Comparison: A new Class D amplifier for an old Class AB player

[paulraphael]

Unless you're looking at 2 amps with identical preamp sections and different power sections, it starts to become very subjective as to what the power section is actually doing to the sound.

Very, very subjective. The ideal way to do a test like this is blind, with someone switching the signal between the amps randomly. And with the amps level-matched to within 1/2 a db or so using a meter.

It's impossible to overstate the influences of bias (even unconscious bias) and of tiny differences in sound level. A difference of 1db sounds much more like a difference in sound quality than a difference in loudness.

These principles are demonstrated over and over and over again in controlled listening tests.

When good test conditions are met, it's generally impossible to tell the difference between power amps that are running in their linear range, unless they have a baked-in eq curve. The former matters only if you plan to overdrive the amp—and then it matters a lot. The latter always matters, if an amp isn't flat, but tells you nothing about amp topology; only about one particular manufacturer's design.

 

[Snaxster]

Very, very subjective.

Paul, is that like 'very unique'? My conclusions weren't very subjective; they were subjective, period. As I said in my first post,

I heard the following comparatively.

and as I said to Andy,

I spoke only of my personal experience, impressions, etc.; not of how things are, but of how they seemed to me.

Though I understand the temptation to ascribe to my test suppositions of... stuff, I don't know how else to say more clearly that what I conducted was purely a subjective assessment.

Anyone wanting to pursue the quantitative and/or conduct a professional usability study is welcome to. Go for it! But it ain't me. I only did my little thing and (as I regret increasingly) reported it here.

 

[Interceptor]

I appreciate the report.

As much as as it frustrates me as an Engineer, not everything can be measured. Good old "what the heck does this sound like" experimentation still reigns.

Passinwind - that block diagram of a PA power amp just scares me. That design means that the input stage really needs to be fed the right levels. That's OK in pro audio, where the 0 dB level = +4 dBm. Using a PA amp at any other input level either invites noise or clipping.

I'd rather use a transformer for balanced to unbalanced, then hit an attenuator, then the first gain stage. Ain't how it is done.

 

[agedhorse]

Passinwind - that block diagram of a PA power amp just scares me. That design means that the input stage really needs to be fed the right levels. That's OK in pro audio, where the 0 dB level = +4 dBm. Using a PA amp at any other input level either invites noise or clipping.

I'd rather use a transformer for balanced to unbalanced, then hit an attenuator, then the first gain stage. Ain't how it is done.

Wait a minute, you have misunderstood how the front end of the QSC amp actually works. The diff amp (called preamp) has a gain of 1 with +/-15 volts (or a little more) rails which means that it's overload is about 14V peak x.707 = 9.5V RMS which is +20 something dBu or 16dB of actual headroom. Since this input is usually driven by a device with about the same maximum output level, there's plenty of margin to work with.

Transformers with wide bandwidth at high levels and low distortion get pretty expensive. They are also a potential for noise coupling into the system unless very well shielded, including electrostatic shielding between primary and secondary.

 

[Interceptor]

I do understand all of that very well including having a good understanding of cost differences. Ever work in a CBS broadcast plant with +8 dBm test tone level? The part that saves the fixed gain first stage from being overly problematic is most MI gear can't put out enough level to get to +22 dBm anyways, plus, in FOH, there's enough compression and limiting to keep levels under control.

It still spooks me because the trade offs are not well understood even by many studio engineers.

 

[agedhorse]

You have to understand at least the very basics of gain structure or else you have very much bigger challenges.

 

[Interceptor]

I agree completely (and did from the onset.) Misunderstandings of gain structure are the root of so many interface issues.

Anyways, even with things all leveled perfectly, different stages do sound different. Otherwise, why is an API or Neve chosen in recording over a generic mic preamp? I've heard great and so so implementations of class D amplification.

Having a comparison as the OP did of a well known power amp against a well regarded head is certainly useful - it helps to dispel the belief that class D is intrinsically inferior.

 

[JimiLL]

Hi guys.
I've been doing a little comparison myself between a cheap Crown XLS 1500 bridged to 4 ohm
driving my two Aguilar GS212 and a pair of recently serviced Crown Macrotech 600 in bridge mode
each driving its own cabinet. So in principle we're talking almost same power i.e. (theoretical) 1x1500 watt into
4 ohm or two times 600 into 8 ohm. And just to make things clear, I don't play particulary loud, more a matter of
not driving your equipment too hard.
Very much to my own surprise the cheap far east produced Crown class D is the winner. Quiet, requires less output
from my Aguilar 659 pre-amp and seems to have endless power and dynamic headroom. Hardly any difference in sound,
meaning both set-up sounds right but it seems the old Macrotech has to be driven harder.
I'm a tube loving guy having played SVT's, Mesa 400, Hiwatt's etc. but being able to carry your pre/power amp in a 3 units
rackbag over your shoulder without comprimizing too much on your preferred sound is hard to beat.
Regards
Thomas from Denmark

Could be different input sensitivity

 

[seamonkey]

As much as as it frustrates me as an Engineer, not everything can be measured. Good old "what the heck does this sound like" experimentation still reigns.

In terms of audio, what things can't be measured?
Can those things be recorded?
I agree that real attempts at DB tests are still a great way to compare audio, but after the test I'm sure a comparative test of whats different in the audio is still possible.

I can't really see players can hear differences in "class" of an amp, and the actual measurements are identical.

 

[Snaxster]

I appreciate the report...

Hello, and thank you. It was my pleasure to do.

Anyways, even with things all leveled perfectly, different stages do sound different. Otherwise, why is an API or Neve chosen in recording over a generic mic preamp?

Speaking of which, I just posted this thread:

Early impressions: Manley TNT 2-channel mic preamp used with bass guitar

 

[paulraphael]

In terms of audio, what things can't be measured?
Can those things be recorded?

Everything can be measured. The only real remaining voodoo lies in the field of human psychoacoustics, which deals with how we odd, upright-walking apes actually perceive sound waves. It's a somewhat young science, and there are new discoveries all the time. But it's starting to seem like the big ones have been made.

We're pretty good now at predicting if, in a blind test, anyone will be able to hear a difference between Thing A and Thing B. In fact, almost all the measuring instruments are more sensitive than human ears; we can measure all kinds of stuff that no one hears. But when there IS a significant measured difference between Thing A and Thing B, we don't have magic formulas to always know HOW the difference will sound, or which will sound better to you or to me.