To Clock or not to Clock! - diyAudio
Go Back   Home > Forums > Amplifiers > Class D

Class D Switching Power Amplifiers and Power D/A conversion

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 28th December 2008, 10:58 AM   #1
Baldin is offline Baldin  Denmark
diyAudio Member
 
Baldin's Avatar
 
Join Date: Jan 2005
Location: Copenhagen
Send a message via MSN to Baldin
Default To Clock or not to Clock!

Hope you all had a very good Christmas


Been thinking a bit duing the hollydays ... I have now tryed some different designs using self oscillation, both UcD and pre-filter designs.
I think it has been bificult to make work and I have all the time had some problems with beeting when using more than one amp. I know that it is some what a matter of proper PCB layout and taming good care of EMI, but it is still an inherent problem with selfoscillating designs.

Now I think I'll try some clocked designas using a good x-tal clock to syncronice the amps.

Then I'm wondering whether the self oscillating designs in theory should have better distortion performance???
Any thoughts??

I actually cant see why that should be. It of cource is a matter of feedback around the output stage in the clocked design, which is inherent in the self osc. designs.


Another thought is about the output filter. If good coils (right ferite material) and good caps are used I can't see these contribute much to the distortion. In a clocked design, I don't think it is likely to get as much feedback from after the filter, and therefore it will probably not be possible to reduce the effect as much as with a self osc. design.

Maybe I'm all wrong here, so please comment with your knowledge.

Best Christmas regards Baldin
__________________
Visit www.sensibleaudio.dk or read latest news on Baldin's Blog
  Reply With Quote
Old 29th December 2008, 04:29 PM   #2
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
...yes, I also think that in clocked designs it is more difficult to realize strong post filter feedback and resulting from this the filter design has to be done with more love.
It is not just the right ferrite. IMHO the ferrite itself is less critical than correct winding. In a ferrite design you will need an airgap, which is strongly linearizing the magnetic path as long as you do not go for very high flux. But you must take care that your windings do not suffer much from the leaking fields. These fields and the resulting eddy currents in the wires do not just cause heat, but also distorsions, because the losses are strongly non linear increasing with the signal magnitude.
Furtheron you should keep the winding capacity low, in order to avoid the undesired 'blibs'.
Unfortunately an optimized design, which is taking into account the points above is leading to larger sized chokes than the popular metal powder torroids of many class D filter designs. BTW: Even in UcD Bruno is using a gapped ferrite design.

Another very important point regarding distorsions and especially regarding audible distorsion is the dead time adjustment. It is really worth to spend R&D efforts in this point.

Last but not least clocked designs have a very low 'natural PSRR'...

In the meantime the battle of self oscilating vs clocked appears to me a little bit like the old discussion about BJTs vs MosFets in class AB designs. If done properly both can give good results and there seems to be also some portion of personal taste in this decision.
In the moment I am slightly more on the side of clocked designs, rather than self oscilating (My Christmas taste 2008).
  Reply With Quote
Old 29th December 2008, 04:32 PM   #3
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
Quote:
Originally posted by ChocoHolic
...
Another very important point regarding distorsions and especially regarding audible distorsion is the dead time adjustment. It is really worth to spend R&D efforts in this point....
I forgot to mention that this is a key topic not just in clocked designs but in self oscillating desgins as well.
  Reply With Quote
Old 29th December 2008, 06:56 PM   #4
Baldin is offline Baldin  Denmark
diyAudio Member
 
Baldin's Avatar
 
Join Date: Jan 2005
Location: Copenhagen
Send a message via MSN to Baldin
Greetings ChocoHolilc

... very god points.
I guess I actually like self osc the best, as it has lots of advantages and more or less only one drawback i.e. the beating problem.

... but is it realy true that the clocked design will inherently have a lover PSRR
Given that you could have the same amount of feedback around the output stage?


Talking about ferites, have you essn the new ML design ... some kinf of 4 level BD design, using an enormous amount of components (as always) and having some huge toroids in the output. .... se pictures

http://www.marklevinson.com/image_li...de-open_lo.jpg

http://www.marklevinson.com/image_li...dulator_lo.jpg

I myself have until now only used toroids, but will try some air gaped ferrits for a (hopefuly) coming design. ...... has been occupied on building myself a home theather / listening room ...... talk about sound improvements

Baldin
__________________
Visit www.sensibleaudio.dk or read latest news on Baldin's Blog
  Reply With Quote
Old 29th December 2008, 08:58 PM   #5
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
The ML modulator is looking pretty simple and straight forward...
Exactly my taste.


... and of course fully in the range of my current know how level about digitals & modulation...


Output filter:
Hm, where from do you get the information that the torroids are ferrite?

PSRR:
...well, your point with same amount of feedback around the output stage might be right. And in fact I am not sure how to model both systems in a directly comparable way.
My point is: In self oscillating designs the duty cycle is instantly moving very fast to the desired direction, if you reduce one or the other rail. In clocked designs this is not happening instantly, but is typically handled via an integrating gain stage, which most often is settled with very slow values - resulting in poor PSRR at higher frequencies.
On the other hand, I can imagine that optimized PID loops, no matter if done analogue or digital, allow similar good results like self oscilating designs. (That's just a guess, I never analysed this in detail).
  Reply With Quote
Old 29th December 2008, 09:29 PM   #6
Eva is offline Eva  Spain
diyAudio Member
 
Eva's Avatar
 
Join Date: Oct 2003
Location: Near the sea
Send a message via MSN to Eva
In my experience, audible crosstalk when the amplifier is idle is not that hard to avoid without synchronization, which only hides crosstalk rather than avoiding it.

The ML design is not looking particularly good to me. Where is the ground plane? Where are the output capacitors? It looks like former analog designers attempting high speed power digital...

BTW: I'm about to assemble the first prototype of a 3000W @ 4 ohms self oscillating amplifier. If my customer gives permission (unlikely) I could show some pictures. Anyway, it looks somewhat like the pink noise amplifier.
__________________
I use to feel like the small child in The Emperor's New Clothes tale
  Reply With Quote
Old 29th December 2008, 10:06 PM   #7
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
Quote:
Originally posted by Eva
...Where is the ground plane? Where are the output capacitors? ...
I would expect both on the bottom side of the PCB.
At least for me such a structure would make sense.
I would neither dare to rate nor just from these few pics.
  Reply With Quote
Old 30th December 2008, 04:56 PM   #8
Baldin is offline Baldin  Denmark
diyAudio Member
 
Baldin's Avatar
 
Join Date: Jan 2005
Location: Copenhagen
Send a message via MSN to Baldin
ML is like overdoing it a bit I think, and not very beautiful in it's simplicity, like UcD.
But as you ChocoHolic, I would not judge it before hearing it. On the other hand I would guess it is actually very good Ö.. ML simply donít make bad products!

Yes you could be right about the output inductors not being ferries, it could very well be a type of plastic former, making it an air coil in essence.

Eva and ChocoHolic, how do you model your amps. See drawing; would you say itís a fair representation, of a self oscillating design including a 2. order output filter??

Where tau is the summed delay from comparator, deadtime, output stage delay, and switching frequency. This will give a large phase shift towards higher frequencies.

Oscillation frequency can then be found as with a normal feedback amplifier (180 dg, unity gain at the summing point).
/Baldin
Attached Images
File Type: jpg d-model.jpg (56.3 KB, 537 views)
__________________
Visit www.sensibleaudio.dk or read latest news on Baldin's Blog
  Reply With Quote
Old 30th December 2008, 07:43 PM   #9
Pafi is offline Pafi  Hungary
diyAudio Member
 
Join Date: May 2004
Location: Budapest
Quote:
would you say itís a fair representation, of a self oscillating design
I'm not they, but I wouldn't say. Natural sampling doesn't cause delay. Otherwise it's a usable approximation, but only at small signal levels and low freq. At high levels K become nonlinear, and the switching residuals start to go down to audio band.

BTW: fsampling=2*fswitching (at least as long as modulation index is small).
  Reply With Quote
Old 31st December 2008, 04:20 AM   #10
diyAudio Member
 
Join Date: Sep 2004
Location: Yahoo, USA
Quote:
Originally posted by Pafi

Natural sampling doesn't cause delay.
Natural sampling introduces just enough delay to add 180 degrees of phase shift at a frequency of half the sample rate. Think about it; if a small signal ideal step suddenly appears at a switching amplifier's input, any output response cannot occur at least until the next natural sampling point.

Regards - a.s.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
My CD Clock - the "Kwok Clock" kwokm Digital Source 13 10th September 2006 12:34 PM
CS8412, PMD100, PCM63, onboard clock + [word] clock output implementation stolbovoy Digital Source 7 3rd December 2004 07:18 PM
Audiocom clock vs. LC Audio clock KevinLee Digital Source 22 26th July 2004 01:29 PM
Kwak clock versus Tent clock rbroer Digital Source 22 21st September 2003 01:20 PM


New To Site? Need Help?

All times are GMT. The time now is 10:11 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2