• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Amplifier "sample test reviews" using Paul's Multitone Analyzer software

OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
As @Sokel points out, the IMD calculation can be specified right on the definition of the test signal. Just add SMPTE to the end of the signal text.

I am specifically concerned of CCIF 19+20kHz IMD calculations.

REW shows this:
Audiobuff_Steppedlevel_IMD_CCIF.png


however Multitone Analyzer shows this (forget the X-axis shift of 3dB)
Audio Buffer CCIF 19+20kHz and  TD+N vs. level.png

There is a huge difference in TD+N calculation

But, in case of 1kHz THD and THD+N level sweeps, both REW and MTA give almost same results:
Audiobuff_Steppedlevel_THDN.png

Audio Buffer THDN vs. level at 1kHz.png


?????
 

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,630
Likes
10,205
Location
North-East
I am specifically concerned of CCIF 19+20kHz IMD calculations.

REW shows this:
View attachment 262553

however Multitone Analyzer shows this (forget the X-axis shift of 3dB)
View attachment 262554
There is a huge difference in TD+N calculation

But, in case of 1kHz THD and THD+N level sweeps, both REW and MTA give almost same results:
View attachment 262555
View attachment 262556

?????

Hard to tell without knowing what settings were used. You may want to post the screen shot of the Spectrum settings panel in MT, as that has a number of values that can affect the computation.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Hard to tell without knowing what settings were used. You may want to post the screen shot of the Spectrum settings panel in MT, as that has a number of values that can affect the computation.
I think not so hard.

All the spectrum components circled in red must be involved in the CCIF IMD calculations. And they are not, as one can see from the plot legend. Numbers in red rectangles are too optimistic, it seems that only difference tone F2-F1, 19.5-18.5=1kHz is taken into account. This is not enough.

CCIF IMD.png


It should be like this:
IMD_CCIF_REW.png


BTW, my MT setting is as installed default, V1.0.71
 

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,630
Likes
10,205
Location
North-East
I think not so hard.

All the spectrum components circled in red must be involved in the CCIF IMD calculations. And they are not, as one can see from the plot legend. Numbers in red rectangles are too optimistic, it seems that only difference tone F2-F1, 19.5-18.5=1kHz is taken into account. This is not enough.

View attachment 262601

It should be like this:
View attachment 262602

BTW, my MT setting is as installed default, V1.0.71

I'll check it out, thanks.
 

Sokel

Master Contributor
Joined
Sep 8, 2021
Messages
5,840
Likes
5,774
The difference looks the same as between CCIF and DIN.
If it helps.

CCIF.PNG


CCIF


DIN.PNG


DIN
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
SYM2 - my Single Ended (SE) to balanced converter with OPA627 and 2 x ADA4898-1

This thing is useful to convert single ended to balanced signal, it has high input impedance of JFET opamp and high output current capability of Analog Devices ADA4898 opamps. "N" is limited by my setup with Topping D10s DAC.

SYM2 THD at 1kHz 3.6Vrms.png


Comparison to REW THD and THD+N (SINAD)
SYM2 1k spectrum.png


SYM2 IMD 250Hz+8kHz at 3Vrms.png


SYM2 IMD 18.5+19.5kHz at 2.2Vrms.png


SYM2 TIM 3.18kHz sq + 15kHz sine at 3.2Vrms.png


SYM2 THD and TD+N at 1kHz vs. level (0dB = 3.6V).png


SYM2 CCIF IMD 18.5+19.5kHz levelsweep.png


SYM2 TIM 3.18kHz sq + 15kHz sine vs. level.png


SYM2 THD and TD+N vs. frequency.png

(Stepped frequency measured at 3.6Vrms fundamental)

... and again same in REW
SYM2 THD and TD+N vs. frequency REW.png
 
Last edited:
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Measurements on NC252MP

The results are nice and predictable. Looking forward to X-axis cal option in [W] ;).

NC252MP THD and TD+N vs. level at 1kHz-4ohm.png


NC252MP THD and TD+N vs. level at 5kHz-4ohm.png


NC252MP THD and TD+N vs. level at 20Hz-4ohm.png


NC252MP THD and TD+N vs. frequency at 25W-4ohm.png


-----------------
One more, DIN IMD vs. level :
NC252MP DIN IMD vs. level at 4ohm.png

Interesting.
 
Last edited:
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
This might be interesting as well - small signal, low power 250mW/4ohm/1kHz
NC252MP THD and TD+N at 250mW-4ohm.png


and very high power 240W/4ohm at higher frequency 5kHz. Very good result.
NC252MP THD and TD+N at 240W-5kHz-4ohm.png


@Sokel + the others, may I kindly ask you not to post videos in this thread, I would like to keep some technical value. Thank you very much.

@pkane : Paul, is the ENOB calculation correct? Please check these two images, thank you.
 
Last edited:
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
This plot shows THD and THD+N vs. frequency of NC252MP at very high power 240W/4ohm. Please note the loss of maximum clean power at lower frequencies below
some 800Hz.
NC252MP THD and TD+N vs frequency at 240W-4ohm.png
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
This plot shows THD and THD+N vs. frequency of NC252MP at very high power 240W/4ohm. Please note the loss of maximum clean power at lower frequencies below
some 800Hz.
View attachment 263065
Interesting, if vexing. Seems like this is a little worse than the previous 20 Hz sweep you did? Looks like distortion dominates given the THD and TD+N curves are very close. Any idea why the knee and rise below 1 kHz or so, then flatlining around 200 Hz? Is this using their SMPS for power?
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Interesting, if vexing. Seems like this is a little worse than the previous 20 Hz sweep you did? Looks like distortion dominates given the THD and TD+N curves are very close. Any idea why the knee and rise below 1 kHz or so, then flatlining around 200 Hz? Is this using their SMPS for power?
I do not think it is worse now, because the previous plot was at 220W (below), and now it is at 240W. I assume the reason is SMPS, quite for sure. The supply voltage either drops or has high ripple or both. It is still within specs, because Hypex declares 250W/4ohm/1kHz/1%. The older measurement was made by a bit different method, true sweep, now in MTA it is done in stepped frequency, frequency measurement points highlighted.
The effect seen in the spectrum, above certain power/frequency point, is a sudden rise of noise floor (and distortion components as well, like in case of clipping).

1675791581433.png


NC252MP thdfreq multiwatt-labeled.png
 
Last edited:

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
Thanks! That leads to a whole host of other questions that just go deeper down the rabbit hole so I'll leave off. You'd think the higher feedback at lower frequencies would reduce the distortion, but if the supply ripple is going up and is correlated with the signal (or the voltage is just dropping and so it heads into clipping) then that makes a lot of sense. The roughly 200~800 Hz transition region is a bit curious, since it basically shelves before and after that, but maybe the ripple (or whatever) maxes out by a little over 200 Hz and then is somewhat constant below that. In the old days I'd double the supply decoupling (or maybe add a capacitive multiplier) and see if anything changes; harder to do with an SMPS, and less impact since the supply is regulated.

Good stuff, thanks for doing all of this!
 
Last edited:
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Thanks! That leads to a whole host of other questions that just go deeper down the rabbit hole so I'll leave off. You'd think the higher feedback at lower frequencies would reduce the distortion, but if the supply ripple is going up and is correlated with the signal (or the voltage is just dropping and so it heads into clipping) then that makes a lot of sense. The roughly 200~800 Hz transition region is a bit curious, since it basically shelves before and after that, but maybe the ripple (or whatever) maxes out by a little over 200 Hz and then is somewhat constant below that. In the old days I'd double the supply decoupling (or maybe add a capacitive multiplier) and see if anything changes; harder to do with an SMPS, and less impact since the supply is regulated.

Good stuff, thanks for doing all of this!

Thank you Don. Let me show 2 less conventional plots, related to the phenomenon we are just discussing.

First one is Noise (relative) vs. power (into 4ohm) at 3 frequencies, 22Hz, 1kHz, 5kHz. Just noise.
NC252MP noise vs. power.png

Green link is for 22Hz and it shows reduced "clean power" at this frequency (elevated noise). Unfortunately the level step was 1dB, which is quite coarse.


The second plot shows H3 harmonics vs. power again at 3 same frequencies
NC252MP H3 vs. power.png

As the effect is most pronounced in H3 (more than in H2), we may say that it is almost symmetrical clipping, related to PSU voltage drop.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
Thank you Don. Let me show 2 less conventional plots, related to the phenomenon we are just discussing.

First one is Noise (relative) vs. power (into 4ohm) at 3 frequencies, 22Hz, 1kHz, 5kHz. Just noise.
View attachment 263155
Green link is for 22Hz and it shows reduced "clean power" at this frequency (elevated noise). Unfortunately the level step was 1dB, which is quite coarse.


The second plot shows H3 harmonics vs. power again at 3 same frequencies
View attachment 263165
As the effect is most pronounced in H3 (more than in H2), we may say that it is almost symmetrical clipping, related to PSU voltage drop.
Cause and effect is pretty clear from those, thanks Pavel! Jives with the earlier plots that imply it is mostly distortion, and I agree looks like supply drop at low frequencies.
 
Top Bottom