Same speakers, different amps, different distortion through REW measurements

[heboil]

Hi everyone -

I recently picked up a Umik-1 mic and have been measuring my room, speakers etc. After level matching and swapping amps (speakers and mic in the same position), I have noticed very similar frequency response across the amps (slightly different), with slight deviations in distortion (THD poking it's head above noise floor and 40 dB with one of the amps). These screencaps are both L&R averaged. Can I infer from this deviation, that one amp is distorting the signal more than the other one at these frequencies? I ran the tests multiple times and got roughly the same results each time.

VS

Thanks.

 

[radix]

Interesting. You're talking about those peaks at 550 and 950 Hz? I wonder if there's an issue with one amp and the crossover or impedance curve of the speakers. Have you looked at your speaker's spinorama to see if there's any correlation? I'm just guessing.

I assume there was no one talking in the background or some other possible noise? If it shows up in both the L and R taken at different times, unlikely to be that sort of interference. You don't hear any hums or buzz on that amp when nothing is playing?

 

[heboil]

Interesting. You're talking about those peaks at 550 and 950 Hz? I wonder if there's an issue with one amp and the crossover or impedance curve of the speakers. Have you looked at your speaker's spinorama to see if there's any correlation? I'm just guessing.

I assume there was no one talking in the background or some other possible noise? If it shows up in both the L and R taken at different times, unlikely to be that sort of interference. You don't hear any hums or buzz on that amp when nothing is playing?

Yeah... that is what I am wondering about. Now, I get mic placement and background noise play a role in what the mic measures, but across the board (even when I move the mic), both sets of measurements are a little different on the distortion spectrum. BTW, no one was talking during the test. HVAC was turned off too. This just happened to be a more obvious delta than some of my measurements in the past.

I have some historical measurements over the past month and the distortion delta wasn't as large, but I didn't have the mic in the same position on the same test with both amps until this last round. I suppose I could re-run the tests a few more times and then average them out against each other. But, looking at that data, I was just curious if those peaks were likely due to some interaction with the amp.

Thanks.

 

[radix]

But, looking at that data, I was just curious if those peaks were likely due to some interaction with the amp.

That's my guess, it's either an interaction with the speaker or the amp itself has a small fault and is just noisy at those frequencies. If you have a different speaker or even a high watt resistor (e.g. a 5W-10W non-inductive), you could see if the amp itself measures clean without the speaker.

 

[MaxwellsEq]

Interesting!

If amplifier J generates less distortion than amplifier L in your setup, that would be more reliably measured at the speaker terminals. This would eliminate background noise, inevitable inconsistencies between in-room samples etc.

A speaker's distortion should normally vastly exceed that of an amplifier.

Do you have any EQ setup on one amplifier that's not replicated on the other?

 

[mhardy6647]

I ran the tests multiple times and got roughly the same results each time.

I suspect that the answer to your question lies right here.
Value of n is important.
You also need to know something about the trial-to-trial variability of the measurement. I am simple-minded experimental scientist, so I default to simple minded statistics unless circumstances dictate goin' high-end!
Looking at standard deviation (with a suitable value of n) makes obvious assumptions about the variability of the measurements, but it is usually straightforward to do and will allow objective (even if not utterly accurate) comparisons of results.
This (i.e., the method described and illustrated in the first post) is a very indirect measurement of THD in the sense that there are many variables contributing to the measured result(s).
If the variation difference in the measured THD at, say, that 550-ish Hz peak between the two amplifiers is less than the measured SD of the methodology for one amplifier, there is no statistically meaningful difference.
Of course, 'statistically meaningful' can be judged objectively by simple methods (or complicated ones, as required, or - if one's like me -depending on how lazy one is ).
Absent some insight into the variability of the assessment method, this is anecdotal -- on the order of the familiar "babies born when the moon is full" observation.