What Class-D amp give best sound?

I wish all the people here, the world, nature and animals an happy and yet very important a better newyear, 2017 was not one of the best, better it was the worst ever.

here I did a full bridge, with a single ended input, a trhee level is also option, but then we need pulses on the zero transistions otherwise we get crossover distortions, we can clock these pulses and inject them there, did see something about the a while ago, do not now where.

The full bridge does supress the even harmonics but I see more uneven harmonics with it, of this sounds bad I do not now, mine ciclotron has also this problem, it is due of balanced technology.

regards

Best wishes for you this new year.

I like how you did the single analog audio stage and derived the bridge drives for the MOSFET totem poles afterward, and that was before you saw the suggestion I wrote.

Yet I didn't expect the outcome. I just came to the idea of the design approach with LTspice and didn't ever get to try a real life circuit, so I can understand that it might not really work so well. I still hope to try it out for myself someday though.
 
Best wishes for you this new year.

I like how you did the single analog audio stage and derived the bridge drives for the MOSFET totem poles afterward, and that was before you saw the suggestion I wrote.

Yet I didn't expect the outcome. I just came to the idea of the design approach with LTspice and didn't ever get to try a real life circuit, so I can understand that it might not really work so well. I still hope to try it out for myself someday though.

Thanks

Here is the schematic for ltspice, just play with it and let now.

regards
 

Attachments

  • ucd orig_sinus_ucd-fullbridge.zip
    2.4 KB · Views: 82

ICG

Disabled Account
Joined 2007
Details. Speakers are 97 dB @ 1w 1 meter. 8 ohm.

globalplayers suggestion of the Sure Electronics AA-AB32313 2x400 (STA516) would be an excellent choice. Don't let the 400W specs deceive you, that's for 3 Ohm, on 8 Ohm it's a lot less. If you want to stay on a tight budget but still improve the sound quality, I would look for some TPA3116(D2) amps, but if you want to spend more, go for Hypex.
 
You're welcome.

The simulation was good. I had some fun with it. Something extra I liked was experience with the .four command to look at distortion.

@Grizlek has also cudo,s from me, because he did help me with this model where I did play with.

In real world it is difficult, I need a good mosfet driver with included dead time adjust, somewhone can advise one who is fast, thanks.
 
You're welcome.

The simulation was good. I had some fun with it. Something extra I liked was experience with the .four command to look at distortion.

@Grizlek has also kudo,s, because he did help me quite good to understand LTspice with this model where I did play with.

In real world it is difficult, I need a good mosfet driver with included dead time adjust, somewhone can advise one who is fast, thanks.
 
I have done some sims, but the UCD version and the delta sigma are the best, delta sigma is even without postfeedback low in distortion.

Here some results, also with squares, maybe time to duild one to see and discover the sound, I am curious about this technology.

Nice to see that current trough mosfets looks the same as with a stereo coder for fm hehe, much more liniair here..

regards.

Thanks for spice files.

This method with pre filter + post filter feedback seems nice, because it keeps filter in feedback.

Can you refer the document this idea originally come from ,is it from IRF ?

Happy New Year To all :D :)
 
Last edited:
Thanks for spice files.

This method with pre filter + post filter feedback seems nice, because it keeps filter in feedback.

Can you refer the document this idea originally come from ,is it from IRF ?

Happy New Year To all :D :)

This one is on UCD principe like phillips did, with a extra intergator op.

Some I do not understand yet is how to setup the integrator, bandwidth etc, what frequency it is tuned at, I need to now that because
this I did not read in the papers, it seems to have a lot possibilities, the 90 degree shift do it keep it that way until bandwidth audio who is setup
in low pass, or has it something to do with the carrier, need the intergrator set up on carrier frequency as I did see in some, the fase shift versions
has a integrator working in the feedback filter, and ucd in the opamp itselfs and feedback.

I have try the sim with IR2010 models, two of them, btl, this did not work well, I think because of the models, maybe I need to rename them before sim.

thanks
 
Last edited:
You claim you need real circuit to vary dead time to see how feedback affects THD.

In reality you can change dead time using the same diode resistor(trimmer) capacitor network + comparator. You vary every rising edge...

Here is one of may schematics using for some buck power supply. This way made for real and works.

HIN and LIN are input to classic low side and high side mosfet driver without internal deadtime, IRF manufactured like IRS2186
 

Attachments

  • Untitled.jpg
    Untitled.jpg
    275 KB · Views: 577
Grizlek

I have drawn one with the ir2010 models and the dead time generator from your sim schematic, I see it does give low distortions, and it is full bridge however when I adjust deadtimes it is difficult, to much and I get shoot through but also when to little.

I think because supply is differential plus and minus 80 volts deadtime setup is difficult, it always crossing wrong? Maybe because of this people do use one supply voltage and a ref level.

Last picture you see what happens, get it not right, not to do with sync such full bridge amp?

regards
 

Attachments

  • ScreenHunter_659 Jan. 04 23.39.jpg
    ScreenHunter_659 Jan. 04 23.39.jpg
    224.3 KB · Views: 537
  • ScreenHunter_660 Jan. 04 23.54.jpg
    ScreenHunter_660 Jan. 04 23.54.jpg
    206.4 KB · Views: 515
Last edited:
Grizlek

I have drawn one with the ir2010 models and the dead time generator from your sim schematic, I see it does give low distortions, and it is full bridge however when I adjust deadtimes it is difficult, to much and I get shoot through but also when to little.

I think because supply is differential plus and minus 80 volts deadtime setup is difficult, it always crossing wrong? Maybe because of this people do use one supply voltage and a ref level.

Last picture you see what happens, get it not right, not to do with sync such full bridge amp?

regards

can I see full bridge schematics.. ? can you also send models for operation amplifiers and IRF ICs like IRF2010.
 
Ohh forgotten to put uit here.

here it is.

Put hierarchy ir2010 into the same folder as the asc file. There are more models but none did work, do not now why.

regards
 

Attachments

  • ScreenHunter_657 Jan. 04 23.03.jpg
    ScreenHunter_657 Jan. 04 23.03.jpg
    140.3 KB · Views: 692
  • ltspice.zip
    42.9 KB · Views: 54
Last edited:
Look at these, picture two is the above menstion one and one is with your ideal switches, there is a loop error in the ir2010 one, I think only build will give answer on the very strange output voltage whie the current looks perfect and low HD.

Good models is what you need when sim a class d or a smps.
 

Attachments

  • ScreenHunter_663 Jan. 05 13.50.jpg
    ScreenHunter_663 Jan. 05 13.50.jpg
    150.9 KB · Views: 625
  • ScreenHunter_662 Jan. 05 13.49.jpg
    ScreenHunter_662 Jan. 05 13.49.jpg
    163.8 KB · Views: 666

Build myself is much more fun, and this amp has not such low dustortions, 0.05 % for 1 watt, 10 % for full power.

The delta sigma ucd did much better, for sim that is.

First picture hysteric delta sigma and the other is een openloop with triangle osc full bridge. All sim so not build, but go do that soon.

regards
 

Attachments

  • ScreenHunter_667 Jan. 05 15.33.jpg
    ScreenHunter_667 Jan. 05 15.33.jpg
    339.4 KB · Views: 563
  • ScreenHunter_672 Jan. 05 21.47.jpg
    ScreenHunter_672 Jan. 05 21.47.jpg
    259.2 KB · Views: 103
Ohh forgotten to put uit here.

here it is.

Put hierarchy ir2010 into the same folder as the asc file. There are more models but none did work, do not now why.

regards

First of all , does it work with my "logic" schematics, without IR2010, this will answer if there is feedback mistake or mistake regarding IR2010 part of schematics. ?!

Do you have good model of IR2010 ?
 
First of all , does it work with my "logic" schematics, without IR2010, this will answer if there is feedback mistake or mistake regarding IR2010 part of schematics. ?!

Do you have good model of IR2010 ?

All the IR series drivers do now well, only in Tina as I did read. I have a model from Analog@IEEE who do work.

I have tryed it in your schematics with the "logic" you did provide, however that are ideal switches.

regards
 
Last edited:
Here is the model and outcome, it is not a post feedback, I can try that also, but fase shifts do make it more difficult, maybe (Bessel?) allpass needed as lag lead? bessel is stable fase shifts..
 

Attachments

  • ucd orig_sinus_ucd-fullbridge.zip
    2.4 KB · Views: 60
  • ScreenHunter_676 Jan. 07 14.34.jpg
    ScreenHunter_676 Jan. 07 14.34.jpg
    234.5 KB · Views: 89