Galvanic isolation means that there is no "copper" connection.... so a balanced connection is in no way equal to galvanic isolation - nor is differential.
//
//
But perhaps a capacitor coupled differential feedback connection could do the job? Capacitor on both hot and cold - no ground - so only 2 wires between the two sections to carry the feedback signal - and no other galvanic connection.
//
//
... the choke being the inductor in the output filter? If so it won't work as before the choke the signal varies in voltage sving which the PWM does not.
//
That's a question of definition. I agree with You if galvanic isolation is defined at the interface between "boxes", otherwise a balanced or differential signal transfer can be isolated between sender and receiver. Safest is optical, but with it comes all the non-linearities of the optics, next is transformers which has their limitations(hysteresis, remanens and so on). A good solution should be based on what signal (freq, waveform, amplitude) You want to transfer and how high common-mode rejection ratio is needed. I have measured a fully functional RS-485-bus(max +/-15V) with over 100V common-mode signal.
Google says: "Galvanic isolation refers to a design technique that isolates electrical circuits, preventing direct current flow while still allowing signal and power transfer between them."
This could be on one and the same circuit board so close vicinity - there should not have to be very high common mode disturbances.
//
This could be on one and the same circuit board so close vicinity - there should not have to be very high common mode disturbances.
//
Before the choke, the signal is PWM and not a a filtered / reconstructed voltage as in post-filter feedback.
However, on second thoughts, I think it may not work well, as the PWM is also known to carry the inevitable voltage fluctuations from the DC bus that digital optos wouldn't pass on, affecting the PSRR of the amplifier, especially in half-bridge configurations.
It looks like TOSLINK is probably the way to go ...
You are of course correct, it's even in the picture in post #1 😎
//
... and passed on to the DAC via ground... hence, "the problem" 😉
//
Galvanic isolation is not an issue, but isolation without distortion is. A fast optocoupler with feedback has been used to isolate video signals and it might just work here, as well.
From pg 49 of:
https://user-web.icecube.wisc.edu/~kitamura/NK/Datasheets/misc/5988-4082EN designers guid.pdf
General:
https://www.vishay.com/docs/83708/appnote50.pdf
From pg 49 of:
https://user-web.icecube.wisc.edu/~kitamura/NK/Datasheets/misc/5988-4082EN designers guid.pdf
General:
https://www.vishay.com/docs/83708/appnote50.pdf
Here we need a feedback loop mechanism that is equal to or better than the quality of the incoming signal.... how is that made? Think about the performance of a top notch DAC... it will not be easy... but perhaps its overkill and other things matters more - like a stable ground ;-)
//
//
I'm not saying that it's trivial but there're a lot of high performance op-amps available these days that could be used to wrap up (and linearise) a fast opto-coupler. The forward path of the amplifier could also be isolated in the gate driver section.
Besides, isolation is not only about breaking the return path, but also about protection of downstream equipment, as Class-D amps can fail in very violent ways. When something like that happens you wouldn't want your expensive DAC to be blown away along with it !!
Besides, isolation is not only about breaking the return path, but also about protection of downstream equipment, as Class-D amps can fail in very violent ways. When something like that happens you wouldn't want your expensive DAC to be blown away along with it !!
I suppose you mean upstream 😉 anyways - you seem to not immediately dismiss the idea - thanks. Could you draw a block diagram of it? Maybe using the scheme in post #1?
I'll try but but I'm prepared to be corrected 🙂
//
I'll try but but I'm prepared to be corrected 🙂
//
The thread title mentioned 'isolation of the output stage', hence the above post.
Nevertheless, it is also possible to simply isolate the audio input using a lower bandwidth but much more accurate and matched pair dual opto-coupler. This way the amplifier and its feedback are not affected by the poles (and delay) of the opto-coupler which maybe important in self-oscillating designs like UcD etc.
However, many people would still prefer the >10MHz wider bandwidth optocouplers like HCPL4562.
HCNR200 AppNote
Nevertheless, it is also possible to simply isolate the audio input using a lower bandwidth but much more accurate and matched pair dual opto-coupler. This way the amplifier and its feedback are not affected by the poles (and delay) of the opto-coupler which maybe important in self-oscillating designs like UcD etc.
However, many people would still prefer the >10MHz wider bandwidth optocouplers like HCPL4562.
HCNR200 AppNote
Last edited:
Title is valid and reflects the core idea but I suppose what is really ultimately wanted is a stable ground to which a DAC is connected to.
Seems like a fine device. But I wonder if it suitable for high quality audio?
Seeing the specs above I doubt it. Can it be used in a way that it is significantly improved? Have you tried it in any audio application?
//
Seems like a fine device. But I wonder if it suitable for high quality audio?
Seeing the specs above I doubt it. Can it be used in a way that it is significantly improved? Have you tried it in any audio application?
//
Well, there maybe other devices that're good enough and we just need to look around better, as isolation, still is (largely) an industrial thing and has not (yet) properly bled into the consumer markets. Even the consumer S/PDIF circuits sometimes are not isolated like the professional ones.
Not this part, but I've used 6N137 to make an audio isolator a few years ago. The idea was successful but this was a switching isolator which is different from the analogue type that directly depends on the matching of CTR ratios between devices.
I even had some issues with adjusting the switching frequency which I had asked about in the following thread, where you could also see the basic circuitry that I'd used.
https://www.diyaudio.com/community/threads/analog-isolation-using-digital-optocoupler.324371/
It's an inexpensive and feasible solution for those who don't mind the switching modulation that goes with it.
So how about differential + caps on each leg? In both directions - from in to output...
Would that work at all?
@MarcelvdG might have an idea on this topic?
//
Would that work at all?
@MarcelvdG might have an idea on this topic?
//
So.... realise I can't do it... :-/
I feel it's not correct but I added it for discussion... at least the FB part can't be OK...
//
I feel it's not correct but I added it for discussion... at least the FB part can't be OK...
//
It is worth noting that this is across 10MHz while opamp noise voltages are usually specified across 20kHz or alternatively in nV/ √Hz at 1kHz. Besides, this part is video qualified.
There's no return path for the bias currents of the input opamp / comparator.
https://www.analog.com/en/resources/app-notes/an-937.html
Last edited: