I am just making first steps in THD measurements with QA401. The bandwidth is apparently limited to 20kHz in contrast to the AP devices. I am wondering if that is sufficient.
Doug Self states in his book that 80kHz is advisable for you otherwise would miss all the harmonics >20kHz which cause IMD <20kHz. That harmonics >20kHz lead to distortion below 20kHz is obvious. However, only audible distortion <20kHz matters but this would completely be detected by my QA401, isn't it? So whatever distortion originates from harmonics >20kHz would bet detected even with a bandwitdh limited to 20kHz?
Can somebody enlighten me in this respect?
Doug Self states in his book that 80kHz is advisable for you otherwise would miss all the harmonics >20kHz which cause IMD <20kHz. That harmonics >20kHz lead to distortion below 20kHz is obvious. However, only audible distortion <20kHz matters but this would completely be detected by my QA401, isn't it? So whatever distortion originates from harmonics >20kHz would bet detected even with a bandwitdh limited to 20kHz?
Can somebody enlighten me in this respect?
I don't really understand your comment... However, after further thinking I am even more confused than in the beginning:
The statement by Douglas Self that IMD will be masked by a limited measurement bandwidth of 30kHz (instead of the proposed 80kHz) is puzzling to me. I my understanding these THD plots in his book were made by a test signal of a single frequency in LF and increasing that to HF to wherever the bandwidth is limited. But with that you would never reveal IMD because there is a single frequency involved. You would need at least two frequencies like 19kHz and 20kHz to see intermodulation effects...
The statement by Douglas Self that IMD will be masked by a limited measurement bandwidth of 30kHz (instead of the proposed 80kHz) is puzzling to me. I my understanding these THD plots in his book were made by a test signal of a single frequency in LF and increasing that to HF to wherever the bandwidth is limited. But with that you would never reveal IMD because there is a single frequency involved. You would need at least two frequencies like 19kHz and 20kHz to see intermodulation effects...
Hi Jan, thanks!
I was trying around with bandwidth before I switched to 192kbps.
But I am still wondering about the bandwidth statement in his book. The difference in HF THD between 30kHz and 80kHz is clear - harmonics above 30kHz are masked in the former case. But maybe harmonics above 30kHz can be regarded as a hint for the IMD in the audible range without actually measuring them.
I was trying around with bandwidth before I switched to 192kbps.
But I am still wondering about the bandwidth statement in his book. The difference in HF THD between 30kHz and 80kHz is clear - harmonics above 30kHz are masked in the former case. But maybe harmonics above 30kHz can be regarded as a hint for the IMD in the audible range without actually measuring them.
THD and IMD come from the same issue: nonlinearity. Hence measuring one gives a guide to the likely level of the other. To measure THD you must be able to see the harmonics, so wide bandwidth is needed. Some IMD products are at lower frequencies than the input so you can measure them with a narrower bandwidth - you just have to assume that the higher frequency products vary similarly.
I think my misunderstanding was that you really measure IMD but as DF96 said, you rather get an idea of IMD caused by THD >20kHz. It makes sense to me.
Another thing is worrying me: In my understanding Doug's THD plots with bandwidth of 80kHz are measured by applying a sine signal starting in LF and incrementally highered to 20kHz. Is that right?
If so, is it a good idea of stopping at 20kHz? Because fundementals beyond 20kHz will cause significant harmonics (since linerarity decreases with frequency) which in turn may cause significant IMD in the audible range. Hence, actually existing and significant distortion is masked by stopping at 20kHz. Or can we assume that usual program material, especially from digital sources, does not include signals above 20kHz or only of a negligible degree?
Another thing is worrying me: In my understanding Doug's THD plots with bandwidth of 80kHz are measured by applying a sine signal starting in LF and incrementally highered to 20kHz. Is that right?
If so, is it a good idea of stopping at 20kHz? Because fundementals beyond 20kHz will cause significant harmonics (since linerarity decreases with frequency) which in turn may cause significant IMD in the audible range. Hence, actually existing and significant distortion is masked by stopping at 20kHz. Or can we assume that usual program material, especially from digital sources, does not include signals above 20kHz or only of a negligible degree?
I don't know which plots by Douglas you refer to, but bandwidth is just that, it is the bandwidth the analyzer looks at. What the actual test frequency or sweep freq is should be apparent from the plot.
I think you worry too much. A 19 + 20kHz test does not reflect actual music where you will not have full power 19 + 20kHz; it is just an artificial measurement condition.
Also, although in theory HD products can in turn generate IM products, the HD products will be very low with respect to the signal and the resulting IMD products even lower.
Jan
I think you worry too much. A 19 + 20kHz test does not reflect actual music where you will not have full power 19 + 20kHz; it is just an artificial measurement condition.
Also, although in theory HD products can in turn generate IM products, the HD products will be very low with respect to the signal and the resulting IMD products even lower.
Jan
It seems that there would be no need to worry about distortion products at 40 or 60kHz from a 20kHz signal, but you have to step back a bit and think about what these measurements are telling you.
The same basic circuit nonlinearities that we're trying to detect (using measurements) can cause harmonic distortion or IM distortion, but as it happens, it is far easier to measure low levels of harmonic distortion than it is to measure IM distortion. So, it's a better idea to equip yourself to do wide band measurements of harmonics, and not IM, and not think that anything above 20kHz is inaudible - that's not the point here.
The whole idea of steady state sine wave testing has little to do with music or audibility, but it has a lot to do with identifying how a circuit misbehaves, so it's best to separate what one would think is 'audible' from what can be measured, in order to gain an insight on how a circuit works or how it fails. You can do listening tests to determine subjective preference, but when optimizing a circuit for minimum distortion, there is no substitute for an instrument that will reliably tell you whether the alteration you just made increased or decrease a certain kind of distortion. In my experience, being able to measure individual harmonics (2nd through 4th) with a low residual distortion floor is most useful.
The same basic circuit nonlinearities that we're trying to detect (using measurements) can cause harmonic distortion or IM distortion, but as it happens, it is far easier to measure low levels of harmonic distortion than it is to measure IM distortion. So, it's a better idea to equip yourself to do wide band measurements of harmonics, and not IM, and not think that anything above 20kHz is inaudible - that's not the point here.
The whole idea of steady state sine wave testing has little to do with music or audibility, but it has a lot to do with identifying how a circuit misbehaves, so it's best to separate what one would think is 'audible' from what can be measured, in order to gain an insight on how a circuit works or how it fails. You can do listening tests to determine subjective preference, but when optimizing a circuit for minimum distortion, there is no substitute for an instrument that will reliably tell you whether the alteration you just made increased or decrease a certain kind of distortion. In my experience, being able to measure individual harmonics (2nd through 4th) with a low residual distortion floor is most useful.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.