Class D and triangle wave linearity

Thread Starter

Wendy

Joined Mar 24, 2008
23,415
I would not class either one of the above as pure class D types, the second especially. A good class D must be frequency stable, and vary over a wide range of PWM modulation. The wide range of PWM is especially critical for maximum efficiencies, this is in fact the main drive for this kind of amplifier.

I have thought of using the FM modulation characteristics of the second design for a optical FM transmitter. Basically the range of PWM is atrocious, and they added a note for a filter as an after thought, this should be included as part of any design.
 

THE_RB

Joined Feb 11, 2008
5,438
You mean the LM393 (dual comparator)? It is good for 36V. Two chips, 40106 and LM393, I didn't spell it out before but that is what I was thinking.
...
You just need a dual comparator. One comparator (with an open collector output) makes a sawtooth oscillator at a fixed frequency. The C charges slowly via R, then the open collector output discharges C very quickly giving a sawtooth output.

Then the second comparator is the PWM generator, detecting the point where the audio signal crosses the sawtooth signal.

Then for a simple output you just need 2 FETs as a half-bridge, and capacitor couple the speaker.

In the interest of education and simplicity you could get it down to not much more than a 8pin comparator and 2 FETs.
 

thatoneguy

Joined Feb 19, 2009
6,359
Telephones sound awful, especially cell phones. ALL the high frequency sillibants of speech are gone so it is difficult to understand what is being said.

ee de'ingner' o' 'ele'one' 'ould be 'o'. (Just vowels.)
Translation to wideband audio: The designers of telephones should be shot.
I can't hear the difference between an f and and s anymore these days anyway. I say "Huh?" and "What?" and "Say again?" a lot, especially if there's background noise.

I'm sure the amp Ron H made would sound fine to me.

As far as keeping Vdd well regulated, why not have the input signal change the duty cycle of a switching supply for the rails, and end up with full Class G (or whatever letter they call it since it isn't a standard yet)? Extremely high efficiency, as you are only switching enough current to reproduce the sound, and aren't holding ±30V rails when listening to quiet music, dynamic headroom would be extreme, as the supply could boost to ±60V for transients.
 

tom66

Joined May 9, 2009
2,595
As far as keeping Vdd well regulated, why not have the input signal change the duty cycle of a switching supply for the rails, and end up with full Class G (or whatever letter they call it since it isn't a standard yet)? Extremely high efficiency, as you are only switching enough current to reproduce the sound, and aren't holding ±30V rails when listening to quiet music, dynamic headroom would be extreme, as the supply could boost to ±60V for transients.
You'll need a delay line so you know when to ramp up the voltage and a supply capable of charging the capacitors rapidly (<10ms), but it definitely could work.
 

THE_RB

Joined Feb 11, 2008
5,438
...
As far as keeping Vdd well regulated, why not have the input signal change the duty cycle of a switching supply for the rails, and end up with full Class G (or whatever letter they call it since it isn't a standard yet)? Extremely high efficiency, as you are only switching enough current to reproduce the sound, and aren't holding ±30V rails when listening to quiet music, dynamic headroom would be extreme, as the supply could boost to ±60V for transients.
You would get the same effect from something as simple as employing closed loop feedback at the amp output, which adjusts the PWM so the voltage at the speaker is the same as the input signal.

So the PWM would self-adjust and compensate for any PSU ripple, PSU sag etc (as it does in an analogue amp). That's how I would do it, but Bill said this was for a minimum parts count type "educational" design?
 

Audioguru

Joined Dec 20, 2007
11,248
Philips made a class-G car amplifier. It had a built-in voltage doubler for 4 times the power of an ordinary amplifier circuit when the music is loud then it cut back to the normal 13.2V during quiet sounds. They discontinued it maybe because it switched too slowly and the switch to high level was heard.
 

tom66

Joined May 9, 2009
2,595
You would get the same effect from something as simple as employing closed loop feedback at the amp output, which adjusts the PWM so the voltage at the speaker is the same as the input signal.

So the PWM would self-adjust and compensate for any PSU ripple, PSU sag etc (as it does in an analogue amp). That's how I would do it, but Bill said this was for a minimum parts count type "educational" design?
The best way I can think of doing this is to use a regulated supply for the audio but power the PWM oscillator off the unregulated supply. A 555 timer will follow the change in supply voltage. It wouldn't eliminate all noise, though.
 

THE_RB

Joined Feb 11, 2008
5,438
That would be a simple way to do it for sure! You'd have to make sure the ramp generator matches (tracks) the supply changes accurately.

I think it's up to how much complexity Bill wants to have in the circuit?
 

Thread Starter

Wendy

Joined Mar 24, 2008
23,415
I am reviving this thread. I have two designs I'll work on, one for the AAC book, one as a high power real world design. Rather than start another set of threads I'll use this one for a while, the AAC book article will be deliberately crude, but functional.

I must be getting healthier, since this is going to require me in front of the computer for long periods of time.
 

Thread Starter

Wendy

Joined Mar 24, 2008
23,415
OK, I picked this design from one of my earlier posts. I am going to use a CMOS 555 for the triangle generator to eliminate power supply instability, then translate it into a protoboard layout.

 

Thread Starter

Wendy

Joined Mar 24, 2008
23,415
OK, next concept circuit.



The one thing I don't like about it is the connection between U2b and R5. It isn't obvious to a beginner that is strictly to make U2b an inverter, and the audio has noting to do with that gate. The schematic is a bit tight, and it has no real gain other than the hellatious current gain common to all Class D amps.

I'll use pieces from it, but back to the drawing board.
 

Attachments

THE_RB

Joined Feb 11, 2008
5,438
Bill, I'm sure i've seen very simple 555 class-D amps that used the 555's CV input (pin 5?) as the modulator.

So the 555 is both the oscillator and the amp, and it's output is controlled by the CV pin. Then you just need one opamp to increase the audio signal to the right scale to drive the CV input.

Have you considered that approach?

(edit) Also, you don't need the inductor and cap on the output, the speaker inductance and its mechanical inertia will eliminate all PWM frequency and just give the audio frequency.
 
Last edited:

Thread Starter

Wendy

Joined Mar 24, 2008
23,415
Perhaps, but this is a demo as well as a working circuit. It will be interesting to try it with and without the filters to see if the output 555's get hot. There is also a pretty good chance of AM interference that might be removed.

Anytime you involve pin 5 you causing the frequency to change, which removes it from the realm of a pure Class D. There will be an element of FM added.

I'm also trying to show block diagrams. This design is primarily for teaching, and breaks down very cleanly into blocks. Come to think of it, I may also include a block diagram in the Theory of Operation.

Post #43 has some links to some designs like you're talking about.

Back to drawing.
 
Last edited:

Thread Starter

Wendy

Joined Mar 24, 2008
23,415
OK, this is supposed to be a final drawing,baring brain farts.

Using a remote computer while flat on my back does slow things (last surgery did not go well, still recovering). My son loaned me his net book, which I can not type (hunt and peck), then set it up for remote access. I am glad to have it.



Now for a protoboard layout.
 

Attachments

Last edited:
Top