Distorsions in class D - diyAudio
Go Back   Home > Forums > Amplifiers > Class D
Home Forums Rules Articles diyAudio Store Gallery Wiki Blogs Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

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
Thread Tools Search this Thread
Old 19th June 2004, 10:31 PM   #1
diyAudio Member
Join Date: Dec 2003
Location: Munich
Default Distorsions in class D

Hi All!

I am triying to learn about switch mode amps.
For my first step I am going for a simple PWM.
One of my concerns were the distorsions, which might be caused by different dead times of the switching bridge.
In contrafiction to normal common sense on this, my simulations did not show massive increasing distorsions with higher dead time.

My simulation is a simple one. Simple PWM, staright forward.
For basic distorsion analysis and learning to understand... without feedback.
I analyzed the direct PWM signal, right after generation from by comparing the audio siganl against a triangle...
And I analyzed the output signal of the power bridge...
At small dead times they are more or less the same.
At high dead times the difference is smaller than expected.
Even at dead times around 800ns!
Also not clear for me is, how I can get quite high distorsion around the 70 harmonics (around 14kHz). I cannot be the high HF-ripple of the signal, because the switching frequency was 70kHz.

The triangle signal for the comparator has a frequency of 70kHz ==> switching frequency 70kHz. I simulated a 200Hz audio signal and checked for 100 harmonics.
This covers the audioable range and can still be handled by the demo version of ORCAD 9.2.

Further on I checked what happens at small signals. About 10% of max. output level.
Here I was quite unlucky.
The distorsion went up to 30...40...50% !!!!
Please refer to the attached file. It contains the simulation schematic and the output file. At the end of the output file you can see the results of the FFT.

Looking forward to your comments
Attached Files
File Type: txt pwm.txt (20.4 KB, 197 views)
  Reply With Quote
Old 19th June 2004, 11:36 PM   #2
Pafi is offline Pafi  Hungary
diyAudio Member
Join Date: May 2004
Location: Budapest
Can I ask other format? I don't have OrCAD.

The distortion caused by dead-time is just like crossover distortion. Amount of harmonics doesn't depend on output level. (They do, but not much.)

Is there any LC filter on output? If not, then absolutely no distortion occured (theoretically).
  Reply With Quote
Old 19th June 2004, 11:38 PM   #3
diyAudio Member
Join Date: Dec 2003
Location: Munich
...for your fun, please find attached the ORCAD project files...
If you have a ORCAD 9.2 light release ("light" means free demo version) or higher, you should be able to play around with my simulation.

Attached Files
File Type: zip diypwm.zip (55.9 KB, 167 views)
  Reply With Quote
Old 19th June 2004, 11:45 PM   #4
diyAudio Member
Join Date: Dec 2003
Location: Munich
Hi Pafi,
unfortunately I do not have any other format.
But the text file of my first post should show the simulated schematic.
Didn't it work to open the file?

I simulated without load and picked the signal after a simple RC-filter of 1k/6.8nF.
Already the filtered signal of the simplest generated PWM shows that distorsions... this is what is confusing me.

Yes I agree the main distorsion were always at the same frequency independend from the dead time and independend from the signal level. But the percentage of that distorsion is depending very much on the signal level. It seems like the absolute signal at 14kHZ is constant and if I decrease the signal then the relation between that undesired 14khz signal and audio signal becomes poor....
  Reply With Quote
Old 19th June 2004, 11:47 PM   #5
diyAudio Member
Join Date: Dec 2003
Location: Munich
....and the link for the demo version of ORCAD.....
  Reply With Quote
Old 19th June 2004, 11:48 PM   #6
Account Disabled
Join Date: Feb 2004
Send a message via AIM to classd4sure Send a message via MSN to classd4sure
Hi Markus,

I too would very much like to know more about this. Why 100 harmonics? From what I've read I dont' think anyone tests with more more than 8 or so, why? I'm not exactly sure. The more you use, the looooooooonger it's going to take to complete.

I've also read something about doing additional "hard filtering" of the output to help sort out switch noise ...

Did adjusting the dead times effect this 70khz noise?
Circuit I'm working on is a little different, the THD decreases with increased power output, but gets rather poor at lower input levels.

I would like some standard types of tests we could do easily with pspice which would be meaningfull to the cause.

-Signal to noise ratio...I understand that's output - input...Must have to account for the gain though? How can this be done on pspice.

-A standard and meaningfull THD test....8 harmonics or 100 or 1000???? Additional filtering required for switch mode? Best settings in setting up the test for the best results.

-Can THD be graphed simultaneously with stepped power levels?

-Anything else I don't know enough about to have thought of.

-Said tests which could easily be reproduced IRL with real components and equipment for direct comparison with simulation, as in, standard scopes.

-At what point, if any, does pspices THD become unreliable? Once it's become so good that you just have to build it in order to improve it further with real test equipment.

-What things such as dead time, on resistance, etc, contribute to THD, and in what way? Where does second harmonic, or third harmonic distortion come from in class D?

These questions need answers.

Thanks for starting this thread, hope it keeps growing for a long time to come.

  Reply With Quote
Old 20th June 2004, 12:18 AM   #7
Pafi is offline Pafi  Hungary
diyAudio Member
Join Date: May 2004
Location: Budapest
maybe a screenshot? I won't install OrCAD again. I did it once and it was a nightmare. (Usage too.)

"Yes I agree the main distorsion were always at the same frequency independent from the dead time and independent from the signal level."

I didn't write about frequency. I wrote about voltage.

"But the percentage of that distorsion is depending very much on the signal level."

Yes. If distorted components don't change, and input level decrease, then the distortion ratio will obviously grow.

This output filter is not real, and has very different effect than LC output filter.

I didn't check your file, but note that time step must be lower than 7 ns, if you want 0,1% accuracy. Is it so now? I guess not. (Distortion on theoretically perfect PWM signal indicates imperfect simulation.)
  Reply With Quote
Old 20th June 2004, 12:22 AM   #8
subwo1 is offline subwo1  United States
diyAudio Member
subwo1's Avatar
Join Date: Jan 2002
Location: North American Continent
Hello, I suspect that the low correlation between dead time and distortion is related to not using feedback. When the loop is closed, more dead time causes less synchronicity between the input and the feedback signals.
  Reply With Quote
Old 20th June 2004, 12:31 AM   #9
diyAudio Member
Join Date: Dec 2003
Location: Munich
Hi Chris!
I also hope that this thread will grow and help for better understanding.
In fact I do not have answers for most of your quetsions.
I analyzed 100 harmonics because I wanted to see the full audioable frequency range.
I di not expect any issues at the 70 harmonic (means at 14kHz).
In fact I am not 100% sure if the simulation is tricking me.... you never know...
On the other hand I picked small steps for the transient analysis and the probe trace look exactly correct to what I would expect.
Up to know I have no major reason for doubts against the simulation. If you would use the default settings of ORCAD then
there might be bigger deviations. But typically you can see if the traces in the probe window show long linearized lines and strange corners. So you normally you notice if the simulation is rubbish.
And additional I think, distorsion below 0.1 % may just have the meaning "low distorsion". ...would not trust if 0.1 or 0.05....
That's another reason why I started this analysis without feedback.
But of course simualtion time is high if you do not have a very fast computer. To be honest, I used my laptop from work, because it is 5 times faster than my private desk top.

I wish I could also use one of our LeCroy scopes which we have in work.... But that's not possible, so I will have to deal with my old
analog tube base Tektronics with defect triggering ...
But I think it will still take me some time before I can go to practice with that amp anyway...

Bye and good night to all
  Reply With Quote
Old 20th June 2004, 12:32 AM   #10
Pafi is offline Pafi  Hungary
diyAudio Member
Join Date: May 2004
Location: Budapest
It's from the txt:

".TRAN 0 40ms 0 5us"

5 us? It's huge! Don't be surprised by results!
  Reply With Quote


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
Class D Vs Class I (Balanced Current Amplifier) rmsaudio Class D 29 26th January 2014 10:25 PM
line-array vs conventional point-source/ distorsions issues crazyhub Multi-Way 26 2nd July 2007 12:03 AM
How about a round-up of Class A kit power amps, or collectable vintage class A? Brisso57 Solid State 4 14th February 2007 10:30 AM
Zen v4 and bass distorsions ... Stabist Pass Labs 4 24th September 2002 11:59 PM

New To Site? Need Help?

All times are GMT. The time now is 06:28 AM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2