Intermodulation Distortion Values below THD Values by SS Amps - is this possible ??

[tiefbassuebertr]

In most cases the mentioned IM values are above the THD values. But there are amps, where this is reversed, e. g. by the Altec 9444B, go to
Altec Lansing Professional Amplifier 9444B Specification List page 6
Is this possible in general, and if yes, what means this for the audible perception by listening tests?
So far I have never noticed this detail.

[Elvee]

It is not only possible, but even very likely.
Some examples are discussed here:
Low distortion oscillator?
And you can see that for a severe case of crossover distortion, the IMD is practically non-existent.

At least the first order IMD, which is nearly always what is specified.

[DF96]

Intermodulation and harmonics are both created by nonlinearity, so if other things are equal they will vary together. Other things are not always equal. By definition, harmonics are at a higher frequency than the input signal so may be affected differently by feedback, as in SS amps this reduces at higher frequencies. So there is a mechanism for harmonics to be greater than IM.

On the other hand, problems with local decoupling can mean that second-order distortion products at low frequencies can feed back in to produce third-order IM (even if third-order distortion is otherwise low). So IM can be greater than harmonics.

[WuYit]

Hi,
the IMD level is several times higher than the harmonic distortion, in particular at high frequencies.

[DF96]

No. Take an amplifier with a flat frequency response, and some second-order nonlinearity. Apply a signal at frequency f1 - you get some DC shift and the second harmonic at 2f1. Now apply a signal at the same level but frequency f2 - again you get DC and 2f2. Now apply both signals together. You get twice the DC shift, both harmonics at the same level as before, and intermodulation at f1-f2 and f1+f2 at the same level as the harmonics. Why is it the same level? Because it is created by exactly the same process: multiplication of trigonometric functions. f1-f2 corresponds to the DC shift, and f1+f2 corresponds to the second harmonic.

[tiefbassuebertr]

Originally Posted by Elvee

It is not only possible, but even very likely.
Some examples are discussed here:
Low distortion oscillator?
And you can see that for a severe case of crossover distortion, the IMD is practically non-existent.
At least the first order IMD, which is nearly always what is specified.

Thank you for this advice. This means for me, that to a certainly THD spectrum there is an associated intermodulation distortion spectrum.

[WuYit]

DF96,
unfortunately yes; the issue is a bit more complicated than in your presentation. A (more advanced) mathematical approach would certainly be interesting to show how the number of intermodulation products vastly exceeds the number of harmonic distortion products, the disproportionality rapidly growing with nonlinearity.

[Elvee]

To go further, here is another example: an underbiased class B amplifier.

Nothing special, nothing exotic, just a very average, very common type of non-linearity.

You can see that the third harmonic is 42dB down (THD is 1.32%).

Now, the IMD test: the first order IM product between the 10KHz and 11KHz is 55dB down.

Of course, if you take into account higher order products, the comparison is more balanced, but when an IMD figure is given, it applies to one product, not the entire spectrum.
The THD figure, as its name implies, applies to the whole spectrum.

[DF96]

Originally Posted by WuYit

the number of intermodulation products vastly exceeds the number of harmonic distortion products

Only when the input signal includes lots of separate signals. This is just a matter of combinatorics. Your original claim was that the level of IM was greater. Are you now revising that to the number of IM?

[WuYit]

No revising. I could have used my preferred term: energy of distortion product.