Why is SINAD measured @ 5W?

[restorer-john]

I do wonder if Anthony Grimani’s CEDIA specification project will get adopted where amps now advertise dB gain and not just (or instead of?) wattage.

dB gain is just a voltage gain number. It tells nothing of power delivered. There was a push a decade or more a go to adpot dBW as a spec. But it never took off, due to domestic amplifiers mostly being bunched together with similar numbers. Consumers don't care for numbers that make no sense to them- they want the bigger is better 'watts' they think they understand.

My attitude is well documented here on ASR in relation to this, but for the sake of consistency and perhaps an acknowledgment that 5W@4R is just as arbitrary as 1W@8R was back in the day, I have no problem wih it.

That said, there needs to be more focus on the FTC compliance being from 250mW to rated power. The Rotel example specs above highlights this. They are generally a very honest manufacturer and their amplifiers are tradionally a bit 'noisier' (residual noise), but always honest low THD. I tested one of my Rotels the other day, and the residual noise meant the THD+N at 1W was circa 0.04% although at high powers, the THD+N was 0.002%. Their rated FTC spec was less than 0.05% from 250mW to rated power (100W) so they were truthful.

 

[MZKM]

Consumers don't care for numbers that make no sense to them- they want the bigger is better 'watts' they think they understand.

Yep, I don’t think it’ll take off, companies would rather have consumers comparing 100W vs 200W amplifier rather than +20dB vs +23dB (and yes, that’s just calculated via voltage gain; all speaker have different impedances and phase, so be near impossible to advertise real-world dab gain, even more so if talking a range of frequencies, as some amps can’t output as much lower into 40Hz like they can 1kHz).

I mostly just care for specs to have their parameters published, even if in the footnotes.

 

[oupee]

Just take it as a reflection. If I test or review a car with 300 horsepower, I will limit the power to only 5 horsepower. What will that tell me about that car? What a difference it will be if I use 150 horses, where the quality of the tires, the road, the weather, the quality of the suspension will start to show .... Testing a car with 5 horsepower will only make a moron. Just for reflection.

 

[EdTice]

Just take it as a reflection. If I test or review a car with 300 horsepower, I will limit the power to only 5 horsepower. What will that tell me about that car? What a difference it will be if I use 150 horses, where the quality of the tires, the road, the weather, the quality of the suspension will start to show .... Testing a car with 5 horsepower will only make a moron. Just for reflection.

If a car has 300 horsepower, should the estimated highway gas mileage be calculated at the car's top speed or at 65mph? Which one tells you more about the anticipated fuel efficiency

 

[ToriCatcher]

That would be the wrong assessment. Directionally, SINAD at 5 watts is a great predictor of the amplifier performance in general. The amp would have to have both low noise and distortion to get high SINAD. Such attention to both of these factors means a very optimized design so rest of the measurements show the same. I can't think of a situation where SINAD was great at 5 watts but rest of the measurements were poor. And vice versa.

Would that also be the case when considering SINAD at different frequencies? Meaning, what would the correlation of SINAD at 5 volts at 1kh, with SINAD at 5 volts at different (in particular very high and very low) frequencies be?
Thanks a lot.
Michael

Edit: this raises by claims of some audiophiles I heard, saying class A amps will have a more consistent performance throughout bandwidth than class D.

 

[ToriCatcher]

Would that also be the case when considering SINAD at different frequencies? Meaning, what would the correlation of SINAD at 5 volts at 1kh, with SINAD at 5 volts at different (in particular very high and very low) frequencies be?
Thanks a lot.
Michael

Edit: this raises by claims of some audiophiles I heard, saying class A amps will have a more consistent performance throughout bandwidth than class D.

If anyone can shed some light on this, it will be greatly appreciated.
Thanks

 

[rwortman]

And, just for fun, why not measure distortion and FR at 5 watts into an actual speaker?

 

[amirm]

And, just for fun, why not measure distortion and FR at 5 watts into an actual speaker?

It is annoying to make such measurements and at any rate, what speaker would be representative of all speakers out there? And how would others repeat the test if they don't have or own that speaker?

 

[amirm]

Would that also be the case when considering SINAD at different frequencies? Meaning, what would the correlation of SINAD at 5 volts at 1kh, with SINAD at 5 volts at different (in particular very high and very low) frequencies be?
Thanks a lot.
Michael

Edit: this raises by claims of some audiophiles I heard, saying class A amps will have a more consistent performance throughout bandwidth than class D.

As a general statement, yes. I run such a test by the way: it is the power vs frequency vs distortion+noise. You can look at a sample of them and see if the conclusions are different.

On Class A, that would only matter if it was as optimized of a design. Many are not so even though they may have more frequency independence, their distortion and noise is higher across the board than state of the art Class D amps.

 

[rwortman]

No speaker would represent them all but it would be interesting to see if the performance degraded less with some amps than others. After all they are not designed for heating resistors. So many people in here think all amps must sound the same based on performance into resistors. I wonder how and in what ways their performance degrades into an active load.

 

[NTK]

No speaker would represent them all but it would be interesting to see if the performance degraded less with some amps than others. After all they are not designed for heating resistors. So many people in here think all amps must sound the same based on performance into resistors. I wonder how and in what ways their performance degrades into an active load.

The issue with using a reactive load is that the reactive load "stores and releases" energy back to the source. Here is a simulation of a square pulse sent to a simulated speaker driver. When the source impedance (R_source) is non-zero, the energy from the reactive load in the form of electrical current will result in a voltage across the source impedance, and this voltage will get picked up by the analyzer. (The lower step level at the sensor is due to the voltage drop across the non-zero source impedance.)



The AP analyzer has no way of telling that this voltage is not distortion from the source, and will count it as distortion (or frequency response error). Therefore reactive loads are not suitable for amplifier distortion measurements.

There is a thread on the effects of "non-ideal" dummy load resistors with measurements.

ASR dummy load configuration

From the left hand column of page 20 of the Hofer article you posted the formula for the Resistor Voltage Coefficient. You neglected to mention or quote, “In most common electronics, the VCr is safely assumed to be negligible”. Still keep your eye on it for ultra-low THD + N applications. The...

www.audiosciencereview.com

 

[rwortman]

The issue with using a reactive load is that the reactive load "stores and releases" energy back to the source. Here is a simulation of a square pulse sent to a simulated speaker driver. When the source impedance (R_source) is non-zero, the energy from the reactive load in the form of electrical current will result in a voltage across the source impedance, and this voltage will get picked up by the analyzer. (The lower step level at the sensor is due to the voltage drop across the non-zero source impedance.)

View attachment 186707 View attachment 186708

The AP analyzer has no way of telling that this voltage is not distortion from the source, and will count it as distortion (or frequency response error). Therefore reactive loads are not suitable for amplifier distortion measurements.

There is a thread on the effects of "non-ideal" dummy load resistors with measurements.

ASR dummy load configuration

From the left hand column of page 20 of the Hofer article you posted the formula for the Resistor Voltage Coefficient. You neglected to mention or quote, “In most common electronics, the VCr is safely assumed to be negligible”. Still keep your eye on it for ultra-low THD + N applications. The...

www.audiosciencereview.com

I am aware of the nature of a reactive load. The job of an amplifier is to control a loudspeaker, not to warm a resister with a nice low distortion signal. It seems to me that investigating the behavior of an amplifier while it is driving a not only reactive but active load might yield some interesting scientific results. Might even shed some light in the hypothesis that amplifiers that measure similarly sound different when connected to the things they are supposed to drive. I don’t expect them to meet their spec’s. A comparison between amplifiers might be interesting.

 

[NTK]

I am aware of the nature of a reactive load. The job of an amplifier is to control a loudspeaker, not to warm a resister with a nice low distortion signal. It seems to me that investigating the behavior of an amplifier while it is driving a not only reactive but active load might yield some interesting scientific results. Might even shed some light in the hypothesis that amplifiers that measure similarly sound different when connected to the things they are supposed to drive. I don’t expect them to meet their spec’s. A comparison between amplifiers might be interesting.

What the AP analyzer will measure is the voltage drop across the amplifier output impedance due to the reactive current coming back from the speaker. The measured "error" voltage is therefore proportional to and dominated by the magnitude of the amplifier output impedance. An amplifier with half the output impedance (double damping factor) will show 1/2 the "distortions" as compared to an equivalent amplifier except when it has twice the output impedance. This measurement is not telling us anything useful.

 

[rwortman]

The speaker load is not simply reactive. The voice coils are generators too so there is an active component. The output impedance of the amp isn’t purely resistive either. You may believe that an amp’s behavior driving a loudspeaker can be completely predicted by its performance into a resistor and it’s output impedance but I don’t.

 

[amirm]

The speaker load is not simply reactive. The voice coils are generators too so there is an active component. The output impedance of the amp isn’t purely resistive either. You may believe that an amp’s behavior driving a loudspeaker can be completely predicted by its performance into a resistor and it’s output impedance but I don’t.

NTK is making a different and correct statement. That the analyzer will get fooled by the back EMF energy from the actual load and present data that is dependent on the feedback level, etc. of the amp, not inherent non-linearities. I did this testing with headphones and the outcome at the end was that the results were not of any value. Of note, different headphones produced very different results due to this factor with the same amp so you can't even integrate the results to any other loads.

 

[rwortman]

The analyzer is not being fooled, it is measuring what is happening. I don’t have the finances to go out and purchase an analyzer but if I had one and people were sending me amplifiers to measure, my curiosity would have me comparing how different amps drive a chosen test speaker. Your house, your choice.

 

[bigguyca]

I like your infographic but I think this thread has gotten a bit too dramatic.

Amir plots the distortion curve for the amplifier from 0.25W up until distortion reaches some threshold ( I believe 1% ). So you can see the whole graph if you want.

Keep in mind that music typically has peaks in the 6dB to 25dB range depending on whether or not you consider EDM to be "music" (Sorry couldn't help myself). I'm not great at math so lets call it 20dB to make it easy. 20dB is factor of 100. So if an amp is rated at 100w into 8ohms, in order to have enough headroom for the peaks, your average output level will be around 1w. More specifically if you were watching a movie, speech would be at about 1w and the blowing stuff up would be at 100w. I like the speech to be intelligible. If sound of a tank firing mortar at a motorcycle at point blank range (and missing) distorts a bit, I probably wouldn't know since I thankfully haven't experienced that in real life.

I am not going to enter the 1w vs 5w debate. But the idea of picking a number that is greater than 1 watt and less than 1/4 of the amplifiers rated power seems pretty reasonable.

It seems likely that essentially no one has seen or heard a normal tank firing a mortar at a motorcycle.

 

[EdTice]

It seems likely that essentially no one has seen or heard a normal tank firing a mortar at a motorcycle.

Yeah. It's like that's the stuff of Hollywood movies or something. Of course, if you did have such an actual experience, it would probably lead to hearing damage to the point where you couldn't really listen in high fidelity so you still wouldn't notice if its distorted.

 

[TheBatsEar]

Found those labels incredibly useful. Could pick a freezer with good isolation and low energy usage in it's class. I don't see a problem.

 

[Gorgonzola]

Would that also be the case when considering SINAD at different frequencies? Meaning, what would the correlation of SINAD at 5 volts at 1kh, with SINAD at 5 volts at different (in particular very high and very low) frequencies be?

As a general statement, yes. I run such a test by the way: it is the power vs frequency vs distortion+noise. You can look at a sample of them and see if the conclusions are different.

As I recall I have seen and even commented that I though these test might well serve to differentiate amplifiers with very similar noise and HD spectra at (say) 5 watts into 4 ohms.

But is it fair to assume that the spectra at different power levels or at different frequencies would be similar in form? Or is it likely that (say) orders of HD would appear in different proportion at different frequencies or power levels?