Another classd from Tekko Audio Labs

[Tekko]

Im once again working on a class d, i´d revisiting the sigma delta class d that at first try failed.

This is the one i tried building but failed: http://ece.wpi.edu/analog/mqps/ClassD_MQP_Final.pdf

Now i built this: http://diymania.hv4all.com/sigma delta modulator/sigmadeltamodulator.PNG and it works nicely, apart from a slight beep, which i blame on a crappy layout on a breadboard with no bypass caps what so ever on the rails.

It runs off a TL074, a 74HC74 with a 8MHz clock and uses a TC4422 as output device. The output filter is yet a dc choke, gonna wind a proper choke later on.

What do you think, worth to continue to work on ? It does sound very good for what it is.

 

[Tekko]

Full schematic:

 

[Lars Clausen]

Tekko: Drop the 1 bit quantiziser, it will only increase quantization noise. Simply short it out, then you are in better shape. 🙂

 

[Tekko]

Now i have improved this thing quite a bit.

Only background nice that is hearable now is a faint hum that i dunno where it comes from since this thing runs off a smps.

Sound is very good for beeing a quick lashup on a breadboard🙂

It looks like this: http://diymania.hv4all.com/sigma delta modulator/sdamp.wmv

 

[fdeck]

Tekko: Drop the 1 bit quantiziser, it will only increase quantization noise. Simply short it out, then you are in better shape. 🙂

I did as you suggested, modeling it with LTSpice. In fact I left out the quantizer and the power driver. Sure enough, it works. Is that all there is to a self-oscillating Class-D?

What determines the modulation frequency?

 

[fdeck]

Also, thanks Tekko for the interesting circuit.

 

[Tekko]

This amp is now stereo and sounding good, but since its a breadboard, the channels interfer with each other = hissing.

Nope i dident drop the quantizer. Removing it would make this a standard self oscillating pwm amp, not delta sigma as it is now.

Im ny case the TL074 limits freq to 2MHz, should be 4MHz, clock is 8MHz as stated before.

 

[Tekko]

Here are some measurements:

http://diymania.hv4all.com/sigma delta modulator/d-amp measures/

 

[phase_accurate]

These results are quite good for a first-order loop. Are you going to increase the loop order ? Do you intend to add dither ? Have you somehow determined the average switching frequency when driven by an input signal ?

Regards

Charles

 

[Tekko]

I have thought of adding more stages, dithering i dunno what that is really.

 

[phase_accurate]

Dithering is the intentional addition of noise inside the loop in order to decorrelate modulator noise patterns (so called limit-cycles) that may lead to "tones".

Regards

Charles

 

[Tekko]

I managed to kill the noise so all that is left now is small artifacts on some frequencies and a faint hum.

 

[fredos]

Why did you try to run it at 8 mhz? You will never be able to drive power stage at that frequency...I already use this topology in my HVI serie amplifier, but take low frequency feedback at the output and hight frequency feedback before the coil...Work very weel at 500 Khz...

Fredos

 

[Tekko]

Hey fredos!

Well the 8MHz xtal module was all i had, i also had a 2MHz xtal module but that one was broken, no output.

Yes i know i cannot run a real output stage at these frequencies, but i can use another 74HC74 to lower the frequency of the master clock to a lower frequency.

 

[Tekko]

Heres a schemo of a new design i´ve whipped together.

http://diymania.hv4all.com/selfosc classd/D-Amp2.png <-- will it work with the level shifters and that stuff ?

 

[phase_accurate]

Why do you use a TL 082 as a sluggish comparator instead of feeding to integrator output to the first schmitt-trigger ?

Regards

Charles

 

[Tekko]

Good question! Never thought of that really.

Tried it now on my simple prototype and it works!!

I´ll go ahead and cad the board.

 

[phase_accurate]

OK at least you understood what I meant to say. For those who didn't the above sentence should read: "....... instead of feeding THE integrator output to ........"

Congrats on your working test setup.

Regards

Charles

 

[Tekko]

Edit: by removing the comparator, THD shot shyhigh, from 0.07% with the comparator, to 8% without the comparator.

 

[phase_accurate]

Maybe your switching frequency is now way off - which you could set by changing the integrator cap.
What ground are the Schmitt-triggers referenced to ?

Regards

Charles