Handling Heavy Amps

 The one area where I think they differ the most with the Pass amps I've had is I think their high fequencies are a little more recessed,with less sparkle. 

@ddafoe The class D goes beyond human hearing so that's not a frequency response thing. Its more likely the difference in the distortion signature of the two amps since distortion is interpreted by the ear as tonality.

@atmasphere

Technical papers that discuss Class D topology suggest that the distortion resulting from the topology’s dead-time constraints is similar in effect to what we used to call TIM or SID distortion in high-feedback, low slew-rate designs.

IOW, pre-GaN-FET Class D artifacts sound to the ear like intermodulation distortion when reproducing rapid rise times -- e.g., transients. It’s certainly possible for a listener to interpret this as brightness or some other relatively simple frequency-response anomaly, but I personally hear the audible effects of Class D distortion more like the complex and nuanced effects created by high-frequency power-line noise.

I realize that I’m hand-waving a bit, but I’m trying to keep this message concise without dumbing-down.

And as for me, I’m negative on pre-Gan D topology. I have 4 Class D amps in service right now and am actively trying to replace them all. Or at least relegate them to home-theater surround channels.  I realize that there are some well-reviewed D-topology components out there, but I haven't yet heard them.

Technical papers that discuss Class D topology suggest that the distortion resulting from the topology’s dead-time constraints is similar in effect to what we used to call TIM or SID distortion in high-feedback, low slew-rate designs.

@cundare2 Those technical papers don't describe everything about class D amps since what you're talking about doesn't seem to apply to our amp. The deadtime in our amps causes lower ordered harmonics instead.

No issues of TIM since the input of the amp is never out of control by the feedback (which is what caused TIM in amps with that problem decades ago). We would not have released the class D if we didn't feel that it kept up with our class A triode OTLs.

@atmasphere

Having never heard your products, I could not with any credibility comment on your comparison to Class A tube OTLs. But please be aware that I’m not trying to criticize your amplifiers. I’m talking about the way that we perceive the sonic artifacts produced by most Class D amplifiers. Please reread.

I tried to be careful with my wording, stating that "traditional" Class D topologies produce distortion that is similar in effect to Aczel-period SID/TIM. But as I tried to convey, this type of distortion is caused by a mechanism distinct from the high-feedback/lower slew-rate TIM that made news back in the 90s, just as is the audible-range intermodulation distortion caused by supersonic, low-amplitude line noise. Class D and line-noise-induced IM have nothing to do with feedback. The sonic effect of all three, however (at least to my ears) is similar. And when you do a deep dive into the technology, that makes sense. That was my point.

I also understand that the much shorter dead time of GaN FET output stages can greatly mitigate this flavor of TIM distortion. Given that "Class D" is an umbrella term that desribes only one characteristic of a complete amplifier design, there have always been at least a few D amplifiers that sounded pretty good, regardless of transistor substrate. Even more so since the increased popularity of GaN FET output stages.

So when you state that your Class D amps don’t produce the types of artifacts that have long compromised D designs, I can only invite you to loan me a unit so that I can hear for myself. Barring that, please try to understand my comments in context.

By the way, I’m not sure what you’re trying to convey when you say that "the deadtime in [your amps] causes ’lower ordered harmonics’ ". One could argue that nasty third-order, and even fifth-order, harmonics are "lower." I suspect you meant that, like a tube amp, it produces even-ordered harmonics.

Hey, gotta be precise when you’re talking to lawyers and engineers! 😉

Anyway, good luck with what I’m sure is a terrific product line.  (It certainly has a good reputation.)  I'm currently looking for a sub-40-pound integrated in the $7-15K range, so if you have any suggestions, I'd certainly take a look.

@cundare2 If you don't mind some correction, the 3rd harmonic isn't 'nasty'. Musically its an octave and a 5th above the fundamental tone. Its well-known that it is innocuous such that this fact is often only mentioned in passing in engineering tomes like this one (see page 10).

The primary distortion that a properly working reel to reel (or any other analog tape recorder) makes is the 3rd harmonic. I don't hear people complaining about reel to reel sound, can we put this one to bed?

The 3rd harmonic is present in any amplifier. It is useful in tube amps for masking higher ordered harmonics and is a big reason for tube amps sounding smooth as a result. It tends to be more suppressed in solid state designs so the high orders in solid state tend to be unmasked, especially if distortion rises with frequency, which is quite common.

The period of amplifiers with TIM issues was about 10-15 years earlier than you suggest. Marshall Leach designed his rather famous amplifier in the 1970s to get around this problem. However its helpful to understand that TIM by itself isn't a thing- tunes out is was just distortion caused by how the feedback is received in the amplifier design. Its caused by the input differential voltage amplifier having to drive a high capacitance, and you can get situations where the feedback is unable to control the voltage amplifier, resulting in distortion. The problem was solved by degenerating the gain of the input differential amplifier, thus reducing the compensation capacitance it had to drive.

We use a lot of feedback in our class D (as its a self-oscillating design) but the feedback is able to control the amplifier properly with any signal right up to clipping. So its immune to this problem.

IMO your 2nd paragraph in the post above doesn't make sense. You do have to be precise when talking to lawyers and engineers :) 

The 'supersonic, low-amplitude line noise' to which you refer really isn't a thing! When designing a class D amplifier, its not enough to build it to meet FCC subpart 15 or EU radiation directives; you have to make it considerably more quiet than that so noise doesn't interfere with digital audio devices connected to the amp. So quite literally the noise you mentioned doesn't exist in a successful design. In our case the class D is quieter than most tube amps, not just in the speaker but also radiated noise (if the tube amp uses a solid state power supply).  So this does mean less radiated noise than most solid state class A or AB amps.

FWIW, due to the fact that GaNFETs have no leads at all, they tend to have very low parasitic inductances, so its easier to control switching noise. This is important because GaNFETs can switch at some pretty high speeds; perhaps a couple of orders magnitude higher than the actual switching frequency of the amp itself. So you have to make sure everything is cleanly done, otherwise you may well have the outputs switching at 60MHz and heating up a bit as a result :)