Why has no one thought of this before?
The entire internet is filled with schematics of some kind of junk.
But here, the maximum result is achieved simply and elegantly.
What's the catch?
Last edited:
Composite amps are nothing new and there's often a performance boost. Look at the incredible amps from Tom Christiansen (Neurochrome) who may know more about this than just about anybody. @tomchr
Yep. Using a composite amp to increase performance is an old trick. Possibly as old as the opamp itself. I've spent the past 11 years developing composite amplifiers with the LM3886. Have a look here for example: https://neurochrome.com/products/modulus-86-10th-anniversary-edition
As you've undoubtedly discovered, stability in a composite amp is a challenge. But it's a challenge that can be overcome with good engineering. Another catch that's specific to the LM3886 is that the LM3886 has a few 'worts' that you need to work around. For example, it has a bit of quasi-oscillation as it exits clipping, especially from the positive rail. This gets worse with a heavier load (lower load impedance). The composite loop needs to be designed to handle this, which means the compensation needs to be a lot more conservative than simulations would indicate.
If you end up building this amp, I recommend that you test it thoroughly before putting it into service. As a minimum, I would look at how the amplifier behaves as it approaches clipping (both soft and hard clipping) with both a sine wave and a square wave. I'd also look at how the amplifier behaves once the output stage slew-rate limits. So basically, test the amp thoroughly with a square wave.
I'd of course also measure the frequency response, distortion, etc. but those are sorta mundane in comparison to getting the amp to perform well on the transient response.
Yeah, but all circuits have weak points. In your first post you basically dismissed all other circuits as junk. I'm sure we can find a weak spot in your circuit too. Does that mean that it is junk?
Tom
It seems relatively simple and elegant solution, especially the dual feedback loop. But I think, the composite architecture isn't enough alone for the better high-frequency behaviour (if it necessary): the LM 3886 is the serious limiting factor at 10x amplification (near insufficient phase margin). The stability at simulator isn't guarantee the good sound at real life. Your enthusiasm is sympathetic for me, but your words are a little bit harsh.
Thanks for the advice.
When I see a circuit with two amplifiers — one having 80 dB gain at 20 kHz, the other 60 dB — and the total feedback depth is around 90 dB, with a phase margin of 20 degrees at the unity-gain frequency of 20 MHz, I consider that circuit pointless.
I’m referring specifically to simulation files.
did this to make the diode limiter operate properly with a floating power supply. It also reduces the loop gain crossover frequency.
Distortion looks high compared... are you sure you looked everywhere?
Why come here bragging about -86dB 3rd, 5th and 7th - that's awful.
//
