NC252MP (class D) vs. A250W4R (classAB) burst measurements into 4ohm//2.2uF load

[tmtomh]

I find it amazing that in a forum where surely many would feel an Apple dongle is inadequate

Your initial premise is flawed, making the rest of your comment a bit of a strawman argument.

 

[Nyattnyatt]

I tried some Piezo film tweeters from Gallo. The amp (class D)I am using has been well reviewed here. It managed to heat up the film and read 4V to 6+V at idle, with a frequency (on my Fluke meter) of 350 kHz, and I shut things down when the film was getting warm. Again, with NO signal. I will try an experiment on a different Class D amp. I also read over 2v on a bullet tweeter that measurably heated the voice coil using an instant read thermometer. Maybe 4 degrees F over room temp. That time, the frequency was around 450 kHz. That is why I am wondering if an external filter could help stop these oscillations.

 

[atmasphere]

I find it amazing that in a forum where surely many would feel an Apple dongle is inadequate and worship some Topping DAC or AHB2 amp etc with S/N of 120 dB, that so many say 1.7V is irrelevant. Please note I am NOT saying it is a problem...but it seems hypocritical to diss a dongle whose performance is most likely 99% transparent to 99% of the people with 99% of the music-and then dismiss the 1.7V etc as negligible.
- Even if it's not a real world problem, wouldn't you prefer the amp NOT do this?
- The bigger question, what happens with other class D amps? Is this inherent to the class? (based on @atmasphere's post I'd think so, unless the design extends the feedback loop out past the filter as a colleague of mine was one of the first to do back in the day. Then I'd *think* this doesn't happen?)

Anyway again a big kudos to @pma for doing these investigations, it's always interesting to find odd behaviors, which sometimes lead to insight which leads to improvement.

The 1.7V is negligible because the amp isn't actually oscillating at that frequency. It is actively trying to prevent one: since you can see its feedback is rejecting the oscillation that is trying to set up.

This behavior is predictable with self-oscillating class D amps. Most self oscillating amps include the output filter in the loop.

A lot of solid state amps from years ago would go into oscillation and blow up when presented with this sort of load. The capacitance would mess with the feedback such that it was pushed beyond the phase margin of the design. Oscillation results.

I tried some Piezo film tweeters from Gallo. The amp (class D)I am using has been well reviewed here. It managed to heat up the film and read 4V to 6+V at idle, with a frequency (on my Fluke meter) of 350 kHz, and I shut things down when the film was getting warm. Again, with NO signal. I will try an experiment on a different Class D amp. I also read over 2v on a bullet tweeter that measurably heated the voice coil using an instant read thermometer. Maybe 4 degrees F over room temp. That time, the frequency was around 450 kHz. That is why I am wondering if an external filter could help stop these oscillations.

Do you know the switching frequency of the amps you tried? 450KHz is high enough (350KHz is too) to actually be the switching frequency. I would expect the inductance of the tweeter to shut down any heating. If the amp was also high power, the residual might have been significant depending on the output filter design.

 

[restorer-john]

A lot of solid state amps from years ago would go into oscillation and blow up when presented with this sort of load. The capacitance would mess with the feedback such that it was pushed beyond the phase margin of the design. Oscillation results.

What specific amplifiers? You say 'a lot' from 'years ago' would 'blow up'. Present some evidence.

 

[atmasphere]

What specific amplifiers? You say 'a lot' from 'years ago' would 'blow up'. Present some evidence.

Apparently you weren't around for that fun.
Here's an example. There are many more. Around this time manufacturers learned to place ZOBEL networks at the output of their amps. That helped but wasn't 'sure fire' if you see what I did there.

 

[Nyattnyatt]

The 1.7V is negligible because the amp isn't actually oscillating at that frequency. It is actively trying to prevent one: since you can see its feedback is rejecting the oscillation that is trying to set up.

This behavior is predictable with self-oscillating class D amps. Most self oscillating amps include the output filter in the loop.

A lot of solid state amps from years ago would go into oscillation and blow up when presented with this sort of load. The capacitance would mess with the feedback such that it was pushed beyond the phase margin of the design. Oscillation results.


Do you know the switching frequency of the amps you tried? 450KHz is high enough (350KHz is too) to actually be the switching frequency. I would expect the inductance of the tweeter to shut down any heating. If the amp was also high power, the residual might have been significant depending on the output filter design.

The switching frequency is 750-800kHz. And the CDT tweeters are piezo film tweeters, so essentially capacitors. No inductance. That’s why I was wondering if an add’l filter (inductors) on the output, after the feedback of the internal output filter, might stop the interaction of the Piezo capacitance combining with the output capacitor that IS in the feedback loop. To be clear, I am NOT a EE, just a serious hobbyist looking for the best sound from my system.

 

[restorer-john]

Apparently you weren't around for that fun.
Here's an example. There are many more. Around this time manufacturers learned to place ZOBEL networks at the output of their amps. That helped but wasn't 'sure fire' if you see what I did there.

Zobel/Boucherot have been on amplifiers forever. Way before I wound my first inductor, at 11yo for a 'zobel' on my first kit build amplifier (1976)

It didn't blow up and neither did many commercial offerings of the time which also contained networks either on the main PCBs or right on the speaker terminals.

Your 'example' linked is not even an example.

Let's not try to re-invent history with blanket incorrect statements. Let's stick to facts and specifics when it comes to statements like yours in post #843 above. I can give you literally hundreds of examples of well designed amplifers from the 1970s onwards that did not oscillate, blow up or otherwise behave poorly in the presence of difficult loads (read: L, C or R). All you have to do is present a single commercially sold topology, with or without an output network as described, that regularly blew up due to oscillation and we will discuss it.

 

[amirm]

I find it amazing that in a forum where surely many would feel an Apple dongle is inadequate and worship some Topping DAC or AHB2 amp etc with S/N of 120 dB, that so many say 1.7V is irrelevant.

It is ultrasonic noise. It is the definition of following science that says to NOT follow what is inaudible. That 120 dB dynamic range is in the audible band where we have research that shows we need almost that much for a silent channel.

Playing music over Wifi means 2.5 and 5 Ghz carrier. You want us to worry about that too???

 

[amirm]

- Even if it's not a real world problem, wouldn't you prefer the amp NOT do this?

If it is free, sure. If it cost money, no, I wouldn't pay extra for it. To be sure, I want it attenuated to pass emissions standards but once there, I am good.

 

[atmasphere]

Zobel/Boucherot have been on amplifiers forever. Way before I wound my first inductor, at 11yo for a 'zobel' on my first kit build amplifier (1976)

View attachment 336401

View attachment 336403

View attachment 336402

It didn't blow up and neither did many commercial offerings of the time which also contained networks either on the main PCBs or right on the speaker terminals.

Your 'example' linked is not even an example.

Let's not try to re-invent history with blanket incorrect statements. Let's stick to facts and specifics when it comes to statements like yours in post #843 above. I can give you literally hundreds of examples of well designed amplifers from the 1970s onwards that did not oscillate, blow up or otherwise behave poorly in the presence of difficult loads (read: L, C or R). All you have to do is present a single commercially sold topology, with or without an output network as described, that regularly blew up due to oscillation and we will discuss it.

As you say there were plenty of amps that didn't blow up with a capacitive load. But there were those that did. There was no attempt on my part to 'rewrite history'; 'a lot' isn't the same as 'all' or 'most'. Just a lot. I don't see an incentive for discussing this matter, since its pretty well known.So I'm not responding past this post; its off topic anyway.
https://www.diyaudio.com/community/...an-amplifier-under-capacitive-loading.359704/
google search
A lot of opamps are quite similar to textbook power amplifiers so you have similar issues. I'm sure amplifier stability with a capacitive load is a thing, else no-one would think to offer an article about it.
https://resources.altium.com/p/oscillations-clipping-and-ringing-amplifier-stability-analysis
A lot of times these problems went unnoticed until something set it off. That's why I remember the Polk Audio Cobra cable but not the amps it upset.
http://www.roger-russell.com/wire/wire.htm

 

[SIY]

That's why I remember the Polk Audio Cobra cable but not the amps it upset.

I'll name names where I had personal experience: Quatre, Rapaport, SWTP. All long gone.

 

[atmasphere]

I'll name names where I had personal experience: Quatre, Rapaport, SWTP. All long gone.

Ah SWTP... I had a Tiger... yeesh!

 

[DonH56]

I'll name names where I had personal experience: Quatre, Rapaport, SWTP. All long gone.

Yup. And the early Infinity (?) Servo-Statik class-D amplifier IIRC, and maybe a Stax amp? Been a long time... I do remember expecting one of those ultra-wideband HK amps to self-destruct, but it actually passed with flying colors.

I thought SWTP had some cool stuff, but as time went on learned their reliability was (shall we say) less than desired...

 

[SIY]

Ah SWTP... I had a Tiger... yeesh!

I thought SWTP had some cool stuff, but as time went on learned their reliability was (shall we say) less than desired...

The design work was very state of the art for the time- full complementary, output triple with gain, very advanced. Look at it sideways or have the phase of the aether cross a threshold and you had a very nice short burst oscillator. One shot, alas.

 

[atmasphere]

The design work was very state of the art for the time- full complementary, output triple with gain, very advanced. Look at it sideways or have the phase of the aether cross a threshold and you had a very nice short burst oscillator. One shot, alas.

Yes, it was pretty impressive. I received mine in derelict condition after a friend botched the assembly. It was a rat's nest! But I cleaned it up, found the open connections on the circuit board, replaced the damaged weirdo driver transistors, rack mounted it and used it to run the woofers in my system. It was a pretty good example of how the feedback network design was a bit too simplistic- easily affected by load, and a bit too close to comfort on the stability side of things to begin with, just as you describe. If I had connected that thing to the infamous Cobra cables it would have been Bad.

By comparison the module that is the topic of this thread is rock stable. The SWTP Tiger was a good example of what was good and bad about 70s amplifiers that pushed the envelope.

My LP mastering lathe ran the Westerex 1700 mastering electronics. The amps in that setup were scary. Designed in the late 1960s as state of the art silicon based amps, they featured a Darlington output section with gain. I flinched every time I turned them on. They had a 30dB feedback loop nested around them; with a capacitive load there would have been lots of smoke from expensive things.

 

[tubesguy]

I'll name names where I had personal experience: Quatre, Rapaport, SWTP. All long gone.

Sounds like an issue of "The Audio Critic"

 

[pma]

Amplifier burst testing

I would like to return to the topic of burst testing. Recently I have been testing my class AB amplifier (A250W) and my class D amplifier (PMA-NC252MP) into 2.2 ohm load and made some comparisons with 4 ohm load results. And also comparisons between the two amplifiers. First, let me show traditional stepped sine THD vs. power plots (noise not included) of both tested amplifiers into 4ohm and 2.2ohm resistor loads, measured at 1kHz and measuring BW=45kHz, with one exception - NC252MP/4ohm was measured with BW=22kHz, which gives it some advantage. The reason is that this measurement is from my archive.

It is evident that into 4ohm load has NC252MP almost identical power, however lower THD. The situation is different into 2.2ohm load. NC22MP has very sharp knee at 175W, though A250W has softer knee at 245W and still good THD = 0.05% at 300W. A250W is limited to 2R2 by the voltage drop on its linear power supply. In the case of NC252MP it is probably the SMPS maximum current limit, but her I am only speculating. Anyway, the class AB amp is the winner here, regarding usable power to very low ohmic load.

I have done some burst testing of the A250W amplifier then. The test signal is 1kHz sine burst (rectangular window) of 10 cycles length. The plots are measured with a DSO at high sampling rate, for the reason that the DSO is much more useful in capturing single-shot time records and also in capturing of single-shot frequency domain spectrum. The lower resolution of DSO does not matter here, as we are measuring at the clipping boundary, where distortion is quite high by nature, however still hardly audible. Burst power is 568W/2.2ohm, with first 2 cycles "clean" and then there is an obvious PSU related amplitude drop. Still, the highest harmonic H3 is more than 40dB (distortion 1%) below the fundamental H1. Please note that harmonic components are odd only, due to symmetrical clipping nature.

Burst testing might be a good alternative method to test small class D amplifiers, that suffer from adequate thermal design and PSU capacity.

 

[pma]

I have done some burst testing of the A250W amplifier then.

Now, the same set of measurements with NC252MP, 2.2 ohm load and same burst test signal. Again, the burst amplitude was set close to clipping boundary (distortion 1%).

Burst power into 2.2 ohm is now 272W, less than half power of the class AB amplifier tested in my previous post. Even if both amps have same power into 4ohm. This clearly shows that:

1) It is a myth that class D amplifiers are better in handling low impedance load,
2) It is a myth that class D amplifiers are better in handling complex loads (see post #1).

Moreover, output spectrum of class D amplifiers is nastier (HF mess).

 

[Julf]

This clearly shows that:

1) It is a myth that class D amplifiers are better in handling low impedance load,
2) It is a myth that class D amplifiers are better in handling complex loads (see post #1).

It shows that in this specific case, this class D amplifier is not better in handling low impedance load, and not better in handling complex loads.
Not sure we can generalize beyond that.

 

[NTK]

Burst power into 2.2 ohm is now 272W, less than half power of the class AB amplifier tested in my previous post. Even if both amps have same power into 4ohm. This clearly shows that:

1) It is a myth that class D amplifiers are better in handling low impedance load,
2) It is a myth that class D amplifiers are better in handling complex loads (see post #1).

Moreover, output spectrum of class D amplifiers is nastier (HF mess).

Shocking news! An amplifier not designed to deliver lots of power into 2 Ω loads behaved as it was designed.

Why not try something designed for it, such as: