Wrong statement. The input signal has no bearing with the "condition" (whatever that means in your foggy language) of the device.
Hi Mark,
Agreed, tone bursts and other tactics can be used.
I am trying to address the fallacy that music is somehow more suitable for testing than a sine wave. We have gated testing that can be used of course as well. There are basic misconceptions that continue to be trotted out that really ought to be laid to rest.
A person who doesn't understand test equipment and procedures is likely to believe anything, and that blocks the progress of otherwise good threads. We get the same arguments from the same people over and over. Education is the answer to this issue. Wouldn't you agree?
-Chris
Agreed, tone bursts and other tactics can be used.
I am trying to address the fallacy that music is somehow more suitable for testing than a sine wave. We have gated testing that can be used of course as well. There are basic misconceptions that continue to be trotted out that really ought to be laid to rest.
A person who doesn't understand test equipment and procedures is likely to believe anything, and that blocks the progress of otherwise good threads. We get the same arguments from the same people over and over. Education is the answer to this issue. Wouldn't you agree?
-Chris
I disagree:
It has constant average power and spends most of its time at the amplitude extremes, which is the opposite of a music signal which has wildly varying power and a statistical distribution that looks like a gaussian centered on zero (if the song was not destroyed by the loudness wars)...
It is zero centered and symmetrical, whereas a music signal is not zero-centered nor symmetrical over a few milliseconds (of course over a longer period it does average to zero).
It has maximum slew rate at zero, which is not necessarily the case of a music signal either. To test for oscillations or ringing, a large sine (or a triangle) sweeping the whole output voltage range plus a small square wave to excite the ringing/oscillations gives more information.
That said, two sines, high amplitude low frequency and low amplitude high frequency, that works. Or one sine that alternates between low power and high power, that works too. Maybe some DC on top, or some amplitude variation. It's simple, the software to analyze it is simple too.
Using music would be very impractical. It's very difficult to separate the signal from the distortion, and null tests depend on getting the phase perfectly right at all frequencies so it cancels, which is impossible. High effort, little payback.
Last edited:
Good luck with that. I won't insult your intelligence by explaining why that won't happen here.
If those are your criteria for a good test signal, then noise fits almost perfectly your romantic description. Actually, testing with noise has a well defined role in systems identification and characterization.
It will of course not tell anything about distortions (same as testing "with music"). Your suggestion for testing with a LF+HF signal is similar to the known, well defined, SMPTE/DIN method to measure the IMD. As I said before, there isn't anything that your method may add to the picture.
P.S. forgot to add that a sine modulated by a square signal is also a well known, standardized, test signal for characterizing DIM (Dynamic InterModulation).
Last edited:
Hi peufeu,
I think we would agree if we frame the question with boundaries.
Music is a collection of sine waves, and it does have a maximum rate of change at zero crossings most often, unless it is a rare inflection point - yes?
More specialized tests always reveal some characteristics of an amplifier, but a frequency response test of an amplifier will also reveal ringing. It has to and does. When designing I do use square waves, and if I were to do an in depth analysis of a circuit or amplifier, I would use these additional techniques. But all this takes time, and someone has to pay for it at some point.
The bulk of what you need to know about an amplifier can be found with one or two sine wave signals. That includes sticky clipping behavior. Most of what someone will hear from an amplifier occurs in the 1 watt range (peaks much higher in music, can be 15 X the average level). So, assuming a listening test occurs in the linear power range of an amplifier (a valid assumption), the 1 watt tests are in fact valid and will describe most of what will be heard. At that level, thermal effects don't occur, nor does clipping with any competently designed amplifier. If the power rating of the unit under test is limited to below 15 watts rms, I would suggest the level of the listening test and quality tests be reduced to match.
I am assuming a well designed test, listening or otherwise. If the test isn't well controlled or design, then all bets are off anyway. The test is a waste of everyone's time.
Again, as long as you are operating equipment within its normal linear range, most of what you need to know will be revealed with a simple THD test at 1 watt, and an IMD test at the same levels. Want to completely characterize something? Okay, we have other techniques and signals we can use. At no time do we have to rely on a music signal.
-Chris
I think we would agree if we frame the question with boundaries.
Music is a collection of sine waves, and it does have a maximum rate of change at zero crossings most often, unless it is a rare inflection point - yes?
More specialized tests always reveal some characteristics of an amplifier, but a frequency response test of an amplifier will also reveal ringing. It has to and does. When designing I do use square waves, and if I were to do an in depth analysis of a circuit or amplifier, I would use these additional techniques. But all this takes time, and someone has to pay for it at some point.
The bulk of what you need to know about an amplifier can be found with one or two sine wave signals. That includes sticky clipping behavior. Most of what someone will hear from an amplifier occurs in the 1 watt range (peaks much higher in music, can be 15 X the average level). So, assuming a listening test occurs in the linear power range of an amplifier (a valid assumption), the 1 watt tests are in fact valid and will describe most of what will be heard. At that level, thermal effects don't occur, nor does clipping with any competently designed amplifier. If the power rating of the unit under test is limited to below 15 watts rms, I would suggest the level of the listening test and quality tests be reduced to match.
I am assuming a well designed test, listening or otherwise. If the test isn't well controlled or design, then all bets are off anyway. The test is a waste of everyone's time.
And this is my basic point. Music won't tell us anything we can't find out with test signals, sine, square or otherwise. Linear region type testing will only require a sine type signal, a pair to generate IM products. That's all you need and that is all I do on a quick test I give my customers. Once you begin to charge time to completely characterize a circuit (amplifier or other), people very quickly become disinterested in demanding more answers.
Again, as long as you are operating equipment within its normal linear range, most of what you need to know will be revealed with a simple THD test at 1 watt, and an IMD test at the same levels. Want to completely characterize something? Okay, we have other techniques and signals we can use. At no time do we have to rely on a music signal.
-Chris
Certainly an understanding of mathematics is, ahem, integral to good measurements.
Too late to edit; It's actually the other way around, a square LF sgnal (3.15KHz) modulated by a sine HF (15KHz) signal that is also a well known, standardized, test signal for characterizing DIM (Dynamic InterModulation). It's in IEC 60268 as DIM30 and DIM100.
- Status
- Not open for further replies.