Should Class D amps be fed analogue or digital?

[dshreter]

That is just the controller chip, but yes, there appears to be a fully digital PWM modulator as part of the block diagram.

That is what is inside of a Sonos Amp.

It seems like the pwm signal can be quantized in fixed height and quantized possibilities for width similar to bitstream digital audio transmission or in a typical class D amp it seems that signal a truly analogue representation.

The D in class D doesn't mean digital. End of story

Thank you, this has been mentioned a few times. I do understand that but it also seems there are topologies that operate digitally. If the pwm signal is quantized, then it the same type of signal as bitstream digital audio. If a generalized digital signal can be converted to bitstream losslessly, then I would imagine there is potential for digital native class D amps. And perhaps there are good examples that already exist if in fact Sonos Amp is of this type.

In speakers like DSP versions of Genelecs, I’m curious how the amplifiers operate in those especially.

 

[Frank Dernie]

That is what is inside of a Sonos Amp.

It seems like the pwm signal can be quantized in fixed height and quantized possibilities for width similar to bitstream digital audio transmission or in a typical class D amp it seems that signal a truly analogue representation.




Thank you, this has been mentioned a few times. I do understand that but it also seems there are topologies that operate digitally. If the pwm signal is quantized, then it the same type of signal as bitstream digital audio. If a generalized digital signal can be converted to bitstream losslessly, then I would imagine there is potential for digital native class D amps. And perhaps there are good examples that already exist if in fact Sonos Amp is of this type.

In speakers like DSP versions of Genelecs, I’m curious how the amplifiers operate in those especially.

I seem to remember a very expensive amplifier in the past (over 10 years ago) that did take a digital signal, convert to PWM and just have the output filter. Lyngdorf were involved with another engineer but they fell out, and the name changed and it all petered out, it became the Lyngdorf Millennium IIRC. I wanted one at the time but mainly since I loved the styling.

 

[dshreter]

I seem to remember a very expensive amplifier in the past (over 10 years ago) that did take a digital signal, convert to PWM and just have the output filter. Lyngdorf were involved with another engineer but they fell out, and the name changed and it all petered out, it became the Lyngdorf Millennium IIRC. I wanted one at the time but mainly since I loved the styling.

This might be the one you’re thinking of:
http://www.soundstagenetwork.com/revequip/robert01.htm

 

[HammerSandwich]

I’m having trouble understanding the merits of when d/a conversion should take place...

See the All Amplifiers Are Analogue... PDF by @Bruno Putzeys. TLDR: You cannot directly compare an analog output to a digital input to perform error correction, so all topologies become equivalent in EE analysis.

In my head, the other approach translates to "Yes! Let's build a reference-quality DAC with 1000x the normal output levels. How hard could it be?"

 

[HammerSandwich]

This might be the one you’re thinking of:
http://www.soundstagenetwork.com/revequip/robert01.htm

Or the Sharp reviewed by Fremer (!?!!)?

 

[EB1000]

The majority of class-d amps are natively analog. They are fed by an analog signal, which gets modulated, usually by analog PWM or selfoscillating modulation. However, it is possible to implement PWM in the digital domain. And some class-d amps do convert the audio signal into digital.

 

[UliBru]

Thoughts about true digital amplifiers can be found at https://www.ti.com/lit/an/slaa117a/slaa117a.pdf
Basically TI has used inventions by Lars Risbo (development at Harman Consumer Europe, CTO Toccata Technology, CTO Texas Instruments, Co Founder Purifi Audio)

 

[Frank Dernie]

This might be the one you’re thinking of:
http://www.soundstagenetwork.com/revequip/robert01.htm

Yes thats it, started life as the TACT Millennium then became Lyndorf Millennium when the partners split. more than 20 years ago!

 

[DonH56]

ASR class D description/introduction: https://www.audiosciencereview.com/forum/index.php?threads/class-d-amplifiers-101.7355/

 

[Matias]

There are some amplifiers on the market that convert digital input to PWM and straight into the amplifier. For example:
https://nadelectronics.com/product/m32-directdigital-amplifier/

But Bruno Putzeys, former Hypex and currently at Purifi, has a white paper arguing against this, that even though it sounds logical and natural, there are shortcomings compared to a DAC stage converting to analog first and the power amplifier using the analog input.
https://www.hypex.nl/img/upload/doc/an_wp/WP_All_amps_are_analogue.pdf

Edit: fixed first link.

 

[UliBru]

But Bruno Putzeys, former Hypex and currently at Purifi...

is now working together with Lars Risbo and Peter Lyngdorf, both arguing for true digital amps.
Probably they develop now together the perfect solution?

 

[Matias]

is now working together with Lars Risbo and Peter Lyngdorf, both arguing for true digital amps.
Probably they develop now together the perfect solution?

So far they released only an analog input class D module, but who knows what else they're cooking!

 

[dshreter]

With my small brain understanding of Class D amplification, they can operate with a digital input that goes straight into the amplification circuit without first requiring digital to analogue conversion. And that when provided with an analogue input, they do what could be described as analogue to digital conversion first before amplification. In other words, Class D natively works with a digital input.

Revisiting my original post, it seems Class D amps are not fundamentally digital at the input stage. Instead they can be either digital or analog native, and there are differing schools of thought as to which is superior.

I would imagine both classes could be built audibly transparent, so which is better maybe more a matter of cost and convenience vs performance considerations. Accepting an analog input provides more flexibility, while digital enables easier integration of DSP.
I’m also curious the relative costs of a high quality analog vs digital pwm generation.

 

[0bs3rv3r]

they will convert analogue to digital then back to analogue if the input is fed analogue signal

That would certainly be a big downside. I suspect the market for a digital DSP based PWM modulation system will be a digital-only amplifier - i.e. no analogue inputs. No-one would want to do that to an analogue signal.

 

[Doodski]

I'm interested to see what our Class D expert @March Audio thinks about this thread.

 

[Frank Dernie]

No-one would want to do that to an analogue signal.

Happens all the time in most LP manufacture and a lot of mixing engineers who have favourite (or fashionable) legacy analogue effects boxes they wish to apply to some of the signals in their assembly of a performance..

 

[DonH56]

Revisiting my original post, it seems Class D amps are not fundamentally digital at the input stage. Instead they can be either digital or analog native, and there are differing schools of thought as to which is superior.

I would imagine both classes could be built audibly transparent, so which is better maybe more a matter of cost and convenience vs performance considerations. Accepting an analog input provides more flexibility, while digital enables easier integration of DSP.
I’m also curious the relative costs of a high quality analog vs digital pwm generation.

They are not fundamentally digital, period. Please look at the link to the brief introduction here on ASR to see how a basic class D amplifier works. The analog input signal modulates the output pulse width (PWM). The output is not a simple bit stream like a "digital" amplifier would produce and the input is not digital; the input analog signal is multiplied by another analog signal (e.g. triangle wave) to module the output pulse width.

Link: https://www.audiosciencereview.com/forum/index.php?threads/class-d-amplifiers-101.7355/ -- look at the pictures showing the signals at various points.

The few amplifiers I have seen claiming to be "all digital" are effectively delta-sigma DACs with a high-power output stage. Those are not class D amplifiers.

HTH - Don

 

[Matias]

I guess it is just naming: class D is analog input by definition, PWM amp using digital input is named something else. That's it.

 

[UliBru]

I guess it is just naming: class D is analog input by definition, PWM amp using digital input is named something else. That's it.

Since several years you can find the term "true digital amplifier".

 

[DonH56]

Class D amplifiers have analog inputs and outputs... The PWM output stream of a class D amplifier is not a true digital signal even before filtering.