How do you perceive THD at 10KHz or 20KHz, especially from simulation? Do these numbers mean anything to you?
My standard minimum for THD at 1KHz is 0.1% at any stage in any amps (depends on the topology). But I cannot do the same thing with THD at 10KHz for example. I cannot relate the number with actual (real life) performance.
So, somebody must be able to "benefit" from this number. Otherwise, why bother measuring it. You might for example (just a guess) use the number to predict certain kind of anomalies in the HF reproduction...
Or, may be somebody have tried to make this measurement to become more meaningfull by lowering the fundamental frequency, or even by adding certain filters? (Just a guess).
I will also find some answers later by simulating as many amps that I'm familiar with and check if there is correlation between this THD and the amp's performance. But from limited simulation so far, I cannot draw a good correlation. I don't even know what should the minimum acceptable distortion be.
My standard minimum for THD at 1KHz is 0.1% at any stage in any amps (depends on the topology). But I cannot do the same thing with THD at 10KHz for example. I cannot relate the number with actual (real life) performance.
So, somebody must be able to "benefit" from this number. Otherwise, why bother measuring it. You might for example (just a guess) use the number to predict certain kind of anomalies in the HF reproduction...
Or, may be somebody have tried to make this measurement to become more meaningfull by lowering the fundamental frequency, or even by adding certain filters? (Just a guess).
I will also find some answers later by simulating as many amps that I'm familiar with and check if there is correlation between this THD and the amp's performance. But from limited simulation so far, I cannot draw a good correlation. I don't even know what should the minimum acceptable distortion be.
Can you put numbers for quantitative measure? For example, 20% is okay, 100% is unacceptable.
I know that this depends on the bandwidth of the amp, but these numbers must have a good use...
The THD itself might not be the weakest link either...
I am very skeptical. I have solid state and tube line level preamps with almost same distortion profile, but they sound VERY different. THD is IMO very poor measure for subjective evaluation of sound quality.
Anyway, try to avoid ANY design with distortion rising with frequency. In this case THD is an INDICATOR that something else is just going wrong.
Anyway, try to avoid ANY design with distortion rising with frequency. In this case THD is an INDICATOR that something else is just going wrong.
THD is just an easy way to measure non linearity. One number doesn't give much information on the shape of the amplifier gain versus input level and frequency. So in my opinion it isn't possible to say from a THD number what is acceptable and what isn't. Though I think that the various distortions that Loudspeakers have completely swamp any low level distortion from most amplifiers though.
Non linearity at 10 kHz and 20 kHz matters because it will cause intermodulation products with music signals that have multiple frequencies.
Non linearity at 10 kHz and 20 kHz matters because it will cause intermodulation products with music signals that have multiple frequencies.
I'm skeptical too right now. But I believe that it is possible to interpret numbers and make it more meaningful with certain methodology.
The THD alone might not be very useful, but coupled with other parameters, why not? I'm most interested in the phase parameter actually. But still understand a little about it.
I think any amp has this behaviour of increasing THD with frequency. The problem is how much is too much. I guess VFET will have terrible THD at above 10kH. And so are simple circuits.
Usually design is about trade-offs. If we know the "minimum acceptable" of each parameter, then it will be easier. For example, once the thd is already with 2 trailing zeros (even less), instead of lowering it further I will try to get good harmonic spectrum and phase behaviour. But noise floor is very important for good bass. My Mimesis clone has -240dB.
Something like what for example? Please note that TSSA has very high THD, at least in simulation.
Except those distortions related with fatigue. IMD, TIM or whatever the name...
But remember that there is no music above 20kHz.
In practice , with NFB amps , this will amount to increase
distorsion at low frequencies to match the high frequency
distorsion level...
Why not consider different possibilities of reduction of distortion, like distortion cancellation. I have not stated that high loop NFB would be my preferred solution
I think that the question is somewhat meaningless. HF distortion is a symptom, not the disease, so you can't categorize the "sound" of it, independent of knowing the cause. THD in particular is a very incomplete number- it lumps together very different pattens of nonlinearity with different causes, different effects, and different cures.
The OP mentions amp simulations. As one who is trying to learn about computer simulations and programs, when modeling an amplifier, and obtaining information such as distortions present, what load is being used? Is it a resistive or complex load, i.e., do you use the complex crossover components and loudspeaker parameters, say, of an LS3/5A (or other known quantity) in the simulations?
Thanks, trying to learn.
Terry
Thanks, trying to learn.
Terry
Harmonic distortion above audible freq
I've always wondered how harmonics of say,a 10khz fundamental, manifests itself in audibility. 2nd harmonic is arguably on edge (or out of) hearing range, and higher harmonics are even further. Without further technical understanding on my part, I would think that modulation interactions and effects in the highest frequencies could be a problem. At most it would sound a bit edgy, unless really high distortion. At any rate, I'm not positive I could hear anything above 16-17khz anyways...
I've always wondered how harmonics of say,a 10khz fundamental, manifests itself in audibility. 2nd harmonic is arguably on edge (or out of) hearing range, and higher harmonics are even further. Without further technical understanding on my part, I would think that modulation interactions and effects in the highest frequencies could be a problem. At most it would sound a bit edgy, unless really high distortion. At any rate, I'm not positive I could hear anything above 16-17khz anyways...
The same nonlinearity that distorts a single sine wave at HF, producing thd, would produce intermods when more than one HF tone is present ... in particular a second order nonlinearity would produce the difference frequency, which is much lower and audible; third order harmonic disto implies third order intermods which are HF but similar in frequency to the two tones, etc, even if all the harmonics are inaudible.
In case there is no slew rate issue, CCIF 19+20kHz is pretty same like THD 1kHz (and NOT like THD 20kHz). I have plenty of measurements that prove this.
Funny you should say that. I was thinking the same recently. I guess we'll have to accept that any truly high-performance amplifier uses nested feedback of one flavor or another.When i read "distorsion cancelation" , i translate with "NFB in disguise"....
Anyway, 10 kHz distortion is just a handy way of getting a handle on high-frequency intermod in simulation. As most designs show distortion increasing with frequency, the point is that if it's OK there, it should be anywhere. Always look at the spectrum rather than just the THD number though.
Incidentally, just about any case where someone thought to have proven the audibility of ultrasonic tones eventually turned out to be caused by plain ol' amplifier HFIM...
When i read "distorsion cancelation" , i translate with "NFB in disguise"....
Then you have too few tools in your toolbox.
When I read “distortion cancellation” I translate it to “making the amp as linear as possible” before NFB is applied.
Cheers
Stein
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.