• 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!

PMA-NC252MP power amplifier

OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Not necessarily surprising given we already have THD+N measurements as a function of power for various frequencies (obviously Amir used 45 kHz bandwidth instead of 24 kHz) which show the high and low frequencies giving out first, but the log scale makes the difference less obvious at first glance.

Yes, it is in conformance with @amirm measurements of various amplifiers with NC252MP modules. I might add some system details and more measurements of my NC252MP amp.

About a half year ago I started to use a measuring system based on Topping D10s DAC and E1DA Cosmos ADC. It is not as good as Audio Precision APX555, however it has lower distortion than Audio Precision APX525. Limits shown below:

D10s-to-E1DA_AA.png D10s-OPA1622-E1DACosmos_thdnfreq.png

STEPS program allows me to make multiple THD vs. level measurements in one graph, however the X-axis is calibrated in Volts, not in Watts. It is the other view to the same thing as shown in my post #92. I would say that the graph in the post #92 is more revealing.

PMA-NC252MP thdlevel multifreq.png



and a common THD and THD+N at 1kHz/4ohm. Not so much telling, IMHO.

NC252MP_E1DA_thdnlevel_1k_4R.png
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Another interesting measurement is a multitone TD+N distortion vs. output power. REW V5.20 can do it (thanks @JohnPM ) - I am not even sure that Audio Precision can do it. Below is the plot recorded from my NC252MP, white curve is the TD+N vs. power, blue line shows system residual noise floor.

PMA-NC252MP multitonelevel.png

[correction - 31 tone multitone: Stepped Multitone measurement using 31 tones from 19.78 Hz to 20 kHz at 1/10 decade]
 
Last edited:
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
This graph shows various spectrum plots in one. The green one is a residual amplifier noise. Please note how the noise floor gets modulated with different input signals. System input scale and measuring configuration is always the same. 0dBFS refers to 45Vrms.

PMA-NC252MP multispectra.png
 

lashto

Major Contributor
Forum Donor
Joined
Mar 8, 2019
Messages
1,045
Likes
535
This graph shows various spectrum plots in one. The green one is a residual amplifier noise. Please note how the noise floor gets modulated with different input signals. System input scale and measuring configuration is always the same. 0dBFS refers to 45Vrms.

View attachment 250509
nice graph, I would vote for adding it to all amps.

The modulated noise floor is pretty bad news, any idea if that's true for all Hypex/Purifi modules?
 
  • Like
Reactions: pma

EJ3

Major Contributor
Forum Donor
Joined
Dec 10, 2019
Messages
2,153
Likes
1,662
Location
James Island, SC
nice graph, I would vote for adding it to all amps.

The modulated noise floor is pretty bad news, any idea if that's true for all Hypex/Purifi modules?
Every CLASS D AMP that I have listened to (not many but still, has given me the Heebie-Jeebies' for some unknown, undefined reason. Perhaps this is why? The Heebie-Jeebies' is the reason that I have not even thought about changing from my vintage triple NAD 2200 setup (with 2 more on the way).
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
I have not even thought about changing from my vintage triple NAD 2200 setup (with 2 more on the way).
Your amp is good enough IMO, if in a good technical shape.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
The modulated noise floor is pretty bad news, any idea if that's true for all Hypex/Purifi modules?

I am not sure, I can only speak about the NC252MP I have. I will make similar measurements on a good performing class A or AB amplifier. Preferably on one of my designs, as I know everything about their circuits.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
I will make similar measurements on a good performing class A or AB amplifier. Preferably on one of my designs, as I know everything about their circuits.
Measurements done. There is almost no trace of noise modulation in my PM-A4 class A amplifier. Slight rise near 60Hz (SMPTE IMD) is an effect of the FFT window. Same measurement setup as in case of NC252MP, but without LC low pass filter (no need). PM-A4 has is thread here:

Noise modulation measurements:
PM-A4 noisemod.png


Another one with multitone
PM-A4 noisemod_multi.png


PM-A4 SMPTE.png
 
Last edited:

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,383
Likes
24,749
Location
Alfred, NY
Every CLASS D AMP that I have listened to (not many but still, has given me the Heebie-Jeebies' for some unknown, undefined reason. Perhaps this is why? The Heebie-Jeebies' is the reason that I have not even thought about changing from my vintage triple NAD 2200 setup (with 2 more on the way).
I suspect I can guess the answer. And it's not stuff going on electrically at 130dB below 45V (0.3 uV).
 

lashto

Major Contributor
Forum Donor
Joined
Mar 8, 2019
Messages
1,045
Likes
535
Every CLASS D AMP that I have listened to (not many but still, has given me the Heebie-Jeebies' for some unknown, undefined reason. Perhaps this is why? The Heebie-Jeebies' is the reason that I have not even thought about changing from my vintage triple NAD 2200 setup (with 2 more on the way).
you might wanna define those heebie-jeebies a bit :)

The audible effects of noise floor modulation are pretty hard to estimate from just a graph. You may look at this thread for some pointers. Generally, the effects should be somewhat similar to those of IMD, usually described as dull/lifeless/harsh/hard. However, those effects are audible at levels waaay above what we see in pma's measurements.

So, it's not possible to conclusively link the above measurements to any audible "heebie-jeebies". But it's also impossible to 100% exclude the "jeebies", there might even be some not-so-audible ones caused by critical band effects like auditory roughness. Anyway, all this stuff is highly speculative, especially at such a low level.

Long story short: if you really-really want to stay on the safe side of those (more or less imaginary) "jeebies", just don't buy classD. Otherwise, we can't really talk about "jeebies" until there are precise definitions/measurements and someone actually hears one (or feels it somehow).
 
Last edited:

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
A class-D amp effectively outputs a pulse train modulated by the input signal (pulse width, density, etc. changes with the input signal). That will lead to a modulated noise floor dependent upon the signal and FFT (signal relationship to FFT length and weighting). But as @SIY points out the modulation is usually well below audibility, masked by the main signal. And it happens for data converters (ADCs/DACs) as well. It looks scary, but not likely an issue in the real world.
 

lashto

Major Contributor
Forum Donor
Joined
Mar 8, 2019
Messages
1,045
Likes
535
And it happens for data converters (ADCs/DACs) as well. It looks scary, but not likely an issue in the real world.
one more reason to use turntables+tube amps :)

There was an interview somewhere with Putzeys about his ultra-expensive Tambaqui DAC and he was very proud that it (suposedly) has "no noise floor modulation". Wish the price was more 'human'..

P.S.
the interview I remember went into more detail about that modulation but here's one quote that also mentions it https://the-ear.net/review-hardware/mola-mola-tambaqui-digital-analogue-converter/
 
Last edited:

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
one more reason to use turntables+tube amps :)
Hmmm... If you want a noise floor around 60~65 dB or so, modulated by various warps, wobbles, ticks, and pops.
 

lashto

Major Contributor
Forum Donor
Joined
Mar 8, 2019
Messages
1,045
Likes
535
Hmmm... If you want a noise floor around 60~65 dB or so, modulated by various warps, wobbles, ticks, and pops.
ticks & pops are just lovely, they remind me of childhood & grandpa's crappy turntable :)

More generally, there are so many complaints about digital sounding "hard" or whatever. Most people around here think that it's all some sort of delusion. I tend to think that there are actually a few "somethings" in these measurements that we are not interpreting right/fully. I don't have any proof or anything close to, just a feeble hunch... which may of course be 100% wrong
 

EJ3

Major Contributor
Forum Donor
Joined
Dec 10, 2019
Messages
2,153
Likes
1,662
Location
James Island, SC
Your amp is good enough IMO, if in a good technical shape.
One of my three triplets is the one Amirm tested in the NAD 2200 Vintage Amplifier test (and I only use the LAB inputs which gave a SINAD of 95. I run them at 4 ohms.
 
Last edited:

EJ3

Major Contributor
Forum Donor
Joined
Dec 10, 2019
Messages
2,153
Likes
1,662
Location
James Island, SC
you might wanna define those heebie-jeebies a bit :)

The audible effects of noise floor modulation are pretty hard to estimate from just a graph. You may look at this thread for some pointers. Generally, the effects should be somewhat similar to those of IMD, usually described as dull/lifeless/harsh/hard. However, those effects are audible at levels waaay above what we see in pma's measurements.

So, it's not possible to conclusively link the above measurements to any audible "heebie-jeebies". But it's also impossible to 100% exclude the "jeebies", there might even be some not-so-audible ones caused by critical band effects like auditory roughness. Anyway, all this stuff is highly speculative, especially at such a low level.

Long story short: if you really-really want to stay on the safe side of those (more or less imaginary) "jeebies", just don't buy classD. Otherwise, we can't really talk about "jeebies" until there are precise definitions/measurements and someone actually hears one (or feels it somehow).
Ah, yes: I wish I could (define those heebie-jeebies), as I do not seem to have this issue with any A/B amps that I know of. It is something that I cannot identify. It's just a somewhat odd feeling that I get for some unknown (to me, anyway) reason. It does not happen instantly but after perhaps 30-45 minutes of listening to a class D amp.
Oh well, maybe it is just me. I am happy with what I have and what I am doing with it. So there's that.
 
Last edited:

lashto

Major Contributor
Forum Donor
Joined
Mar 8, 2019
Messages
1,045
Likes
535
Ah, yes: I wish I could (define those heebie-jeebies), as I do not seem to have this issue with any A/B amps that I know of. It is something that I cannot identify. It's just a somewhat odd feeling that I get for some unknown (to me, anyway) reason. It does not happen instantly but after perhaps 30-45 minutes of listening to a class D amp.
Oh well, maybe it is just me. I am happy with what I have and what I am doing with it. So there's that.
'somebody' else with pretty clear classD 'jeebies' issues :)
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Measurement of THD+N vs. output power into 4ohm, at 50Hz, 1kHz and 5kHz, measuring BW=22kHz

PMA-NC252MP_THD+N_50-1k-5k.png
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Based on some interesting findings I collected during testing of Paul's @pkane Multitone Analyzer software (here) I have decided to post some of those measurements into this thread, because they are directly related to the NC252MP module behaviour at high power, depending on frequency of the test signal. Maximum clean power of NC252MP is a bit reduced as the test frequency goes down below 1kHz and it is of course most pronounced at 20Hz. There is also an interesting point (power), at any frequency, when the noise floor of the power spectrum suddenly jumps of tenths of dB upwards. So I made measurements with smaller level step (0.4dB) and made them for separated THD, N and THD+N vs. power. Using THD+N (or SINAD) vs. power only confuses the issue, as the mixture of distortion and noise masks the origin of the issue. Clipping in time domain was measured as well.

NC252MP THD 4freq vs. power.png


NC252MP N 4freq vs. power.png


NC252MP THDN 4freq vs. power.png


NC252MP 1kHz clipping.png


NC252MP 20Hz clipping.png


The behaviour seen here above is almost certainly due to SMPS PSU voltage drop at lower frequencies, when capacitor bank (charging and discharging rate) is unable to maintain the clean output voltage at same level as when the test frequency is above 1kHz.

It is of course not a big issue during real life listening, however it is interesting at least.
 

Davide

Senior Member
Joined
Jul 6, 2020
Messages
460
Likes
171
Location
Milan, Italy
Based on some interesting findings I collected during testing of Paul's @pkane Multitone Analyzer software (here) I have decided to post some of those measurements into this thread, because they are directly related to the NC252MP module behaviour at high power, depending on frequency of the test signal. Maximum clean power of NC252MP is a bit reduced as the test frequency goes down below 1kHz and it is of course most pronounced at 20Hz. There is also an interesting point (power), at any frequency, when the noise floor of the power spectrum suddenly jumps of tenths of dB upwards. So I made measurements with smaller level step (0.4dB) and made them for separated THD, N and THD+N vs. power. Using THD+N (or SINAD) vs. power only confuses the issue, as the mixture of distortion and noise masks the origin of the issue. Clipping in time domain was measured as well.

View attachment 263292

View attachment 263293

View attachment 263294

View attachment 263295

View attachment 263296

The behaviour seen here above is almost certainly due to SMPS PSU voltage drop at lower frequencies, when capacitor bank (charging and discharging rate) is unable to maintain the clean output voltage at same level as when the test frequency is above 1kHz.

It is of course not a big issue during real life listening, however it is interesting at least.
I assume the test is on one channel only, so I assume that with both channels running the problem may be more apparent as they share same power supply.
 
Top Bottom