New self oscillating post filter feedback topology... - 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 23rd April 2010, 07:00 PM   #1
diyAudio Member
 
Join Date: Apr 2010
Default New self oscillating post filter feedback topology...

Hello all,

I m quite new here but quite interested in this topic.
My history: I ve a masters degree in power electronics, about 10 years ex in power supply engineering. Starting many years ago, I ve done some SIMETRIX work on sigma delta self-oscillation with feedback be4 filter, built one 5kW subwoofer amp based on this very easy control topology using 1200V APT IGBTs at 600V bus / 20kHz many years ago (still working - pics avail.), then come over to play around with UCD style post filter feedback control to try to get a feeling on how it would behave on things like step
response, no load, slew rate etc.... in transient sim...

..attached you can find "my approach" to UCD in "block style" - not component level - in SIMETRIX. Its a 180V bus powered full bridge using 200V components at ~470kHz capable of giving >3kW at 4 Ohm, much like UCD 2K. FETs in SIM are simplified by switches (20mOhm / 1MOhm), series schottky is missing (no effect for control loop analysis, but for switching losses), antiparallel diode would be 200V ultrafast or 300V carbide in reality. S9 and R4 are only a replacement for a comparator so R4 voltage will be
square either 0 or 12V. R10, C5 will give about 200 ns to simulate delay between comparator inputs crossing and actual bridge nodes moving. L3 / R2 is the load impedance to play around with - can be ohmic or "loudspeakeric" or passive X-over network or....
Result: works in simulation.

Next step: I invented some new control topology that I will call self oscillating Class D (SOCD).
Goal: PA full bridge amp using 180 V bus, nearly same performance as UCD in things like step response, output resistance / damping, frequ. response >30.000Hz.... gain 35dB, workable from 2 Ohms to 32 Ohms at up to 120VRMS output voltage and fully stable at no load with slight overshoot accepted at no load /square wave input (which is not really
the typical application....). Less frequency decline than UCD at very low or very high duty cycles. Still self-oscillating, but no need to infringe bruno's patent should it be built (actually planning on this).
You can find the "block-style" simulation file attached. TX 1 is a transformer on MPP material instead of 2 chokes. Why: Only 70.7 % turns is needed for the same "effective" inductance if 2 chokes wound on 1 core. Have only one component instead of 2. (Core would still be as heavy as 2 single inductors cores coz the sum energy stored equals 2 inductors). One winding is 13.3uH, but acts like 26.5uH. L4 / R12 is an air coil. It can be used to tune the oscillation f. w/o soldering just by mechanically pulling and pushing it. R36, R13 and especially R35 could face some real watts during bursts of 5-digit frequency waves at high power, so use 10W non inductive
types, although average power will be much lower. The resistors help to improve transient response at very high ohmic loads.
Result: Sim results look quite ok, goals more or less achieved. The air coil will also give an effective 4th order filter for higher harmonics >1MHz, so EMI will be much improved over single LC filter.
Why am I posting this:

I d like to hear and test this thing, but as I am quite on the lazy side, I need someone to do some layouting and soldering / assembling this in 2 channels for me and send over 1 stereo unit to me. My part would be:
Define schematics and define actual components, define air coil dimensions, help to get and provide components like hi side driver / FETS / diodes / foil caps etc., help and give (remote) advice on layout, power up the unit carefully in my lab and send back final "rework instruction" to you so you can really get to see it working. AND as a special bonus: I'd send you over a 5kW PFC powered 3 ph input isolated 177Vdc Power supply for operating the amp. (It will actually be 3 single phase telcom rectifiers with series'sed outputs, so would also work on just one "household" phase - with less output power).
Anybody here interested?
Attached Images
File Type: jpg UCD.JPG (282.7 KB, 1646 views)
File Type: jpg SOCD.JPG (285.4 KB, 1601 views)
  Reply With Quote
Old 23rd April 2010, 08:46 PM   #2
stoc005 is online now stoc005  United States
diyAudio Member
 
stoc005's Avatar
 
Join Date: Feb 2007
Location: Midwest in the USA
I've been simulating a similar full bridge since the fall. Mine looks very much the same as yours except the filter is conventional and I run at up to 60V, single supply. The only reason (but there are a few more advantages too....) to do a full bridge is to be able to run off a single supply, IMHO. I'd like to understand the output filter in the second picture a bit more.
The bridged to single ended conversion by the opamp, and I have the same thing, bothers me since the only reason it is there is to provide the differential to SE conversion and no gain. I wish there was a way to do this with passives.....Yes, I know....a transformer.
I think your design should work quite well.
BK
  Reply With Quote
Old 23rd April 2010, 09:28 PM   #3
diyAudio Member
 
Join Date: Apr 2010
@stoc005: TX1, C18, C9 is a 1st order LC slightly damped by R13. Air coils, C17, R36 is a low impedance resonator that will set the resonant frequency by well defined phase decay - so resonant freq. is (almost) independent on output load as long as load does not get significantly lower than 1.5 Ohm or so... evrything optimized by many sim runs.. R35 C7 is a "bonus" damping and can be avoided if no load step response is no issue (if focus application is only 2 - 8 Ohms...)
the opamp doesn't really worry me since its cost is bout one "buck" .. as you call it over there.. i'm more worried on getting a precise 100V / us comparator
  Reply With Quote
Old 24th April 2010, 11:46 AM   #4
savu is offline savu  Romania
diyAudio Member
 
savu's Avatar
 
Join Date: Feb 2007
Location: Cluj - Napoca
Send a message via MSN to savu Send a message via Yahoo to savu
Quote:
Originally Posted by ViennaTom View Post
Hello all,

I m quite new here but quite interested in this topic.
My history: I ve a masters degree in power electronics, about 10 years ex in power supply engineering. Starting many years ago, I ve done some SIMETRIX work on sigma delta self-oscillation with feedback be4 filter, built one 5kW subwoofer amp based on this very easy control topology using 1200V APT IGBTs at 600V bus / 20kHz many years ago (still working - pics avail.), then come over to play around with UCD style post filter feedback control to try to get a feeling on how it would behave on things like step
response, no load, slew rate etc.... in transient sim...

..attached you can find "my approach" to UCD in "block style" - not component level - in SIMETRIX. Its a 180V bus powered full bridge using 200V components at ~470kHz capable of giving >3kW at 4 Ohm, much like UCD 2K. FETs in SIM are simplified by switches (20mOhm / 1MOhm), series schottky is missing (no effect for control loop analysis, but for switching losses), antiparallel diode would be 200V ultrafast or 300V carbide in reality. S9 and R4 are only a replacement for a comparator so R4 voltage will be
square either 0 or 12V. R10, C5 will give about 200 ns to simulate delay between comparator inputs crossing and actual bridge nodes moving. L3 / R2 is the load impedance to play around with - can be ohmic or "loudspeakeric" or passive X-over network or....
Result: works in simulation.

Next step: I invented some new control topology that I will call self oscillating Class D (SOCD).
Goal: PA full bridge amp using 180 V bus, nearly same performance as UCD in things like step response, output resistance / damping, frequ. response >30.000Hz.... gain 35dB, workable from 2 Ohms to 32 Ohms at up to 120VRMS output voltage and fully stable at no load with slight overshoot accepted at no load /square wave input (which is not really
the typical application....). Less frequency decline than UCD at very low or very high duty cycles. Still self-oscillating, but no need to infringe bruno's patent should it be built (actually planning on this).
You can find the "block-style" simulation file attached. TX 1 is a transformer on MPP material instead of 2 chokes. Why: Only 70.7 % turns is needed for the same "effective" inductance if 2 chokes wound on 1 core. Have only one component instead of 2. (Core would still be as heavy as 2 single inductors cores coz the sum energy stored equals 2 inductors). One winding is 13.3uH, but acts like 26.5uH. L4 / R12 is an air coil. It can be used to tune the oscillation f. w/o soldering just by mechanically pulling and pushing it. R36, R13 and especially R35 could face some real watts during bursts of 5-digit frequency waves at high power, so use 10W non inductive
types, although average power will be much lower. The resistors help to improve transient response at very high ohmic loads.
Result: Sim results look quite ok, goals more or less achieved. The air coil will also give an effective 4th order filter for higher harmonics >1MHz, so EMI will be much improved over single LC filter.
Why am I posting this:

I d like to hear and test this thing, but as I am quite on the lazy side, I need someone to do some layouting and soldering / assembling this in 2 channels for me and send over 1 stereo unit to me. My part would be:
Define schematics and define actual components, define air coil dimensions, help to get and provide components like hi side driver / FETS / diodes / foil caps etc., help and give (remote) advice on layout, power up the unit carefully in my lab and send back final "rework instruction" to you so you can really get to see it working. AND as a special bonus: I'd send you over a 5kW PFC powered 3 ph input isolated 177Vdc Power supply for operating the amp. (It will actually be 3 single phase telcom rectifiers with series'sed outputs, so would also work on just one "household" phase - with less output power).
Anybody here interested?
I would be interested in building this baby.

mail me at savu_silviu@hotmail.com for more details

PS. hope you live in europe

Best regards,

Savu Silviu
__________________
1'st rule of chess:
The only way to get smarter is by playing a smarter opponent.
  Reply With Quote
Old 24th April 2010, 08:13 PM   #5
81bas is offline 81bas  Germany
diyAudio Member
 
Join Date: Jan 2006
Location: Kiel
Quote:
Originally Posted by ViennaTom View Post
the opamp doesn't really worry me since its cost is bout one "buck" .. as you call it over there.. i'm more worried on getting a precise 100V / us comparator
It is even possible to build this schematic without any opamps, feeding "differential" bridge output to the differential comparator inputs via resistive voltage dividers (with appropriate "phase lead" networks)...

Wounding full bridge output filter on the single core is a known trick, but using of 4th order LC filter to set selfoscillation frequency is really new and good idea!
  Reply With Quote
Old 25th April 2010, 09:58 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
I have developed my own self-oscillating topology too, at the voltage and power levels that you mention. I more or less managed to solve all the problems related to making it work in the real world and making manufacturing not too complex.

If you do THD analysis, you may notice that neither the 4th order output filter nor less than 1:2 oscillation frequency drop are good ideas. A second active pole may be a good idea, though.

Feding carrier residual to the comparator through op-amps is not a good idea either. I had noise immunity problems (the circuit disturbing itself due to EMI) in some of my early prototypes. There don't seem to be many ways to make it work fine, only a few.

Don't forget about parasitics when modelling output filter. Parasitic L and C will reduce the quality of the carrier residual waveform and degrade THD.
__________________
I use to feel like the small child in The Emperor's New Clothes tale
  Reply With Quote
Old 26th April 2010, 11:37 AM   #7
diyAudio Member
 
Join Date: Apr 2010
"If you do THD analysis, you may notice that neither the 4th order output filter nor less than 1:2 oscillation frequency drop are good ideas. A second active pole may be a good idea, though.
Feding carrier residual to the comparator through op-amps is not a good idea either. I had noise immunity problems (the circuit disturbing itself due to EMI) in some of my early prototypes. There don't seem to be many ways to make it work fine, only a few.
Don't forget about parasitics when modelling output filter. Parasitic L and C will reduce the quality of the carrier residual waveform and degrade THD. "

I think THD is mainly dependent on blanking time and that is dependent on how well timed you can get the switches to work. I think I can't do any THD analysis just by simulation.
I also think its possible to even avoid the opamp, but have not worked out this idea...
Attached you can find some nice step response results of SOCD simulation. Input = Square 2Veff = 2 Vpk, Output = 2, 4, or 8 Ohms resp.
Attached Images
File Type: jpg SOCD SR 2Ohm 1.JPG (166.3 KB, 1329 views)
File Type: jpg SOCD SR 2Ohm 2.JPG (160.5 KB, 1039 views)
File Type: jpg SOCD SR 4Ohm 1.JPG (165.0 KB, 970 views)
File Type: jpg SOCD SR 4Ohm 2.JPG (163.8 KB, 87 views)
File Type: jpg SOCD SR 8Ohm 1.JPG (166.3 KB, 72 views)
File Type: jpg SOCD SR 8Ohm 2.JPG (159.8 KB, 67 views)
  Reply With Quote
Old 26th April 2010, 12:04 PM   #8
AP2 is offline AP2  Italy
Banned
 
Join Date: Dec 2009
Location: Italy
Hi,
DT is also responsable of thd Vs. load (when is <8R).
Even the "R" filter interacts with this at very low load
  Reply With Quote
Old 28th April 2010, 04:47 PM   #9
diyAudio Member
 
Join Date: Apr 2010
Default SOCD - getting rid of the opamp...

For most of you don't like the OPAMP within the feedback, I just eliminated it. Result is attached. Still working - same behavior...
Attached Images
File Type: jpg SOCD new.jpg (163.4 KB, 606 views)
  Reply With Quote
Old 28th April 2010, 08:03 PM   #10
81bas is offline 81bas  Germany
diyAudio Member
 
Join Date: Jan 2006
Location: Kiel
Quote:
Originally Posted by ViennaTom View Post
For most of you don't like the OPAMP within the feedback, I just eliminated it. Result is attached. Still working - same behavior...
  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
Post filter AND pre filter feedback..! Lars Clausen Class D 30 4th April 2013 04:03 PM
question :post filter feedback titanchen68 Class D 38 2nd August 2006 02:00 AM
Question- post filter feedback error correction? darkfenriz Class D 11 13th June 2006 10:31 PM
Voltage Mode Pre and Post Filter Feedback Genomerics Class D 30 6th April 2006 07:27 PM


New To Site? Need Help?

All times are GMT. The time now is 04:25 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