PMA-NC252MP power amplifier

[pma]

Could you add your own DC detection and pull nFatal low via J7.5 to trigger the built in DC protection shutdown?

I can add my safe DC protection circuit, however as I have stopped selling audio components in 2012, this info would probably not being much helpful. However, I am open to technical discussion of the issue.

 

[somebodyelse]

I phrased it poorly, but intended it more generally to cover anyone like @Paul Ebert who may wish to improve the DC protection on a hypex amp they own. A quick read of an old datasheet made me think nFatal on J7.5 could be used to trigger the built in DC protection (main SMPS shutdown, to be reset by disconnecting mains) instead of using relays or FETs between the amp outputs and the speakers. I'm asking in case I have misunderstood the datasheet, or if you know that it doesn't behave that way, but it's probably something that should be asked of Hypex. Now I know of the issue I might look more closely at the DC protection on my UcD180/SMPS400A180

 

[pma]

To test amplifier with various complex impedance loads, I have prepared a combination of 4.7ohm/200W resistor in parallel with series connection of 2.2uF polypropylene capacitor and 1.6uH inductor with 0.3ohm resistance. 5 amplifiers were tested and NC252MP is the only one that failed in this test - it is unstable and distortion is higher by 40 dB compared to resistive load. The new dummy load impedance plot is

Capacitive reactance is to simulate speaker like this

Test results

Hypex NC252MP	class D	Failed
Hypex UcD180HG	class D	Passed
AIYIMA A07	class D	Passed
A250W4R	class AB bipolar	Passed
PM-AB2	class AB Mosfet	Passed

 

[pma]

Measurements into 2ohm resistive load

There were some issues in that review file

AUDIOPHONICS MPA-S250NC Amp Review

This is a review and detailed measurements of the AUDIOPHONICS MPA-S250NC stereo Ncore class D amplifier. It is on kind loan from a member and costs 415,83 € tax excl. (US $444). The enclosure is absolutely adorable! While it is fair bit deep, the other two dimensions are quite small...

www.audiosciencereview.com

to measure NC252MP into 2ohm with PowerCube/AP, so I have decided, after several preliminary posts on the subject, to measure my NC252MP implementation into 2ohm load. I have improved my test rig a bit and here are the results:

THD+N vs. power into 2ohm at 50Hz (blue), 1kHz (orange) and 5kHz (green) with 22kHz measuring BW

The distortion knee at 1kHz is at 220W/2ohm


THD+N vs. frequency into 2ohm at 140W with 22kHz measuring BW

 

[Head_Unit]

For decades, the common practice was to use at least 80kHz measuring BW. This has changed after the introduction of class D amplifiers, that were unable to compete under standard conditions of distortion measurements.

As a comparison, below please see THD+N vs. power measurements Hypex NC252MP [vs] A250W4R class AB amplifier...With the higher measuring BW, the class D amplifier cannot compete to an average class AB amplifier.

I'm somewhat confused. My understanding of this history is that the very high frequency "pilot tone" of a Class D could drive audio analyzers into slew rate limiting, hence the need for filtering. I never thought of it as "oh we're going to change the bandwidth since they can't compete" although looking at the filter bandwidths that sure is a side effect.
- Then it appears some such filters do or did have quite a wide bandwidth but some barely 20 kHz, and not explained why aren't they all wideband-??? https://www.ap.com/analyzers-access...mily-switching-amplifier-measurement-filters/
- Your curves show THD+N: how much is the actual distortion performance of the Class D inferior to the AB, and how much is it the ultrasonic noise pushing up the numbers? What do these curves look like if both are measured with the filters?

 

[pma]

- Your curves show THD+N: how much is the actual distortion performance of the Class D inferior to the AB, and how much is it the ultrasonic noise pushing up the numbers? What do these curves look like if both are measured with the filters?

Measurements in post #164 are measured with a 25kHz analog LPF filter behind the amp and 22kHz measuring bandwidth (digital processing). Thus, the THD+N plot is pointless above 11kHz concerning distortion, because H2 of 11kHz is exactly at 22kHz and we have nothing higher calculated, above 11kHz we can see only noise contribution. However, this is how the Hypex presents their measurements, to make them good looking. I have chosen the 22kHz BW for the reason of possibility of a comparison with datasheet plots. Earlier, I have posted a lot of measurements with 45kHz and 90kHz bandwidth. The THD+N worsening with 45kHz and 90kHz BW goes at expense of both noise and distortion. Distortion components above 22kHz go quickly high in amplitude and the same can be said about noise bottom, which is noise shaped with considerable elevation above 22kHz. Nothing is simple and you should make your own conclusions. I am not the one make definitive statements.

Comparison of processing BW22kHz x BW45kHz

Another plot - comparison of THD+N vs. power at 1kHz into 2ohm and 4ohm.

 

[Davide]

Could you add your own DC detection and pull nFatal low via J7.5 to trigger the built in DC protection shutdown?

I repeat this question. The idea seems good to me because it means having to put only a detection circuit and not also an interruption one, which is problematic or more expensive.
Neurochrome's Guardian-86 board for example costs $99 per channel (plus VAT plus shipping to Europe in my case). More than the NC252MP in practice. Nonsense.
The various chinese boards available online have zero technical data neither schematic, therefore impossible to understand if they are suitable.
The circuit shown on sound-au has less than ideal trigger threshold in my opinion, at least when compared to Guardian-86.
So the question is, is it possible to make a sense circuit as good as Neurochrome's to drive nCore's nFatal PIN? And if so, how?

 

[pma]

Another interesting measurement a la PowerCube, NC252MP THD and THD+N vs. output voltage at 1kHz into complex load 3.3 ohm magnitude, -45° phase at 1kHz

Distortion is low with the knee at 31.70V, which makes 304.5W/3.3ohm

 

[pma]

So the question is, is it possible to make a sense circuit as good as Neurochrome's to drive nCore's nFatal PIN? And if so, how?

Everything is possible. Are you asking for a design or what?

 

[Davide]

Everything is possible. Are you asking for a design or what?

I'm asking if it is technically feasible and if so what could be such a solution, understood as an electrical circuit or a commercial product.

 

[pma]

I'm somewhat confused. My understanding of this history is that the very high frequency "pilot tone" of a Class D could drive audio analyzers into slew rate limiting, hence the need for filtering.

The LPF filter that I use to cut switching frequency residuals is a 2nd order RLC filter with slope -12dB/oct. It suppresses the 400kHz switching frequency effectively, but is not any brickwall near the border of audio band, leaving the distortion components above 20kHz to still have an effect. The comparison of NC252MP frequency response without this LPF and with the LPF is below:

and the effect of cutting the switching frequency residuals:

NC252MP pink noise without LPF NC252MP pink noise with LPF

 

[pma]

Another interesting measurement a la PowerCube, NC252MP THD and THD+N vs. output voltage at 1kHz into complex load 3.3 ohm magnitude, -45° phase at 1kHz

Though the NC252MP has been doing well in the 3.3ohm/-45° complex load test (4.7ohm//33uF at 1kHz), it is unable to work with 4.7ohm//2.2uF, as I have shown and stated earlier, nothing has changed here.

After connecting the 2.2uF capacitor in parallel with the load resistor, the NC252MP starts to oscillate and the zero input output noise spectrum looks like this:

We can see oscillations near 70kHz with 1Vrms level and unfortunately also the spurious frequencies in audible band near 8kHz, which are visible also in audio analyzer measurements -

and they result in a very high noise level through the 22kHz band, of 5.7mVrms. The spikes are unstable and vary in amplitude with time.

So, NC252MP and capacitive load of 2uF is no, no, though it worked well with 4R7//33uF.

 

[Silly Valley]

Well, where does this leave us? Is this a corner case and less than ideal from an engineering standpoint or does this have the potential to be an issue for average 4 ohm speakers?

 

[EJ3]

Thank you, I think if you read this thread you would understandd why the case is "big" with big heatsinks. I want to run it at lowest temperature under any possible conditions like long-term full power. I
and the temperature will be increased much much more. The second reason is that the the 19" wide case fits to my audio components. A small box does not. I have no need to use small boxes and to reduce life span and product reliability. However I know - who cares .
So this is my design philosophy, I do not care about miniaturization at expense of reliability.

I am one who cares. I run my 3 (that I have operating in my main system at any one time [I have six & 2 NAD 2100's [so that I can swap a spare 2200 in when I send one out for a refurbishment {about every 6 or 7 years now] ) NAD 2200's pretty hard.
Since around 1984 the only ones that have blown were because the power company's transformer on the power pole blew out.
Around 2010 is when I got around to having Peter Williams at Quirk Audio retromod them (which was more extensive than a cap replacement). But they have never been any trouble unless an outside force created that trouble.
This is the service life that I expect. And only dainty amps will do this (maybe) in a dainty case.

 

[EJ3]

I myself anyway, like to see the 8 ohm rating followed by how much extra is available into a 4 ohm load (I remember being shocked how some of the now vintage US high end behemoths doubled their power from 8 to 4 to 2 ohms and some even managed not to melt into 1 ohm as presented by certain panel speakers back then. Next to see how the distortion changes into evermore cruel loads and out put levels.

That's my excuse and I'm sticking to it 100W into 8 ohms at the distortion 'knee' and hopelly double into 4 with little to no increase in distortion Reading other replies does now make me think that as few speakers are nominally 8 ohms now, that measurement rating is out of date - a factor of my being around and about for far too long now

Anyway, please keep doing what you're doing, 'cos I and others find it fascinating

And double again from 4 Ohms to 2 Ohms with little to no increase in distortion.