Multi stage output digital bi amp? - 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 8th October 2012, 06:13 PM   #1
Siggma is offline Siggma  United States
diyAudio Member
 
Siggma's Avatar
 
Join Date: Oct 2012
Location: Portland, Oregon
Default Multi stage output digital bi amp?

First, I'm not an amplifier designer so I'm out of my element but this seems like a logical evolution to me. I hope this hasn't been posted or asked before. I see several threads relating to using multiple class D amplifiers in a system or using multiple stages to reduce THD but I have a different idea.

Since the output of a class D amplifier is driven by a clock, is there a way to separate frequency components before the output stage and add extra output stages for each speaker you intend to drive. One could possibly make a nearly zero phase sound system if the signal can be processed fast enough.

The idea is to have an input that is sampled at an appropriately high rate, say 400Khz or more. The sampling block would contain a discrete sample-and-hold for each output stage. For a two way speaker the amp would have two output stages driven off a single power supply. Each digital filter would then clock the appropriate output based on a comparator in the digital filter. I'm not a digital designer but it seems like it would be fairly easy to use a sample-and-hold --> bridge (DC analog of signal)--> comparator --> class D output.

The filter would need to take a sample then hold it. The comparator, synched to the sample clock, would then enable it's output based on the current signal. If the delta between current signal and previous sample is above a threshold this is a tweeter sample. If it's below a threshold this is a woofer sample. The comparator would enable it's output to drive the appropriate output stage in the same way a single channel class D does. I'm not even sure if this kind of circuit would work but the idea is to obtain an output drive signal based on the frequency of the input.

With the addition of more advanced signal processing before the crossover an inverse phase curve could be applied to correct for system group delay and/or one could also apply simple frequency EQ.

Additionally, If one can apply an inverse of the system group delay to the appropriate driver output it would effectively reduce it to near zero, minus any resistance or reactance between the speaker and the amp or other unaccounted for variables.

A zero phase system?
Is such a thing possible, likely, even in the works?
Seems to me like the next logical step in sound system evolution.
  Reply With Quote
Old 8th October 2012, 09:21 PM   #2
diyAudio Member
 
Join Date: Oct 2010
Location: Canada
Yes an analog crossover will introduce a phase shift. A digital crossover can be done without delay. I believe it makes more sense to do the processing of the signal before the power stages (analog crossover or DSP) and let the power stages be less complicated. With digital sound processing (DSP) it can do similar to what you describe as its a mini computer with RAM that can compute this information (correct delays, crossover, and EQ) and output it to multiple power stages. Just my 2cents anyways.

Also most class d amps are going in the direction of self oscillation, means there is no fixed clock... it varies.
  Reply With Quote
Old 9th October 2012, 07:59 PM   #3
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
Universe does not work like that.

The class D schemes that perform best don't use a fixed clock, but rather a self-oscillating free-running delta sigma approach (it compensates open loop gain automatically for power supply variations, it corrects the error of every cycle in the next cycle, and it corrects output filter resonance (ages ahead of triangle wave fixed frequency schemes), with just a few op-amps and a zero cross comparator).

To perform any form of useful digital filtering for speakers, a sampled input (either from digital source or from A/D conversion), RAM memory to store data, and a a high bit depth ALU (arithmetical logical unit, capable of signed addition and multiplication), are required. But delta sigma self oscillating class D takes an analog input, so D/A conversion is required too.
__________________
I use to feel like the small child in The Emperor's New Clothes tale
  Reply With Quote
Old 9th October 2012, 08:20 PM   #4
AP2 is offline AP2  Italy
Banned
 
Join Date: Dec 2009
Location: Italy
Your idea is very original, in this case the freq. carrier should be much higher, if you want to reserve a channel for 400Khz. in theory is a system already in use, such as in some measuring instruments. I believe that the problem of delay is already solved with the use of IIR in dsp. for class D, it all seems very simple (as it says EVA), strange that some engineers have lost 10 years, in an attempt to play a good self-oscillating hehe!

Regards
  Reply With Quote
Old 9th October 2012, 09:29 PM   #5
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
Of course there are many ways to do it, you can obfuscate it, or make it switched-capacitor-filter-like for 40dB SNR. But there will always be an inherently universal beauty in the way the real-world error components present at the output of the system (distortion) are dealt with in delta-sigma systems, and this beauty is lost in every other scheme. It's an analog calculator that can be made very accurate with little effort, replacing millions of logic gates.

btw: I've done some progress in DSP programming during the last year, I'm doing digital feedback loops among other things, that's why I don't show up often. I'm finding out what I already knew: Getting both high precision and high bandwidth from a digital loop is very expensive in terms of complexity, you can have either one, but not both. For example I use a digital loop for temperature limiting in amplifiers, because there are thermal time constants in the range of minutes involved (a real pain to do this with op-amps).
__________________
I use to feel like the small child in The Emperor's New Clothes tale
  Reply With Quote
Old 9th October 2012, 09:48 PM   #6
sreten is offline sreten  United Kingdom
diyAudio Member
 
Join Date: Nov 2003
Location: Brighton UK
Quote:
Originally Posted by Siggma View Post

First, I'm not an amplifier designer
Hi,

Or a speaker designer or anyone au fait with the real issues regarding
phase and group delay and the consequences of digital manipulation.

Not that I know that much about it, because its pretty inconsequential
in the big scheme of things that matter regarding quality reproduction.

rgds, sreten.
__________________
There is nothing so practical as a really good theory - Ludwig Boltzmann
When your only tool is a hammer, every problem looks like a nail - Abraham Maslow
  Reply With Quote
Old 9th October 2012, 11:11 PM   #7
Siggma is offline Siggma  United States
diyAudio Member
 
Siggma's Avatar
 
Join Date: Oct 2012
Location: Portland, Oregon
Quote:
Originally Posted by sreten View Post
Hi,

Or a speaker designer or anyone au fait with the real issues regarding
phase and group delay and the consequences of digital manipulation.

Not that I know that much about it, because its pretty inconsequential
in the big scheme of things that matter regarding quality reproduction.

rgds, sreten.
Huh?
  Reply With Quote
Old 9th October 2012, 11:25 PM   #8
Siggma is offline Siggma  United States
diyAudio Member
 
Siggma's Avatar
 
Join Date: Oct 2012
Location: Portland, Oregon
Quote:
Originally Posted by Eva View Post
Universe does not work like that.

The class D schemes that perform best don't use a fixed clock, but rather a self-oscillating free-running delta sigma approach (it compensates open loop gain automatically for power supply variations, it corrects the error of every cycle in the next cycle, and it corrects output filter resonance (ages ahead of triangle wave fixed frequency schemes), with just a few op-amps and a zero cross comparator).
With a high clock rate, I'm thinking 2 Mhz or a multiple of the lowest oscillator frequency, it could easily be integrated into a self -oscillating design. The crossover would simply act as an OR switch for the drive clock of an existing amp. Duplicate the output stage for a tweeter and add some OR logic to select which output to clock based on the crossover. Seems simple enough if the drive oscillator is high enough to accurately separate the samples.

Eh? Maybe?
I guess the "proof is in the puddin". I'll have to build one. I was hoping someone else would get all excited and do it for me.
  Reply With Quote
Old 10th October 2012, 03:09 AM   #9
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 any 2-way system, poles and zeros at very specific locations are required in each output for proper acoustical sum (at least for matching the acoustical phase of both drivers on-axis across crossover frequency region, the best compromise). I have already told briefly how the samples have to be processed, it's not just a matter of selecting them, frequency division requires complex math and memory of previous samples to calculate new LF and HF samples from input samples.
__________________
I use to feel like the small child in The Emperor's New Clothes tale

Last edited by Eva; 10th October 2012 at 03:12 AM.
  Reply With Quote
Old 10th October 2012, 02:50 PM   #10
Siggma is offline Siggma  United States
diyAudio Member
 
Siggma's Avatar
 
Join Date: Oct 2012
Location: Portland, Oregon
Quote:
Originally Posted by Eva View Post
In any 2-way system, poles and zeros at very specific locations are required in each output for proper acoustical sum (at least for matching the acoustical phase of both drivers on-axis across crossover frequency region, the best compromise). I have already told briefly how the samples have to be processed, it's not just a matter of selecting them, frequency division requires complex math and memory of previous samples to calculate new LF and HF samples from input samples.
So you're saying the error that would occur with two different clocks would induce random phase and probably frequency distortion at the output because there is no way to synch the samples?

I'm still convinced there is a way to do it given enough time and consequently experience. I suppose only time will tell. Thank you for your input, no pun intended...
  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
Input stage for multi-zone amp. wsmckenz Class D 2 3rd February 2010 12:45 AM
Bench Digital Multi Meter TheDriver Parts 20 24th July 2008 12:52 AM
Multi-stage amplification sam9 Solid State 2 5th August 2007 05:49 PM
Help needed - H-bridge, ClassD output stage IC for digital amplifier gretzteam Digital Source 12 19th January 2003 07:08 PM
Help needed - H-bridge, ClassD output stage IC for digital amplifier gretzteam Solid State 0 16th January 2003 01:32 AM


New To Site? Need Help?

All times are GMT. The time now is 09:29 AM.


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