Switching current source for Class A?

[kenpeter]

Originally posted by QED047
Something would surely have to servo the DC to keep it in check? And I'm guessing that that would clutter the circuit beyond recognition.

Nah, just search the tube forum for MJK's variant of the Anti-triode.
Its based upon Waveborn's gyrator load, but an active helper too.
Got incredibly useful self-servo action at DC...

Very few changes to implement... Here's a copy of his diagram:
I believe this one would tend to center out around 658V...

The IGBT in this diagram is doin what your switcher ought to.
To be clear, its drawn upside down to the way yours works...
The plate of his triode would be your MOSFET's source.

[QED047]

Originally posted by kenpeter
[B]

To be clear, its drawn upside down to the way yours works...
The plate of his triode would be your MOSFET's source.

658V? Sorry to be so dumb, but I don't know where to begin trying to analyse this configuration. Can you describe the principle involved?

[QED047]

Thinking on a bit, the problem with the power plots for the abused centre-tap in post#60 may be to do with the way I return the flyback voltage. The recycled energy is a bit toxic when it makes its way out the diode's cathode. Returning it to the PSU and modelling it properly is a bit of a challenge. I'm not seeing anything like it on my breadboard.

Perhaps I should be re-introducing the recycled power into the supply resevoir via an inductor? The flyback voltage goes as high (or low) as you let it, so a capacitor is still needed on the cathode. But it would seem to be a good idea to use an LC combo to pour the volts in gently?

[sawreyrw]

The power in each sense resistor is 1.25 watts, so a 3 watt resistor will not be abused at all. I'm not sure what you mean by 'recycled power', but the cathode of the diode must be connected to a very low inductance point. Otherwise you will have a huge L di/dt spike.

[SQLGuy]

400Hz because you want to use this to control aircraft flight control servos? That sounds a bit too obscure, but 400Hz is standard for use on aircraft.

Cheers,
Paul

[QED047]

Originally posted by sawreyrw
The power in each sense resistor is 1.25 watts, so a 3 watt resistor will not be abused at all. I'm not sure what you mean by 'recycled power', but the cathode of the diode must be connected to a very low inductance point. Otherwise you will have a huge L di/dt spike.

Not real abuse! It was just Ken's colourful way of promoting the idea of splitting the current sense into two parts with the load in the middle. And yes, inductance alone will not do. I was thinking of some kind of safe "holding tank" to return the flyback e.g. capacitor(s) and couple that inductively to the main supply. Should work in theory - the DC will be the same at each end of the coupling inductor.

Originally posted by SQLGuy
400Hz because you want to use this to control aircraft flight control servos? That sounds a bit too obscure, but 400Hz is standard for use on aircraft.

True, but this has a more musical (and whimsical) association. You have to know the right Melody.

[SQLGuy]

Well, 440Hz is a concert A, that used to be standard for orchestral tuning. However, a conductor on another forum mentioned that pitch has been rising for the past few years (i.e. most orchestras tune sharp compared to what was done back in Gershwin's days). I would guess 400Hz to be a G or so, but...?

[QED047]

Originally posted by Eva
Hi QED047. You are going the right way. In fact, a self oscillating class D amplifier is not that different from your approach, it just uses a voltage comparator and a four quadrant switcing mode current source (one MOSFET to each rail). I only like triangle wave modulators for SMPS, not for audio.

Right on! The most "trust" I have is for the simple follower. This is why I shy away from typical Class D. This trust is both in my own engineering competence and in the even more subjective area of listening.

Originally posted by Eva
btw: Some of them actually do class A followers because they can't understand anything more complex This is a nice way to tease them or maybe to tempt them to enter the dark side of electronics muahahaha

I'm not really all that far from that group. However, silly as it may be, so long as the signal path still has a clear line of sight to the load, I don't mind augmenting it with non-linear techniques if they can be used effectively.

Originally posted by Steven
Ronan van der Zee showed a similar idea in his thesis, see chapter 6.6.3 (page 96). You can download the complete thesis here: http://doc.utwente.nl/14149/1/t000000d.pdf
or check this: http://purl.org/utwente/14149.

Steven

Many thanks for the link. I've been looking at the parallel amp with interest. I've re-modelled my reference Class A (follower and linear current sink) along these lines. For anyone unfamiliar with the parallel amp, in one implementation, a conventional output stage capable of sinking and sourcing current is tied to the load via a current sense. This sense point is used to try to keep the current at a minimum by operating a half-bridge on the other side. The two transistor switches in the bridge source or sink current via an inductor at a variable switch rate. These take over the "heavy lifting" duties and the linear amp serves only as a reference signal.

The motive in Ronan van der Zee's thesis was to improve the efficiency of a typical (already quite efficient) class AB amp. The use of a single ended reference appeals to me (as always because of its simplicity) and as I'm not after extreme efficiency, a few tens of watts here are no problem at all. All in all this is a very interesting idea.

Attached is a schematic of an LTspice sim plus some output.

[QED047]

A fair bit of ripple can be seen with this simple circuit...

[QED047]

And here's the sim file for anyone that wants a play (I take it we can only have one attachment per post?)