Dear rhythmsandy,
To be honest, having minimal gain and not using much feedback sounds counter-intuitive to me. If I were you, I would use a lot of feedback, especially considering that it's a class-D amplifier that has several mechanisms contributing to pulse-width error. Besides, without much feedback, even power supply fluctuations would get through to the output, almost unopposed.
I would suggest adding enough feedback (with enough compensation) to get the best out of what you have. If you fear poor stability margins and instability, you may try self-oscillating designs such as UcD or the hysteresis control.
All the best.
To be honest, having minimal gain and not using much feedback sounds counter-intuitive to me. If I were you, I would use a lot of feedback, especially considering that it's a class-D amplifier that has several mechanisms contributing to pulse-width error. Besides, without much feedback, even power supply fluctuations would get through to the output, almost unopposed.
I would suggest adding enough feedback (with enough compensation) to get the best out of what you have. If you fear poor stability margins and instability, you may try self-oscillating designs such as UcD or the hysteresis control.
All the best.
You can parallel more FETs to an existing class-D amplifier and use a bigger or multiple output filters.
But if you want to buffer an analog amplifier, you just have to attenuate it to compensate for the class-D amp gain because the PWM modulation is done at low signal levels and not at the output level. Some class-D amplifiers may be unity gain stable and some may not but it is unlikely the input common mode range is anywhere close to rail-rail. So it's pointless to drive a class-D amp with an analog amp. Just use the class-D amp.
But if you want to buffer an analog amplifier, you just have to attenuate it to compensate for the class-D amp gain because the PWM modulation is done at low signal levels and not at the output level. Some class-D amplifiers may be unity gain stable and some may not but it is unlikely the input common mode range is anywhere close to rail-rail. So it's pointless to drive a class-D amp with an analog amp. Just use the class-D amp.
What if he just hates nfb sound?
Ah, yes, class D may then be the wrong amp for him
Just thinking why not reduce the feedback in the class D.
Also the waveforms shown in the ADH video are nonsense. This is not what Class D amplifier output looks like at all. It actually looks like a square wave, the amplitude is the total Rail-to-Rail voltage, with a varying duty cycle. After it is filtered, it just looks like a regular sinusoidal audio waveform. The basic operation of a Class D amplifier is the same as a switch mode power supply, only that the output is varied dependent on the input, rather than set to a fixed voltage.
Last edited:
It is difficult to say exactly what Devialet are doing but I suspect it is similar to the Yamaha hybrid amplifier where the power supply rails are switch-mode controlled so that the analog output devices operate with a maximum of about 4 Volts drop and therefore very little heat is wasted. In any case, the class-D outputs are not tacked on the output of an analog amplifier, but perhaps the other way around so that the analog amplifier dictates the final result.
Their video talks about the Class-D and Class-A amplifiers running in parallel, but who knows if that's how it actually works. They claim to have it patented, so one could always look up the patent for a clearer description.
Either way, I don't think it's the kind of thing most hobbyists could knock up from scratch.