SystemD_2kW, any interest for an open design? - Page 9 - 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 29th November 2012, 07:54 PM   #81
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
Quote:
Originally Posted by phase_accurate View Post
...
This one here is a patent by Toshiba which comes quite close to our proposals. Although the technical implementation is different to our proposals the general idea is quite close. From the legal perspective the claims would have to be analysed in detail in order to be sure.
...
Espacenet - Bibliographic data
Yes, this seems to be close. The sketch and abstract on the cover look very close to Charles approach. Less close to the method according posting #3, but in fact all claims would have to be checked before commercially using any of our discussed ideas for the dynamic hysteresis.

P.S. I think legal perspective does not just mean the claims, but also who can afford the stronger attourneys for unlimited time.

In any case it should be clear to every reader that the coming amp is not intended for commercial business. Neither from legal perspective nor from design perspective.
It is simply for DIY enthusiast, who know about dangerous voltages.
  Reply With Quote
Old 30th November 2012, 01:08 AM   #82
diyAudio Member
 
nigelwright7557's Avatar
 
Join Date: Apr 2008
Location: Carlisle, England
IR do a 1.7KW amplifier project using IRs2092 and gate buffers to up the gate drive.
  Reply With Quote
Old 30th November 2012, 05:20 AM   #83
diyAudio Member
 
Workhorse's Avatar
 
Join Date: Dec 2004
Quote:
Originally Posted by nigelwright7557 View Post
IR do a 1.7KW amplifier project using IRs2092 and gate buffers to up the gate drive.
Variable frequency response w.r.t. load, pre-filter feedback projects
  Reply With Quote
Old 30th November 2012, 05:24 AM   #84
diyAudio Member
 
Workhorse's Avatar
 
Join Date: Dec 2004
Quote:
Originally Posted by Ouroboros View Post
I wonder if you could use a window comparator, followed by an S-R latch as the modulator instead of a single comparator with hysteresis? I used the same arrangement in the past where I wanted variable hysteresis without all of the tolerance issues that setting an accurate hysteresis level involves. You have to provide symetrical threshold voltages to the window comparator to set the (emulated) hysteresis effect, but that shouldn't be too difficult, although very high speed op-amps would be required to follow the switching waveform. A hastily drawn Visio pic is attached.
Brilliant, the threshold could be set very easily thru symmetric current sources/mirrors.
  Reply With Quote
Old 30th November 2012, 01:14 PM   #85
diyAudio Member
 
Ouroboros's Avatar
 
Join Date: Dec 2003
Location: Nottingham UK
How do we determine the transfer function for the required hysteresis level in relation to the output signal peak value? I see that the original schematic used the input signal level to determine this, whereas the main effect on the self-oscillation frequency will be how close the output signal approaches to the supply rails. For most of the output signal range, the self-oscillation frequency does not change much, but the frequency starts to drop quickly as the peak of the signal gets closer to the supply rail voltage. I guess that the control signal into the (non-linear) signal processing block that determines the hysteresis level has to be from circuitry that provides a 0V-referenced voltage, proportional to the absolute value of the difference between the output signal level and the relevent +ve or -ve dc supply rail voltage?

Hope this makes sense, I've been thinking about this over lunch!
  Reply With Quote
Old 30th November 2012, 01:39 PM   #86
diyAudio Member
 
Ouroboros's Avatar
 
Join Date: Dec 2003
Location: Nottingham UK
Quote:
Originally Posted by Ouroboros View Post
I see that the original schematic used the input signal level to determine this
Sorry, I meant in the schematic that Charles posted in #52.
  Reply With Quote
Old 30th November 2012, 01:50 PM   #87
diyAudio Member
 
Join Date: May 2002
Location: Switzerland
I'd say it is actually a function of the audio (!) signal at the input of the modulator. If taken from the output filter you'd have a lagging control signal which might also cause distortion at higher frequencies. It would be best to regard the AMP as LTI system and determine the LTF from input to modulator and use this LTF to EQ the input signal accordingly before feeding it the hysteresis modulator. The next step to increase accuracy would be to determine whether a simple proportional factor is OK or if some nonlinear processing would be needed as well.

Regards

Charles

Last edited by phase_accurate; 30th November 2012 at 02:04 PM. Reason: typos
  Reply With Quote
Old 30th November 2012, 07:24 PM   #88
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
My first sims of the clamping method took the signal from the input with an OP amp, but it was much less convenient to parametrize.
The difference between output signal and rail is the key to understand the frequency shift and output voltage ripple
Towards the close rail you need very long ON times to derive a certain change in the output voltage, the fastening towards the other direction cannot compensate this.
Note, towards the close rail the driving voltage across the output chokes converges to zero and consequently the necessary ON time converge towards unlimited, while towards the far away rail the voltage across the choke will not more than double vs. idle situation. Besides the voltage across the choke, also the superimposed low frequency current influences the the relation of ON times and different dv/dt of both directions - but it does not change the overall picture in a fundamental way.

Above observation and considerations together pushed me to improve the hysteresis modulator by designing a circuit which uses the difference between output and rail as control criteria instead of the input voltage.
  Reply With Quote
Old 1st December 2012, 10:26 AM   #89
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
Hi Charles,
no matter if Ouroboros and myself are pointing more to the difference between output and rail, this still allows a math modelling similar to your proposal.
It would just need to introduce a term with a difference vs a constant which is reflecting the (weighted) rails.
Well, rail sagging would again complicate the equations, but I think everything is possible.
It is just a question from which direction we look at the system.
When coming from distortion, it is absolutely making sense to describe the correction term as a function of the audio signal at the input.
When looking to it from the perspective of output ripple and frequency shift it is more intuitive to use the difference between output and rail.
By the end of the day we do not change physics by our math model.
Also the similarities of the simulation results of both implementations does not show massive short comings in any of both approaches.
  Reply With Quote
Old 1st December 2012, 12:46 PM   #90
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
I was not satisfied with the performance of my 320kHz design and at the same time I was not happy with the 400kHz version which would chase IRFP4668 at 400kHz/160V/60A hard switched.

In the mean time I have settled a new adjustment which is fitting mostly my taste.
It has a comfortable idle switching frequency around 320kHz, good step response and bandwidth around 80kHz.
Harmonics at 400W/2R are in the -80db range, means further optimization by simulation would be completely meaningless, because in this range simulation and reality typically differ already pretty much.
Output voltage ripple is slightly below 500mVrms during idle.

No lunch for free.
These selfoscillating topologies are a tight knot in it self. No matter what you change, all other parameters change as well. Swichting frequency, loop gain, carrier shape, output ripple, step response...
Achieving overall good results with 320kHz does consequently lead me to a design with higher output filter inductance. 18uH.
Ok, that's not really high and for ripple current load of the e-caps this is even fortunate .
The downside of this higher inductance value is that it impacts the physical size of the output inductor and it also limits the power bandwidth of the amp.
The new adjustment has a bandwidth of 80kHz, but the high filter inductance is limiting the -3db power bandwidth to approx 24kHz (into 2R), which is just acceptable. Well, my intended MKP2 filter caps will also be close to melt down (at least running beyond spec), when applying something like 20kHz with full power as continuous load...

Attached the updated circuit and some simulation results. Please note the harmonics where derived from a more extensive simulation, which includes a switching stage with IRFP4668.
Attached Images
File Type: jpeg 18uH_Version.jpeg (235.5 KB, 491 views)
File Type: jpeg Step_18uH.jpeg (278.1 KB, 468 views)
File Type: jpeg Harmonics_18uH.jpeg (119.6 KB, 406 views)
  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
O2 Front Panel INTEREST POLL your own design Llama16 Group Buys 1 12th September 2012 06:23 PM
Probing interest in My inrush limiting design nattawa Power Supplies 2 19th November 2011 02:08 PM
Custom open baffle 12" woofer - any interest? SimontY Group Buys 2 4th March 2009 02:53 PM
Open baffle design lawbadman Multi-Way 1 17th January 2008 01:15 AM
Open call for suggestions on Open Source DIY Audio Design gfergy Everything Else 1 15th April 2007 08:33 AM


New To Site? Need Help?

All times are GMT. The time now is 01:38 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