• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

What kind of gain/power circuit on a class AB amp allows THD+n to look like this, and does it cause audible artifacts with variable amplitude signals?

mike7877

Addicted to Fun and Learning
Joined
Aug 5, 2021
Messages
692
Likes
138
So this is looking more and more common these days on AB amps, instead of a steady line which bottoms and then rises a bit before rising sharply with clipping.

1697308543236.png



My feeble mind sees this and thinks this must be variable gain of some kind, which allows for better THD+n at the lower end. At first I thought supply voltage might be fluctuated as well, but after further consideration, not at the size/weight/price point of these amps. Although it would increase efficiency, but I think equal efficiency to previous designs with better low end performance is probably good enough for the first steps in the minds of the designers (change one thing, observe, tweak, move on is the wise move, especially when there's little to gain from changing a lot at once)

OK, so... This real-time gain switching... it looks like it's somewhat gradual, happening whenever THD+n is getting worse after the first bottom (the rise of each little peak). This is happening with the gradual increase in amplitude of a sine wave.

Here's the thing, though... I don't know over how many seconds this increasing amplitude sine wave takes to reach clipping, but I assume its a couple or few seconds. However long it takes, this is nothing like real music. Just like a single tone, is nothing like real music.

What we need, at the very least, to test these types of amplifiers, is some dynamics. Not necessarily 1812 overture level (with the real cannons), but something like music before 1991 (or whenever it was that compression got turned to 11 and the peaks and troughs weren't more than 3db apart anymore - I'm exaggerating a bit on how bad it's gotten, but most music mixed since has only gotten more disgusting this way).

To keep things fair with the normal AB amps without this fancy feature, I don't suggest we change to measuring distortion on songs. BUT, if you were to cleverly apply side chain compression (with the right release and attack) to attenuate a 1khz wave 1:1 using a properly dynamic song (literally anything average from the 70s or 80s)...

THD+n could be compared and a graph generated using different points in the test for different amplitudes (like it is now, except now it's a constantly increasing amplitude sine wave instead of a variable sine wave)

I could be wrong in the premise (that this is variable gain), and so this whole thing could be pointless. But, if I'm not...

If you think this is a good idea, let me know.
If you think this is a good idea which needs a bit of tweaking (like I do, it's not perfect yet, and I don't know what the software is even capable of), post how you think it should be changed.
If you think there is absolutely no reason to test like this, explain why.

And if I am wrong in the premise - what is the cause of this previously rare / anomalous feature of THD+n charts of power amps (like the Topping LA90)? Would testing whatever it is in (roughly) the same way I described still be beneficial? If not, is there something else we should possibly be considering because of how these amps behave? (the relationship used to be linear, and now it's clearly not - very different)
 
OP
M

mike7877

Addicted to Fun and Learning
Joined
Aug 5, 2021
Messages
692
Likes
138
It’s a measurement artifact from the AP’s auto ranger. You only see it in amps with such low distortion / noise that the measurement is governed by the AP.

Michael

Shouldn't the software be subtracting (or adding) the error every time it ranges x amount so that the graph is an accurate representation
 
OP
M

mike7877

Addicted to Fun and Learning
Joined
Aug 5, 2021
Messages
692
Likes
138
Is the truth of the matter with these amplifiers, then, that the actual THD+n is the 'total auto-ranged range' less than what's actually plotted as the bottom? So for the Topping LA90 Discrete, which is 120-121dB into 5W 4 ohms, with an approximate auto-ranged range of 14dB and a plotted -114dB THD+n at peak power on the graph (of 56 watts into 2x 4 ohm loads), actually has a THD+n of -128dB at 56 watts into 2x 4 ohm load?
 
OP
M

mike7877

Addicted to Fun and Learning
Joined
Aug 5, 2021
Messages
692
Likes
138
Visually:
1697315554352.png

Is this right?
(except for not inverting those gains, and upon closer inspection, maybe lowering the second/last two portions by a total of maybe 2-3dB?)
 

DVDdoug

Major Contributor
Joined
May 27, 2021
Messages
3,023
Likes
3,977
You can't vary the gain of the amplifier because that would result in distortion. O, automatic changes in volume like al fouled-up automatic volume control, etc. Or compression or expansion...
 

staticV3

Master Contributor
Joined
Aug 29, 2019
Messages
7,913
Likes
12,671
Is the truth of the matter with these amplifiers, then, that the actual THD+n is the 'total auto-ranged range' less than what's actually plotted as the bottom?

Visually:
1697315554352.png

Is this right?
I believe that's right, though I don't have measurements to back this up.
 
OP
M

mike7877

Addicted to Fun and Learning
Joined
Aug 5, 2021
Messages
692
Likes
138
You can't vary the gain of the amplifier because that would result in distortion. O, automatic changes in volume like al fouled-up automatic volume control, etc. Or compression or expansion...

True, very true. Unless it was multi stage and there were complex interactions - a divider of sorts I believe would be required

Edit: the point was to find where the extra performance was coming from at the bottom end. Turns out, there is no extra performance at the bottom - it's at the top!
 
Top Bottom