Class D at low volume


Hi,

How do class D amps behave at low volume levels?  My question is general rather than related to a particular amp.  I know there are exceptions, but as a rule, SET’s and class A SS excel at low volume.  What about class D?  Is low volume performance of class D predetermined, all else being equal? Do class D amps have a comfort zone?  Do they distort more at low volume or is it uniform throughout?  For the purpose of this question I am only referring to analog input amps and not the ones that take in PCM (e.g. NAD M2).

The second part of my question is as follows.  I’m interested in some higher end commercial class D amps from the likes of lab.gruppen, powersoft, mc2, XTA, etc. due to their network-ability as in, I can control them, DSP them, and stream digital, all via RJ45, at the same time dispensing with all the extra boxes and cables.  But, they are all of very high power from 100‘s to 1000‘s of watts.  Does this mean that in a domestic setting at low volume they operate much closer to their noise floor or is this different with class D?

Thanks
serge_s

Showing 5 responses by bombaywalla

03-03-15: Kijanki
Class D (PWM) can be easily constructed without negative feedback at all.
I wasn't aware of this. how does the PWM system know that it's tracking the input if there is no feedback? Can you please point me to some references?
but current modules resemble more of Delta-Sigma A/D converters.
so, in current class-D modules, they noise shape the input music signal to move the noise out of band & the resulting pulse train from the delta-sigma block is used to drive the power switches? I suppose that delta-sigma technology makes sense for audio because the ear is very sensitive to distortion & noise? what frequency is the delta-sigma clocked at - also 1MHz like the new gen SMPSs? thanks.
03-04-15: Mitch2
I think the audio SMPS has finally developed to a point where it is has a low enough noise floor & can handle large currents in a compact size.
The Acoustic Imagery Atsahs use the Hypex/Ncore SMPS 1200 switch mode power supply and in my system they sound at least as good as similar powered, high quality solid state, Class A and Class A/B amps I have owned. They are dead nuts quiet and I notice absolutely no listener fatigue. When I listen, I sometimes wonder why some are trying to use linear supplies with the NC1200 amp module, when it sounds this good with the stock SMPS.
Someone i talked to last year about class-D amps & this someone I have some respect for informed me that the best part of the Hypex/NCore modules was their SMPS.... ;-)
SMPS have line and load regulation. Amplifier with SMPS keeps composure during power peaks since voltage does not sag, like it often happens in linear power supplies.
linear power supplies have line & load regulation as well, don't they? one could argue that it might be easier to design better line & load reg for SMPS compared to linear power supplies but i believe that a good linear power supply can also have very good line & load reg. I think a few such supplies are available for Empirical Audio gear from a Paul XXX (I'm forgetting his name) & I also see a good linear power supply in Sanders Magtech & ESL power amps.
Yes, traditionally the manuf of class-A, AB amps have not bothered to put regulated supplies hence the poor line & load reg.
03-04-15: Kijanki
Bombaywalla, Linear power supplies have regulation in preamps, DACs etc. but MOST of power amps are unregulated because of the amount of heat dissipated in power supply (and loss of efficiency).
agree. There might a small change happening here - Sanders Sound Systems Magtech amps have a regulated power supply (& I believe Roger Sanders has applied for a patent for that?).

Look at fig. 2 showing how PWM signal can be obtained by just using linear ramp. Icepower modulator uses sinewave, has two feedbacks etc. - but only for improvements.
http://en.wikipedia.org/wiki/Pulse-width_modulation
thanks for this reference, Kijanki I'm quite familiar with class-D PWM using a ramp waveform.
My question was using delta-sigma instead. There might be some info on that same page you sent the link for. I'll read up.

Also had a question re. how class-D PWM can be designed without feedback like you wrote in your initial post. Got any references that I can read?

My amplifier with bandwidth of 60kHz shifts phase of the 20kHz signal by 20 degree.
agree! The "funny" thing here is that I wrote something very similar in another thread but had Almarg state a contrary view stating that it wouldn't be such a big deal given that the hi freq energy is much less at these frequencies. Perhaps that is true - it's true that the there is generally less energy at the high frequencies. So, even tho' your amp produces a 20 deg phase shift at 20KHz, it might matter very little(?).
What I wrote in that thread specifically was
....This also means that since the ARC Ref 150 bandwidth is just 3X (rather than 8X or 10X) the music bandwidth (of 20KHz) one can expect to hear the amp impart its own phase shift onto the higher frequencies of the music. This can manifest itself in a few ways - the highs could sound rolled-off or they could sound warmer or there could be less sparkle/shimmer compared to an amp of higher bandwidth.
.
Almarg's comments to my post:
http://forum.audiogon.com/cgi-bin/fr.pl?aamps&1423113432&openflup&50&4#50
It's my understanding that class D power conversion modules typically use high levels of global feedback. The input signal is constantly being compared to the output signal (thousands of times per second) prior to the output signal being amplified. The output signal is only sent to be amplified once it matches the input signal and any necessary adjustments have been made. This high level of global feedback, which designers of more traditional amp topologies typically try to avoid at all costs, is a mainstay of class D amp design. My theory is that this high reliance on high levels of global feedback is responsible for class D amps' astounding neutrality and their performing so close to the ideal of 'a straight wire with gain'. Also, my theory is that this design results in excellent frequency response at all volume levels.

As I've stated I have no technical training on amplifier design or electronics. The above is based on personal reading about class D amplification combined with any small amount of rational thought and common sense I may possess. I may have this entirely wrong, however, and would welcome comments and thoughts on my theory from those with more technical knowledge on class D amps than myself.
while it is true that class-D amps use feedback to make them work & while you are right in stating that without global negative feedback a class-D power would not work, I don't think it is correct to correlate the use of global negative feedback to a class-D power amp's sonic quality.
The way class-D architecture was invented/designed/formulated, global negative feedback is part of its entity. So, just because you read global negative feedback you shouldn't relate it in the same way as you would to GNFB in a class-A, AB power amp. The class-D architecture needs GNFB while class-A, AB architectures have topologies that can do with little or no GNFB.
The class-D power amp is a continuous-time, discrete-voltage pulse width modulated system. The 1st gen of class-D power amps used (& still use) analog/linear power supplies (like the type you see in class-A, AB amps). And, now I'm observing that the next gen of class-D power amps are using switch-mode power supplies (which are themselves class-D power supplies). I think the audio SMPS has finally developed to a point where it is has a low enough noise floor & can handle large currents in a compact size.

Sonic qualities of a class-D power amp have to do with
* power supply design
* noise attenuation at the final output (the amp binding posts) - correct choice of filter
* managing the switching noise in the power output transistors i.e. reducing the switching noise impact on the analog circuits that form the overall class-D power amp
* routing of noisy & quiet signals

just to name a few items.