There's been a few threads about THD and its pros and cons and they always devolve into philosophical discussions.
So what about IMD (InterModulation Distortion)? I came across an article by Nelson Pass where he argues that it's a much bigger problem than THD and I agree with him, since IMD is a non-musical distortion. As we all know, all musical instrument produce harmonics, so harmonic distortion blends into the music, whereas IMD is going to be "off key".
I did a crude measurement of the IMD using the SMPTE method on my latest build. It came out to about 0.02% on a resistive load and slightly higher on a speaker. I think it was at around 5W. Is that good? Should I be worried? How do you mitigate IMD? Does a high OLG increase the risk of IMD? I know that Pass argues that you should get rid of the feedback altogether, but that opens a whole new can of worms.
I'm here to learn from you smarties. 😀 (And comments like "Don't measure, just listen" aren't useful. I want to understand the theory)
So what about IMD (InterModulation Distortion)? I came across an article by Nelson Pass where he argues that it's a much bigger problem than THD and I agree with him, since IMD is a non-musical distortion. As we all know, all musical instrument produce harmonics, so harmonic distortion blends into the music, whereas IMD is going to be "off key".
I did a crude measurement of the IMD using the SMPTE method on my latest build. It came out to about 0.02% on a resistive load and slightly higher on a speaker. I think it was at around 5W. Is that good? Should I be worried? How do you mitigate IMD? Does a high OLG increase the risk of IMD? I know that Pass argues that you should get rid of the feedback altogether, but that opens a whole new can of worms.
I'm here to learn from you smarties. 😀 (And comments like "Don't measure, just listen" aren't useful. I want to understand the theory)
My impression is that it is only an issue when you have too less NFB, just as all the designs that Nelson Pass does. When you have 60dB NFB over 1KHz, all distortion are so low that you don’t need to care.
By mitigating HD? Same physical non linearities are responsible for both. If you have a complete knowledge of the HD spectrum at different amplitudes i bet you can mathematically derive all sorts of IMD spectra from it. Other than offering a different visualization i am not at all sure IMD offers any additional data.
Interesting. Nelson Pass seems to treat them as two distinct issues in this article: https://www.passdiy.com/project/articles/audio-distortion-and-feedback But what you say makes sense: To get intermodulation you need nonlinearities, and nonlinearities will show up in the HD.
A quick experiment shows that the IMD is pretty much proportional to the THD. So that's all there's to it? Are there specific linearities in specific parts of the amplifier that are more prone to generate IMD? I've heard people who poo-poo THD as a yard stick proposing IMD at a better metric.
Low THD (proper bias)
High THD (no bias)
Low THD (proper bias)
High THD (no bias)
*T*HD can be a poor metric. .1% 2HD with .01 or less at all others sounds better than .07% 3HD and .03% 5 and 7HD. WHY? Because 2nd order IMD (same mechanism as 2HD) is almost inaudible as it only affects phase slightly and 3rd order IMD (same mechanism as 3HD) produces audible products that are not in the original spectrum. 5th and 7th or IMD can produce even more products, and they don’t all just push above 20 kHz where you can’t hear them.
