• 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!

Class B amplifier with SINAD of 120, how is that possible?

John Atkinson at Stereophile has been doing that for decades with his test equipment.

Yes, because it has infinite bandwidth, a perfect square wave can’t be made, but apparently good test equipment comes really close.
 
10kHz square is a very good test signal. It may show hidden stability issues that are not clearly seen from measurements with strictly limited bandwidth.
Or show you things that don't occur in real life, having you chase a ghost.
 
A pure class B amp has an efficiency of about 78%. The higher the bias of the class A section of a class AB amp, the lower the efficiency becomes since the class A section never turns off.
 
Is there a schematic of this thing somewhere? Why do people feel so confident asserting alleged facts about this amp's implementation?
Not an expert by far. But it seems that has been difficulty identifying a pure class B historically. But may be off base here but have huge interest as an owner of a Nakamichi System One.
 
I think the lesson here is that the class of the amp doesn't tell you very much about its ultimate performance.

apparently good test equipment comes really close.

Hot, pedantic take here, but no analog signal comes "really" or even "kinda" close to infinite bandwidth in practice. ;)
 
Hot, pedantic take here, but no analog signal comes "really" or even "kinda" close to infinite bandwidth in practice.
Yes but 50kHz or 1MHz makes a big, big difference. And do not forget it is a test signal and you need a fast test signal to test caveats of the circuit design, especially in case of linear amplifiers. There is no excuse for not using the analog square wave generator.
 
Maybe post here. Much to be said and learned.
Good suggestion. Merged threads
 
It is a class B, no idle current of the output stage, except for leakage current. I built about 20 pcs of these amplifiers when I was a student and made some pizza lunch profit of them.
Whew, guessed right! ;)
 
I have always favored a 1 kHz and/or 10 kHz square-wave test, bandlimited to 50 or 100 kHz. The harmonics (odd only in an ideal square wave) fall as 1/N for the Nth harmonic so 1/3, 1/5, 1/7, 1/9.... the amplitude means the really high stuff is getting pretty low in amplitude. Five harmonics give a fairly good indication, though I tended to use 10 just for grins (THD measurements use 10 harmonics, so I figured it was good for a square wave test as well), thus a 10 kHz "square wave" limited to 100 kHz was sometimes interesting. For analog systems it was a reasonable, given turntable cartridges can have response to >100 kHz, and these days of higher sampling rate digital sources means to me it might be worthwhile. "Back then" I did two tests, one into a load resistor, and the second in the resistor in parallel with a large capacitor to emulate a worst-case speaker load.

squarewave_stack.JPG


I think @amirm did some tests early on, but for competent amps didn't show anything, so abandoned them as too time-consuming for the reward? Stability is pretty much a given these days.
 
NAD built those and they were plagued with thermal runaway.

And some of the "giant killer" Protons. Quite a few Korean OEM amplifier/receiver makers too.
 
Stability is pretty much a given these days.
I does give me pause personally, with spec-oneupsmanship among these companies compelling engineers to find ways to apply more and more feedback... especially with Topping's history, I'm not sure how thorough their testing processes are.
 
There is a microprocessor running and I *think* a relay that turns on the output. These are extras beyond the amp.

The internal picture shows a power relay, a speaker relay and a bunch of other input/gain switching relays. So, the 9W is reasonable including all the opamps, micro and relays along with amp operational current.
 
I have an interesting Cambridge Azur 840A v2 integrated here. Given to me for parts.

It too has a "patented" amplifier innovation and they call Class XD. This was Cambridge's first amplifier to incorporate this circuit. A Douglas Self design element I believe.

1726605862414.png



Patent appears to have lapsed, so I've attached some details for general interest. Including their white paper below (PDF)

1726604791524.png


Power amp stage schematic attached below. Notice the XD circuit can be switched on and off, like a VW defeat device. LOL.

1726605955835.png
 

Attachments

  • 840A_Class_XD_Amplifier_White_Paper.pdf
    764.4 KB · Views: 59
  • 840A power stage schematic ASR.pdf
    361.2 KB · Views: 44
Last edited:
Why do you think Topping chose class B instead of class AB?
More audio output power with less power consumption = higher efficiency and possible cost savings by drawing less current from the psu at all but maximum signal levels.

I see mention of some amazing DIY class B designs with vanishingly low distortion and high power. It would be really nice to see schematics, which should be possible since these are DIY designs, and presumably not commercial designs.

I wanted to correct a mistake I'd made previously. It seems the definition of class B operation does not necessarily include a crossover notch caused by the output devices 'sticking' in a zero-conduction state for any period of time. It is technically possible to get everything lined up with microscopic precision so that the push side of the circuit turns on the instant the pull side of the circuit cuts off.

1726608212137.png


But how is that achieved in real life, with devices that degrade over time, resistors that drift in value with heat, electrolytic capacitors that dry out with time and heat, etc? That's what would be interesting to know from Topping. My meager understanding is that it's much easier to get and keep a class AB output stage working well than a class B output stage.

Eh have anyone actually confirmed that it really is class B ? If it has bias current at all will it not by definition be class-AB
Apparently Topping reps were confused at first, calling it class AB but then coming back and correcting themselves, stating it's class B after all. Lots of confusion.

Here's how they say they did Class B a Nakamichi in the mid '70's.
That sales brochure says, "closer to class B than ever before." It doesn't say the output stage is out and out class B. It looks like very cold biased class AB, with at least a tiny amount of idle current drawn by both output devices at the waveform zero crossing.

It is a class B, no idle current of the output stage, except for leakage current. I built about 20 pcs of these amplifiers when I was a student and made some pizza lunch profit of them.
What is leakage current, please? Where does that appear in this circuit? Thanks.

AB amp will be simpler to build but will run hotter in idle.
By definition. Class AB keeps more current going through the output devices, even when no signal is present. The advantage is that with no region where both output devices are in cutoff (not conducting current) there's no crossover notch to fill in. I'd read somewhere (still trying to find where) that negative feedback is powerless to correct crossover notch distortion, but that might be one of those things you read that aren't actually true. Who knows...

One thing I have heard from several engineering types is that with negative feedback in an audio amplifier, it's best to use either very little or none of it (if your amplifier circuit is linear enough that way, very difficult to achieve in real life) or use a LOT of NFB. If you use just a little NFB, higher harmonics are generated in the output waveform, which is not desirable. However, as more NFB is applied, all harmonics are suppressed more, including those higher ones. Pretty soon those higher fundamentals are buried in the noise floor and the main concern becomes stability, freedom from ringing, overshoot or oscillation.

10kHz square is a very good test signal. It may show hidden stability issues that are not clearly seen from measurements with strictly limited bandwidth.
Yes indeed it is! I've asked Amir twice to measure this B100 amp with a 10kHz square wave at a moderate signal level. No reply to that. I wish he would do that. I wonder, do all these wonderful little amplifiers measured here have exemplary output of a 10kHz square wave? No ringing, no overshoot, no slanting, no blunting of the leading edge or the trailing edge, no rounding? Or do we assume that the 10kHz square wave response of all these amplifiers is essentially perfect, so why bother? (I hate assumptions...)

If it is in fact Class B it has the same inefficiency as Class AB when playing loud.
Yes, because a class AB amp goes into class B when playing loud. That's exactly why it's not class A.

10kHz square is a very good test signal. It may show hidden stability issues that are not clearly seen from measurements with strictly limited bandwidth.
Or show you things that don't occur in real life, having you chase a ghost.
Well, there's a controversial statement. I'm not going to argue but I think I'll just quietly disagree. The challenge with applying great gobs of negative feedback to a non-linear circuit (like a class B output stage) has always been stability, overshoot, ringing, oscillation. I'd really like to know how Topping did it, and others would too, it seems. Or did they? How's the 10kHz square wave look? ;)

Stability is pretty much a given these days.

Is it? Even in a design that uses more NFB than anyone has dared use before? The whole problem with using great gobs of NFB is stability, and a 10kHz square wave is a test of high frequency stability. It seems reasonably to me to check the stability of the latest, greatest wonder amps.

OK, that's all. Thanks for your input, everyone. This Topping B100 amp is one very interesting product. Maybe it is exactly what Topping says it is, maybe it isn't.
 
Last edited:
More audio output power with less power consumption = higher efficiency and possible cost savings by drawing less current from the psu at all but maximum signal levels.

I see mention of some amazing DIY class B designs with vanishingly low distortion and high power. It would be really nice to see schematics, which should be possible since these are DIY designs, and presumably not commercial designs.

I wanted to correct a mistake I'd made previously. It seems the definition of class B operation does not necessary include a crossover notch caused by the output devices 'sticking' in a zero-conduction state for any period of time. It is technically possible to get everything lined up with microscopic precision so that the push side of the circuit turns on the instant the pull side of the circuit cuts off.

View attachment 392902

But how is that achieved in real life, with devices that degrade over time, resistors that drift in value with heat, electrolytic capacitors that dry out with time and heat, etc? That's what would be interesting to know from Topping. My meager understanding is that it's much easier to get and keep a class AB output stage working well than a class B output stage.


Apparently Topping reps were confused at first, calling it class AB but then coming back and correcting themselves, stating it's class B after all. Lots of confusion.


That sales brochure says, "closer to class B than ever before." It doesn't say the output stage is out and out class B. It looks like very cold biased class AB, with at least a tiny amount of idle current drawn by both output devices at the waveform zero crossing.


What is leakage current, please? Where does that appear in this circuit? Thanks.


By definition. Class AB keeps more current going through the output devices, even when no signal is present. The advantage is that with no region where both output devices are in cutoff (not conducting current) there's no crossover notch to fill in. I'd read somewhere (still trying to find where) that negative feedback is powerless to correct crossover notch distortion, but that might be one of those things you read that aren't actually true. Who knows...

One thing I have heard from several engineering types is that with negative feedback in an audio amplifier, it's best to use either very little or none of it (if your amplifier circuit is linear enough that way, very difficult to achieve in real life) or use a LOT of NFB. If you use just a little NFB, higher harmonics are generated in the output waveform, which is not desirable. However, as more NFB is applied, all harmonics are suppressed more, including those higher ones. Pretty soon those higher fundamentals are buried in the noise floor and the main concern becomes stability, freedom from ringing, overshoot or oscillation.


Yes indeed it is! I've asked Amir twice to measure this B100 amp with a 10kHz square wave at a moderate signal level. No reply to that. I wish he would do that. I wonder, do all these wonderful little amplifiers measured here have exemplary output of a 10kHz square wave? No ringing, no overshoot, no slanting, no blunting of the leading edge or the trailing edge, no rounding? Or do we assume that the 10kHz square wave response of all these amplifiers is essentially perfect, so why bother? (I hate assumptions...)


Yes, because a class AB amp goes into class B when playing loud. That's exactly why it's not class A.


Well, there's a controversial statement. I'm not going to argue but I think I'll just quietly disagree. The challenge with applying great gobs of negative feedback to a non-linear circuit (like a class B output stage) has always been stability, overshoot, ringing, oscillation. I'd really like to know how Topping did it, and others would too, it seems. Or did they? How's the 10kHz square wave look? ;)

OK, that's all. Thanks for your input, everyone. This Topping B100 amp is one very interesting product. Maybe it is exactly what Topping says it is, maybe it isn't.
But has true Class B ever physically existed or just slightly overlapping AB, virtual Class B? Does an Class AB amp ever go into true Class B?
 
Back
Top Bottom