structure that digital often seems to lack.

At power levels above 25-30%, the higher ordered harmonics come into play. SETs will typically have about 10% THD at full power. The higher orders contain in addition to more even-ordered harmonics, also the odd orders, the 5th, 7th and 9th (used by the brain to sort out how loud the sound is). Their presence is masked to our conscious hearing by the presence of even ordered harmonics, for example the 4th, which is louder than the 5th, masks the presence of the 5th.

However music is processed by the limbic system in the brain, and the presence of the odd ordered harmonics is not missed even though consciously we are not aware of distortion. Music has a lot of transients (dynamics) which demand greater power out of the amp. If you are playing the amp on a moderate to high efficiency speaker, depending on the power capabilities of the amp and the efficiency of the speaker, the power levels above that 25-30% level will only be showing up on the transients.

What this means is that the artificial loudness cues (odd ordered harmonics) will only show up on the transients. **This is why** SETs are often cited for having dynamics that seem far above what one should expect for such a low powered amp. We have all seen those comments in plenty of reviews.

The bottom line is what we are seeing in this example is distortion masquerading as dynamics.