Let us concentrate on the topic.
Generally speaking, IMD will give more information about a system, because higher order products will not necessarily result in DC or duplicate components.
But, when the performance of an amplifier or system is characterized by a single figure, the THD figure is much more relevant than the value of a single IMD product, because it encompasses all of the higher products.
This is clearly shown in the previous simulations.
It is extremely difficult to find realistic examples where the IMD figure is much higher than the THD figure.
jcx gave an example, but it is pretty unrealistic, it is just (cleverly) hand crafted to support his argument.
By contrast the example I gave is based on normal amplifier circuits, and shows the effects of typical distorsion mechanisms.
Generally speaking, IMD will give more information about a system, because higher order products will not necessarily result in DC or duplicate components.
But, when the performance of an amplifier or system is characterized by a single figure, the THD figure is much more relevant than the value of a single IMD product, because it encompasses all of the higher products.
This is clearly shown in the previous simulations.
It is extremely difficult to find realistic examples where the IMD figure is much higher than the THD figure.
jcx gave an example, but it is pretty unrealistic, it is just (cleverly) hand crafted to support his argument.
By contrast the example I gave is based on normal amplifier circuits, and shows the effects of typical distorsion mechanisms.
DF96,
High oder nonlinearities produce increasingly more IMD. The combinations of intermodulation rise fast with the number of input tones and the order of nonlinearity. Just a moderately high order nonlinearity with a moderate number of input tones can produce tremendously more IM products than harmonics. A single-tone pure sine wave test does not excite much nonlinearity and is very far from the harsh reality (actually worthless).
That`s what my memory suggests, but I am not able to prove it. I`d like to know this correctly, but have little knowledge and interest in elaborate mathematical descriptions. To me, the distortion perception and audibility aspect is much more relevant than numbers. While low order harmonic and IM distortion constituting a considerable amount of energy is perceived as benign, switching transients consisting of high frequency IM products having very small energy are devastating. Distortion due to clipping is more tolerable as well.
High oder nonlinearities produce increasingly more IMD. The combinations of intermodulation rise fast with the number of input tones and the order of nonlinearity. Just a moderately high order nonlinearity with a moderate number of input tones can produce tremendously more IM products than harmonics. A single-tone pure sine wave test does not excite much nonlinearity and is very far from the harsh reality (actually worthless).
Nothing elaborate. Just a bit of trigonometry. In my day this was part of our school leaving exam, so I would probably have learnt the necessary maths when I was about 14 years old. People without this maths will never understand amplifier distortion.
Agreed. As I said, it is a matter of combinatorics.
Not true. A full amplitude sine wave will exercise most sources of non-linearity. The main exception is the re-entrant distortion caused by poor decoupling which I mentioned in an earlier post. Two sine waves at similar frequencies will excite that one. IM may be easier to measure than harmonics, and it may sound worse, but it is essentially just another manifestation of the same non-linearity.
Hi,
not having the limitations and defects of technical analysis, mathematics is a great and precise tool to describe the physical reality.
Distortions rise significantly with signal complexity, algorithmically, the function is strongly dependent on the sample and sinusoidal waves and music signals have very different statistical properties.
There is no general relationship between the level of harmonic distortion and IM distortion. Clearly, IM distortion is chiefly determined by amplifier (dynamic) high frequency performance.
not having the limitations and defects of technical analysis, mathematics is a great and precise tool to describe the physical reality.
Distortions rise significantly with signal complexity, algorithmically, the function is strongly dependent on the sample and sinusoidal waves and music signals have very different statistical properties.
There is no general relationship between the level of harmonic distortion and IM distortion. Clearly, IM distortion is chiefly determined by amplifier (dynamic) high frequency performance.
"algorithmically" - are you suggesting that the amplifier somehow calculates via an algorithm how much distortion to generate?
There is a simple relationship between the level of harmonic distortion and IM distortion. Other things being equal, they would be at exactly the same level, as they are usually generated by the same mechanisms. Other things are not always equal, as I have said, hence they can vary from each other. The number of distortion products, but not their level, obviously relates to the number of distinct elements in the input signal (which is perhaps what you mean by "signal complexity"). IM distortion is not chiefly determined by high frequency performance, as IM can occur at any frequency. I am repeating myself, so there seems little point in continuing this discussion. Please read a good book on trigonometry, then you will begin to understand amplifier distortion.
There is a simple relationship between the level of harmonic distortion and IM distortion. Other things being equal, they would be at exactly the same level, as they are usually generated by the same mechanisms. Other things are not always equal, as I have said, hence they can vary from each other. The number of distortion products, but not their level, obviously relates to the number of distinct elements in the input signal (which is perhaps what you mean by "signal complexity"). IM distortion is not chiefly determined by high frequency performance, as IM can occur at any frequency. I am repeating myself, so there seems little point in continuing this discussion. Please read a good book on trigonometry, then you will begin to understand amplifier distortion.
I assume, the main reason, why in most cases the mentioned IM values are above the THD values is the fact, that the mentioned THD value was only checked by 1000 Hz (especially by older publications), while the IM value was checked by 19/20 KHz.
The reason for that is again that most engineers generally assumed, that the THD value by 20 KHz is of no interest, because the associated distortion components are all in the inaudible range.
This fact I didn't notice at first glance.
The reason for that is again that most engineers generally assumed, that the THD value by 20 KHz is of no interest, because the associated distortion components are all in the inaudible range.
This fact I didn't notice at first glance.
Last edited:
That s right..
An amplifier is an analog multiplier, basically.
That s also right.
Generaly, THD and IMD are of the same order, amplitude wise.
A couple of words on an old thread:
A 19Khz-20kHz twin-tone IM measurement is primarily a measure of an amplifier's linearity at 1 kHz. Secondarily, it measures linearity at 19/20 kHz.
It goes like this. The non-linearity at 19/20 kHz causes an error that has a 2nd order (f2-f1) product at 1 kHz. The amplifier's loop gain at 1 kHz corrects this error. What's more this distortion is of the same order as a 2nd harmonic. If the amplifier has low 2nd order harmonics (a bridged amplifier, for instance), but high 3rd harmonic then THD will be higher than the CCIF-2 IM measurement.
A 19Khz-20kHz twin-tone IM measurement is primarily a measure of an amplifier's linearity at 1 kHz. Secondarily, it measures linearity at 19/20 kHz.
It goes like this. The non-linearity at 19/20 kHz causes an error that has a 2nd order (f2-f1) product at 1 kHz. The amplifier's loop gain at 1 kHz corrects this error. What's more this distortion is of the same order as a 2nd harmonic. If the amplifier has low 2nd order harmonics (a bridged amplifier, for instance), but high 3rd harmonic then THD will be higher than the CCIF-2 IM measurement.
in the meantime I think, THD- and IM values don't determine the sound quality of any amplifier.
check out Fig. 4 under
Linn Klimax 500 Solo monoblock power amplifier Measurements part 2 | Stereophile.com
and fig. 4 under
Pass Aleph 1.2 monoblock power amplifier Measurements | Stereophile.com
so as the IM spectrum and THD+N
Without listening test only distortion and noise waveform (with fundamental notched out) provides any advice to the expected sonic quality in the frequency area above approximately 0,7-2KHz.
Linn's Klimax SOLO 500 vs. Pass Labs Aleph 1.2 - even without a listening test against each other, the decision should be actually an easy task.
Check out also this thread:
Choosing of best sounding OP AMPs for the lowest possible THD+N -really the best Way?
check out Fig. 4 under
Linn Klimax 500 Solo monoblock power amplifier Measurements part 2 | Stereophile.com
and fig. 4 under
Pass Aleph 1.2 monoblock power amplifier Measurements | Stereophile.com
so as the IM spectrum and THD+N
Without listening test only distortion and noise waveform (with fundamental notched out) provides any advice to the expected sonic quality in the frequency area above approximately 0,7-2KHz.
Linn's Klimax SOLO 500 vs. Pass Labs Aleph 1.2 - even without a listening test against each other, the decision should be actually an easy task.
Check out also this thread:
Choosing of best sounding OP AMPs for the lowest possible THD+N -really the best Way?
Last edited:
Hi
I haven't read the whole thread so this might already have been mentioned: IM on a first principle is simply the sum and difference of two frequencies. However, the number of harmonics generated this way extends all up and down the frequency spectrum and is truly an impairment to fidelity.
You can imagine just having some rectification noise and a single audio frequency and that intuitively there should be some "maximum" number of IM products? Then add some THD and the interaction between those harmonics and the mains and then with each other. It becomes quite a mess.
The big problem with IM is that it is not harmonically related to the music frequencies and is therefore very fatiguing to our brain to get rid of it or filter it out. Even with a very low IM figure an mp can sound grating.
As others have said, every measure that reduces THD will also reduce IM, so the goal should be to reduce THD as much as possible and IM reduction will track that.
I haven't read the whole thread so this might already have been mentioned: IM on a first principle is simply the sum and difference of two frequencies. However, the number of harmonics generated this way extends all up and down the frequency spectrum and is truly an impairment to fidelity.
You can imagine just having some rectification noise and a single audio frequency and that intuitively there should be some "maximum" number of IM products? Then add some THD and the interaction between those harmonics and the mains and then with each other. It becomes quite a mess.
The big problem with IM is that it is not harmonically related to the music frequencies and is therefore very fatiguing to our brain to get rid of it or filter it out. Even with a very low IM figure an mp can sound grating.
As others have said, every measure that reduces THD will also reduce IM, so the goal should be to reduce THD as much as possible and IM reduction will track that.
Decision for whom? People have widely divergent ideas of good and bad. Very hard to tell who would prefer what. Are you just expressing prejudice?
I don't think so in cases, if listening mainly acoustical instruments and voices (Aleph 1.2 then is clearly better in sound).
But in cases, if listening mainly heavy metal resp. rock music and electro pop/techno, you are right; there is no need to predict which power amp device will be preferred according the THD waveform character (fundamental notched out).
But in cases, if listening mainly heavy metal resp. rock music and electro pop/techno, you are right; there is no need to predict which power amp device will be preferred according the THD waveform character (fundamental notched out).
Hi
tiefbassuebertr's statement regarding rock music requirements is wrong and seems like he has an aesthetic bias.
The fact is that most pop and rock and even rap is very dense in the midrange frequencies, which is also where our hearing is most sensitive. It follows that the highest fidelity possible is required to make this come across as intelligible music and not just a smear of mud.
Contrary to that requirement is what is needed for very open music, the oft-cited vocal plus single instrument. The number of frequencies present at any given time is low and almost any amp will sound good with that. You can even tolerate much higher THD from the electronics when listening to this kind of source material.
For myself, I prefer to have the lowest THD I can get, which generally offers the lowest IM which is more important. Electronic distortions are ALL unnatural to our ears and brain since we did not evolve having to sort them out.
tiefbassuebertr's statement regarding rock music requirements is wrong and seems like he has an aesthetic bias.
The fact is that most pop and rock and even rap is very dense in the midrange frequencies, which is also where our hearing is most sensitive. It follows that the highest fidelity possible is required to make this come across as intelligible music and not just a smear of mud.
Contrary to that requirement is what is needed for very open music, the oft-cited vocal plus single instrument. The number of frequencies present at any given time is low and almost any amp will sound good with that. You can even tolerate much higher THD from the electronics when listening to this kind of source material.
For myself, I prefer to have the lowest THD I can get, which generally offers the lowest IM which is more important. Electronic distortions are ALL unnatural to our ears and brain since we did not evolve having to sort them out.
I'd like to listen to "Linn's Klimax SOLO 500 vs. Pass Labs Aleph 1.2 ". Also the first post mentioned Altec Lansing 9444. Unfortunately, in flyover country, I'd have to buy them. Checking on e-bay, anything with a famous name starts $500 up, even non-functional.
I view amplifier HD & IMD levels as a bit academic, since speakers are so bad. I was able to hear 1% HD of my dynaco ST70 (repaired), versus much better Peavey CS800s (specified .03% @ 240 W/ch). Who knows what HD is at 1/4 W that I listen to it at. And I have a 1950's record player (Howard) that was probably producing 20% HD, but then 78 records weren't hifi anyway.
I view amplifier HD & IMD levels as a bit academic, since speakers are so bad. I was able to hear 1% HD of my dynaco ST70 (repaired), versus much better Peavey CS800s (specified .03% @ 240 W/ch). Who knows what HD is at 1/4 W that I listen to it at. And I have a 1950's record player (Howard) that was probably producing 20% HD, but then 78 records weren't hifi anyway.
Last edited:
This is a very interesting subject, perhaps deserving its own thread. Too easy to fall into the trap of oversimplification.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.