I don't really know what to say but I need a prod up the behind to actually get down and do this..... please give me one.
I'll be posting the design info at...
http://groups.yahoo.com/group/classdamplifier/
The rest is open to discussion.
I'm going to be doing a clocked thing using PWM with multiple stages operated in phase shifted parallel, if that makes sense.
I've got LTspice and some other bits as well as schematic capture and PCB design software (I can release Gerbers). I'll try to present the documentation in PDF format
Ask some questions and we'll see how it goes. All and any input appreciated.
Thanks
DNA
I'll be posting the design info at...
http://groups.yahoo.com/group/classdamplifier/
The rest is open to discussion.
I'm going to be doing a clocked thing using PWM with multiple stages operated in phase shifted parallel, if that makes sense.
I've got LTspice and some other bits as well as schematic capture and PCB design software (I can release Gerbers). I'll try to present the documentation in PDF format
Ask some questions and we'll see how it goes. All and any input appreciated.
Thanks
DNA
I played around with a multi phase design for a while but always found that the changing output impedance resulted in distortion folding down into the audio band. To see this effect in simulation I had to add non-linear inductors and model speaker back EMF effects.
I'm sure there are solutions for these problems but I gave up on it because there did not seem to be any advantage and there are a lot of added costs.
ClassDunce
I'm sure there are solutions for these problems but I gave up on it because there did not seem to be any advantage and there are a lot of added costs.
ClassDunce
ClassDunce said:
I'm not sure I understand what you are saying.
However, if you want to do a 'multi phase' design you have to bear in mind that the filter inductors in each stage integrate the volt seconds applied to them to give a final output current.
If you simply apply the same, but phase shifted, pwm signal to each stage then it is likely that one of the stages will 'hog' the output current because off differential delays and switching times.
If you add a current mode control loop to each stage then you can force them to share the current.
DNA
Long time, haven't done much
Got back to it and started writing some of the theory behind what I'm doing.
Done a bit of design work to try and get things together and then modelled things in spice using 'proper' bits and pieces.
The piccy is an FFT of four stages operated in parallel with current sharing. The base modulator frequency is 200KHz for an effective 800KHz.
You can see a bit of a lump at 200KHz with the rest of the switching hash up at 800KHz. These components are 40dB down or about 10mV rms(?). That's one of the reasons for doing it this way.
This is 1.5Vpp in for 60Vpp out at 10KHz off 80V rails. Mosfets are IRF640/9640.
If the theory, and small signal model, agree with the switching model then the -3dB bandwidth is about 200KHz. I'd be bothered about current slew rate limiting.
As is always the case with spice.... don't believe everything you see. I'm fairly confident that things are not too wide of the mark.
DNA
Got back to it and started writing some of the theory behind what I'm doing.
Done a bit of design work to try and get things together and then modelled things in spice using 'proper' bits and pieces.
The piccy is an FFT of four stages operated in parallel with current sharing. The base modulator frequency is 200KHz for an effective 800KHz.
You can see a bit of a lump at 200KHz with the rest of the switching hash up at 800KHz. These components are 40dB down or about 10mV rms(?). That's one of the reasons for doing it this way.
This is 1.5Vpp in for 60Vpp out at 10KHz off 80V rails. Mosfets are IRF640/9640.
If the theory, and small signal model, agree with the switching model then the -3dB bandwidth is about 200KHz. I'd be bothered about current slew rate limiting.
As is always the case with spice.... don't believe everything you see. I'm fairly confident that things are not too wide of the mark.
DNA
Attachments
This is the output voltage at start up along with the stage currents.
Each stage here is modelled the same as all the others but D gets caught first and is then brought back into line with the others by the control loops.
The spikey bits are body source diode reverse recovery.
DNA
Each stage here is modelled the same as all the others but D gets caught first and is then brought back into line with the others by the control loops.
The spikey bits are body source diode reverse recovery.
DNA
Attachments
Hi,
may I suggest that you take a serious look at Bruno's US pat no. 6,803,816. His method of using autotransformers for automatic current sharing between phases of multiphase system is a very good idea. I will try to implement it in multiphase nonsynchronous buck converer where I expect it will alow me to reduce volume of magnetics components by more than half.
I suggest you do some simulations. Coupling coefficient can be as high as 0.99997 (that is what I have measured on my prototypes). Maybe you will find that you do not need output filter at all.
Best regards,
Jaka Racman
may I suggest that you take a serious look at Bruno's US pat no. 6,803,816. His method of using autotransformers for automatic current sharing between phases of multiphase system is a very good idea. I will try to implement it in multiphase nonsynchronous buck converer where I expect it will alow me to reduce volume of magnetics components by more than half.
I suggest you do some simulations. Coupling coefficient can be as high as 0.99997 (that is what I have measured on my prototypes). Maybe you will find that you do not need output filter at all.
Best regards,
Jaka Racman
Jaka Racman said:
I went and had a look, perhaps I should go and have another look.
Patents make hard reading. Didn't see the autotransformer thing.
Just for the fun of it here's what happens if I take out the filter capacitors.
I was surprised and I'm not so sure I want to think about it....
DNA
Attachments
Hi,
attached is picture from patent showing how phase summation circuit looks like. I added dots to indicate winding polarity. On the right side of the picture you see combined pwm signal from four phases. As you can see there is hardly any filtering necessary.
Maybe you can ask Bruno to mail you his AES article. It is much more descriptive than patent.
BTW, such autotransformer combining is nothing new in other fields. In my lab I have 7kW single phase variac which consists of two physical variacs on the same axis and whose outputs are combined with such autotransformer. I also have 1978 Motorola RF handbook where autotransformers are used to combine RF power of several smaller RF amplifiers.
Best regards,
Jaka Racman
attached is picture from patent showing how phase summation circuit looks like. I added dots to indicate winding polarity. On the right side of the picture you see combined pwm signal from four phases. As you can see there is hardly any filtering necessary.
Maybe you can ask Bruno to mail you his AES article. It is much more descriptive than patent.
BTW, such autotransformer combining is nothing new in other fields. In my lab I have 7kW single phase variac which consists of two physical variacs on the same axis and whose outputs are combined with such autotransformer. I also have 1978 Motorola RF handbook where autotransformers are used to combine RF power of several smaller RF amplifiers.
Best regards,
Jaka Racman
Attachments
Oh, sort of. Well, not really
One of the things you get is ripple cancellation.
Take two phases 180 degrees out of phase and the zero voltage sum has zero ripple.
Take four phases 90 degrees out of phase and you get zero ripple at 0V out and zero ripple at +/- half the supply voltage out and so on.
Perhaps when you are talking about 'autotransformers' you are referring to what I'd call coupled inductors.
If you look at a full bridge circuit with a differential output filter then it should be possible to wind the two filter inductors on the same core.
Ideally they experience the same volt-seconds so it works.
DNA
One of the things you get is ripple cancellation.
Take two phases 180 degrees out of phase and the zero voltage sum has zero ripple.
Take four phases 90 degrees out of phase and you get zero ripple at 0V out and zero ripple at +/- half the supply voltage out and so on.
Perhaps when you are talking about 'autotransformers' you are referring to what I'd call coupled inductors.
If you look at a full bridge circuit with a differential output filter then it should be possible to wind the two filter inductors on the same core.
Ideally they experience the same volt-seconds so it works.
DNA
Hi,
you might also call them coupled inductors, but I prefer term autotransformer (and if I recall correctly Bruno also) because they carry no net DC current so they use high permeability core. You also dimension them like transformers ( Volt second product). If you have time and will, do a simulation, you will find that phase currents do not look like you expect them to be (classic ripple cancelation thing). Use small LC filter at the output for first simulation.
Best regards,
Jaka Racman
you might also call them coupled inductors, but I prefer term autotransformer (and if I recall correctly Bruno also) because they carry no net DC current so they use high permeability core. You also dimension them like transformers ( Volt second product). If you have time and will, do a simulation, you will find that phase currents do not look like you expect them to be (classic ripple cancelation thing). Use small LC filter at the output for first simulation.
Best regards,
Jaka Racman
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.