Class D blokschema

Jump to Latest
J

jhmkoop

Member
Joined 2007
#1
I am not a real DIY-er but for financial and audiophile reasons I have put together a Hypex Ncore amplifier. Because I wanted to know how everything works I also read the publications of Hypex about this kind of class D amplifiers. What I understood about it I have recorded in a block diagram. See: https://www.dropbox.com/scl/fi/5gxy...ey=ljwb6u7x08l7n1ou0wnghtl59&st=4th8fvyt&dl=0
My question to this community: is that somewhat correct? Feedback is welcome.

What also surprised me is that apparently there are no class D amplifiers with a direct digital input, they are all analog amplifiers with possibly an A/D converter in front of them. In my opinion, such a class D amplifier with a direct digital input would be possible according to block diagram 2. See: https://www.dropbox.com/scl/fi/1ibp...ey=vufbmcdbf78zp2xd0oco6v5cs&st=roewukiz&dl=0 Again, feedback is welcome, even mocking laughter (with an explanation why) is very welcome
 
Last edited:
#2
This is an English language forum ...

I am not a real DIY-er but for financial and audiophile reasons I have put together a Hypex Ncore amplifier. Because I wanted to know how everything works I also read the publications of Hypex about this kind of class D amplifiers. What I understood about it I have recorded in a block diagram. See: https://www.dropbox.com/scl/fi/5gxy...ey=ljwb6u7x08l7n1ou0wnghtl59&st=zethgzzx&dl=0
My question to this community: is that somewhat correct? Feedback is welcome.

What also surprised me is that apparently there are no class D amplifiers with a direct digital input, they are all analog amplifiers with possibly an A/D converter in front of them. In my opinion, such a class D amplifier with a direct digital input would be possible according to block diagram 2. See: https://www.dropbox.com/scl/fi/1ibp...hart.pdf?rlkey=vufbmcdbf78zp2xd0oco6v5cs&dl=0 . Again, feedback is welcome, even mocking laughter (with an explanation why) is very welcome

Jan
 
#5
jhmkoop said:
" "I edited and checked the links in my post, they should be working now.
Click to expand...
I do not want to be negative but this is about the worst way to share documents in this forum. In order to be able to properly view the document I had to rotate it. And to do that I had to sign in into Dropbox. I used my Google account and I have no idea if I already had an account, if I signed up, or I signed in. This is about as bad as sending people to facebook to see information. Which you cannot without signing up. I don't have Dropbox, I don't want Dropbox and still I had to use it to see your documents.

This forum fortunately has an option to directly attach documents to posts. Apologies for my rant, it is not directed personally to you but to everyone posting links to sites where you have to sign up or identify yourself.

I wanted to join the discussion but not join any data collecting evil big tech company.
 
  • Like
Reactions: Ixnay and jan.didden
#6
jhmkoop said:
Again, feedback is welcome, even mocking laughter (with an explanation why) is very welcome
Click to expand...
It is not a stupid idea and I am wondering why it is not discussed more often. In the 1980-ies when digital audio was beginning to emerge, it was a discussion topic among fellow audio interested students how far you could keep the audio chain digital. As in why not build a power DAC and put an passive filter in front of the speaker. In those days class-D amplifiers did not exist yet. Not commercially.

As for your idea, you use a feedback loop to feed back the analog signal, convert it to PWM and compare it with the original PWM. Is that necessary?

If you have a classic DAC (not that 1-bit nonsense etc) you simply feed the digital value to the DAC and you get an analog signal. The signal is low-pass filtered and not further processed. There is no feedback to correct the analog signal in any way. The steps are exactly 1/65365 of Uref and that is it. Oversampling in the time domain is exclusively to simplify the low pass filtering. It is easier to filter 176 kHz if your wanted band is up to 20 kHz.

Now if you convert a 192 kHz signal directly to a PWM signal of several hundreds of kHz using a DSP, is feedback still necessary? It is a honest question, I don't know. Distortion in a class D amplifier is caused by non-ideal conversion of analog to PWM. The analog-to-sawtooth comparator being used to generate the PWM and all that, I don't see why a PWM signal calculated from digital is distorted.

Sure the PWM signal is quantized in time. That is, the resolution of the width is 1/2^n where n is the number of bits used during conversion. Sure the number of pulses is not infinite in a time interval. But I don't know what the effects are on the analog signal after it has been low-pass filtered by the output filter.

It might be counter-intuitive. Just like the quantized voltage steps in a DAC output contribute to noise, but not to distortion.

For a modern DSP it should be a piece of cake to convert a digital signal to a distortion-less PWM. And even if I am wrong and the hardware is not powerful enough (not even NVIDIA?) at least it should be possible to calculate it theoretically or simulate it.

So my question is: If a digital signal is converted to PWM is there any distortion induced? If so, which, how and why? If not, why do digital-to-PWM converters not exist.
 
#7
jlinkels said:
So my question is: If a digital signal is converted to PWM is there any distortion induced? If so, which, how and why?
Click to expand...
So it is not a problem to make PWM without distortion, the problem is to amplify this PWM to the required power without distortion.

I have been working with power PWM DACs for a long time, and based on it I made a DSP Amp, and in order to get an acceptable level of distortion at the amplifier output, I had to use feedback, otherwise the distortion level can hardly be considered acceptable.
This thread contains measurements of what I was able to squeeze out of this direction.

https://www.diyaudio.com/community/threads/3-way-dsp-amp.415065/#post-7734938
 
#9
uriy-ch said:
So it is not a problem to make PWM without distortion, the problem is to amplify this PWM to the required power without distortion.

I have been working with power PWM DACs for a long time, and based on it I made a DSP Amp, and in order to get an acceptable level of distortion at the amplifier output, I had to use feedback, otherwise the distortion level can hardly be considered acceptable.
This thread contains measurements of what I was able to squeeze out of this direction.

https://www.diyaudio.com/community/threads/3-way-dsp-amp.415065/#post-7734938
Click to expand...
That is quite impressive.

If I understand correctly. You built a power DAC using the TAS5352. The TAS5352 itself does not feed back the analog signal. (At least I don't see a feedback connection in the TAS5352 AN) But you take the analog signal after the low-pas filter, convert it to digital and correct the distortion in your DSP?
 
#13
jhmkoop said:
What also surprised me is that apparently there are no class D amplifiers with a direct digital input, they are all analog amplifiers with possibly an A/D converter in front of them.
Click to expand...
These exist for years and they likely will be the standard in the near future. They are called FDA (Full Digital Amplifier) or PowerDACs. There are numerous attractive benefits but there is also the feedback dilemma. Bummer. The concept was far too modern to many (audio is grey) and it also suffered from the then common class D antipathy that of course slowly has dissolved.

The large class D amplifier IC manufacturers often produce a version of IC's with digital inputs. Also one of the manufacturers discontinued all the digital input ICs so maybe the time is not right yet.

In ready made stuff the Denon PMA-60 particularly made a good impression as FDA. In the past I tried Nuforce DDA-100 and DDA-120 and these were also quite good.
 
Last edited:
You must log in or register to reply here.
Top Bottom