multitone spectrum proposal

[Lambda]

I can create one and the same dense multitone with regard to frequencies and magnitude spectrum with varying crest factors from close to infinity (basically almost a Dirac pulse) to almost unity (endless sine sweep, overlapping the range ends). Of course they give totally different result even when normalized to the same RMS level.

Can you pleases do this and upload it here to lustrate it

 

[NTK]

Why?
To me this sounds misleading.

As chaining the scale also changes the "noise floor"

The "grass" level, in this case, is not the noise floor. The grass are the intermodulation products of the multitones. Therefore, what the multitone measurement is about is the heights of the intermodulation products relative to the heights of the tones, which will not change by changing the reference point of the dB scale.

 

[Hayabusa]

The "grass" level, in this case, is not the noise floor. The grass are the intermodulation products of the multitones. Therefore, what the multitone measurement is about is the heights of the intermodulation products relative to the heights of the tones, which will not change by changing the reference point of the dB scale.

The actual combined level of the multi tone should be the reference level in this case I would say..

 

[NTK]

The actual combined level of the multi tone should be the reference level in this case I would say..

What do you mean by the "combined" level? Absolute peak? Or RMS?

[Edit] I agree with KSTR that the crest factor is important, and both the RMS and crest factor should be stated in the measurement. And since this is a measurement of nonlinear behavior, it is valid and applicable only at the particular measurement signal level.

 

[Hayabusa]

What do you mean by the "combined" level? Absolute peak? Or RMS?

RMS would be most fair I would say.

 

[Lambda]

The actual combined level of the multi tone should be the reference level in this case I would say..

eater way way it should be clearly labeled.

What do you mean by the "combined" level? Absolute peak? Or RMS?

I would say Absolute peak. but this way a pure sine wave would also not go to 0dB
As KSTR says (if i get him right) the difference between RMS and Peak can be significant depending on how the signal is generated.

The "grass" level, in this case, is not the noise floor. The grass are the intermodulation products of the multitones.

You ar right.
As long as the inter modulation products are higher then the noises random noise what they usually are.

 

[amirm]

The actual combined level of the multi tone should be the reference level in this case I would say..

It already is:

See the label?

 

[Lambda]

See the label?

I don’t get it...
What is the combined level? +20dB?


because the combined level can't be the same as a singe tone level?

 

[amirm]

I don’t get it...
What is the combined level? +20dB?

It says right on the graph: 5 watts@4 ohm. I adjust the level being fed to the amplifier until the power meter in AP shows 5 watts. This way, all amplifiers are tested at the same level, obviating the need to indicate this in the vertical axis. By setting that axis to 0 dB reference, it then becomes much easier to read the relative level of intermodulation products.

People who use random vertical axis with dBu/dBv etc. cause no end of confusion as people just look at the floor of the graph, and think that is the negative distortion level.

 

[amirm]

because the combined level can't be the same as a singe tone level?

I rely on RMS detector in AP to provide the equiv. sine wave level. You can see the effect of lower amplitude of individual tones in DAC measurements where the distortion is lower for each spike due to that factor.

 

[Lambda]

0 dB reference, it then becomes much easier to read the relative level of intermodulation products.

So Your reference test level is 5W (with unknown crest level?)
I think it would make more sense to show the distortion level relative to the 5W reference level?

If 5W is reference for this test why not make 5W 0dB?

Also, it might be a small difference, but setting the RMS output to 5W
Since the 5W includes signal+distortion+noise
and we don't have the numerical values for distortion, intermodulation products, and noise
we don't know the absolute level that you're labeling as 0 dB.

 

[amirm]

If 5W is reference for this test why not make 5W 0dB?

That is what it is for heaven's sake:

You see the peaks at 0 dBrA?

Also, it might be a small difference, but setting the RMS output to 5W

There is no such thing as RMS watts. There is RMS voltage, but not RMS watts.

 

[amirm]

and we don't have the numerical values for distortion, intermodulation products, and noise

That is what the other tests are for. This test is meant to show a composite view of how distortion varies with frequency. Indeed, the origin of multitone signal was a quick, manufacturing test to show multiple things at once. If frequency response has error for example, the peaks of each spike varies.

 

[Lambda]

There is no such thing as RMS watts. There is RMS voltage, but not RMS watts.

You just said:

I rely on RMS detector in AP

So you uses the RMS " voltage" level to calculate Watts.
If you want to be be Pedantic what You label Watt are not Watts at all it might be VA. but you don't measure current?

You see the peaks at 0 dBrA?

So 0dBrA is surely not 5W or 5VA or the RMS voltage correlating to theoretical 5W.
If this is So clear what please tell me what is the actually Absolute level of 0dBrA?
Maybe in use "random" vertical axis with dBu/dBv

With an not random but absolute scale like dBu/dBv it would be trivial.

 

[amirm]

So you uses the RMS " voltage" level to calculate Watts.
If you want to be be Pedantic what You label Watt are not Watts at all it might be VA. but you don't measure current?

With resistive loads, we don't worry about power factor as it is 1. So V*V/R works.

 

[amirm]

If this is So clear what please tell me what is the actually Absolute level of 0dBrA?

Level of what? Voltage? If so, it is the RMS voltage corresponding to 5 watts @ 4 ohm.

It really doesn't matter much anyway as all comparisons are made the same way, with the same output voltage and load.

 

[NTK]

You just said:

So you uses the RMS " voltage" level to calculate Watts.
If you want to be be Pedantic what You label Watt are not Watts at all it might be VA. but you don't measure current?

So 0dBrA is surely not 5W or 5VA or the RMS voltage correlating to theoretical 5W.
If this is So clear what please tell me what is the actually Absolute level of 0dBrA?
Maybe in use "random" vertical axis with dBu/dBv

The voltage levels of the multitones can be easily back calculated.

The power is 5 W into 4 Ω, therefore the multitone V_rms = √( 5 × 4 ) = 4.472 V.

Since all the individual tones have equal voltage amplitudes, therefore V_rms = √( 32 × V_tone² ) ⇒ V_tone = 0.79 V.

So 0 dBrA is 0.79 V (rms).

 

[DonH56]

So Your reference test level is 5W (with unknown crest level?)
I think it would make more sense to show the distortion level relative to the 5W reference level?

If 5W is reference for this test why not make 5W 0dB?

Also, it might be a small difference, but setting the RMS output to 5W
Since the 5W includes signal+distortion+noise
and we don't have the numerical values for distortion, intermodulation products, and noise
we don't know the absolute level that you're labeling as 0 dB.

The AP, like most if not all audio analyzers, measures voltage and calculates RMS values accordingly. Conversion to watts comes from knowing the load resistance, then P = V^2 R which provides average (not "RMS") watts. With a purely resistive load it is watts, not VA, since a resistive load has unit power factor. Since it is generating an RMS value, crest factor is irrelevant, as it is buried in the RMS calculation. The crest factor depends upon the signal, and in this case number of sine-wave tones, since the crest factor of a multiplicity of tones is much higher than that of a single tone. The RMS value goes as the square root of the product of the tones squared, Vrms = sqrt(n*V^2) for n equal tones of value V.

Unless noise and/or distortion are very high, they won't affect this measurement to any appreciable degree, and in any case since RMS voltage is calculated they are included in the result so using a 0 dB reference is valid.

Don't get buried in the weeds looking at the forest...

 

[olieb]

So 0 dBrA is 0.79 V (rms).

Yep. If all peaks of all tones happen at the exact same time V_max (peak) would be 35.8V.
But the tones might be chosen so that this does not happen. It would be interesting to know the crest factor.

 

[Lambda]

The power is 5 W into 4 Ω, therefore the multitone V_rms = √( 5 × 4) = 4.472 V.

Since all the individual tones have equal voltage amplitudes, therefore V_rms = √( 32 × V_tone² ) ⇒ V_tone = 0.79 V.

So 0 dBrA is 0.79 V

We are ignoring the noise and distortion also contributing to the 5W but ok i don't want to be pedantic.

So lets say its So 0 dBrA is 0.79 V RMS
Tahts far From Waht Amir is saying:

Level of what? Voltage? If so, it is the RMS voltage corresponding to 5 watts @ 4 ohm.

The voltage corresponding to 5 watts @ 4 ohm would be ~4.47V
But 32 Tones at 4.47V would add up to ~25V...