PDM filters

[WA1GFZ]

Steve,
The TI samples I just received will do 1, 2 or 4 channels. More than two sounds like a lot of hardware. The only  advantage I see of multiple channels is running at a lower frequency for maybe better control near 100 % negative modulation. fc

[steve_qix]

Steve,
The TI samples I just received will do 1, 2 or 4 channels. More than two sounds like a lot of hardware. The only  advantage I see of multiple channels is running at a lower frequency for maybe better control near 100 % negative modulation. fc

Will the TI samples to 4 channels, each to a 100% duty cycle?  If so, that's a good part!  What is the part number?

So far, the multi-channel PWM chips I've seen don't go to 100% duty cycle.  These may be better!

[WA1GFZ]

Steve the part number is TPS40090 but the maximum duty cycle is 87.5%.
Look at the whole system. Your power supply will have a good amount of ripple on it so going to 100% just puts you into the ripple. 87.5% doesn't give you enough  peak output voltage then just raise the unregulated voltage a bit to compensate voltage you need.
I would do current transformers on each phase to provide feedback to the controller since it is a low voltage part. This would also allow the use of slope compensation if necessary.
The error amp is faster than the 25701 so suspect the comparator is up to the job. The internal op amp would make a nice balanced audio interface ahead of the error amp.
I found some app notes on the TI web site

[W1DAN]

Steve:

I am interested in your bi-phase PDM modulator more...

Why do you need two PDM generator chips to create the bi-phase switching pulses? I would think you could take a single PDM generator chip's output and then create an out of phase pulse by using an RC network and a comparator.

I do not have much time these days, but would like to do this someday.

Thanks,
Dan
W1DAN

[WA1GFZ]

Dan,
Both outputs can overlap to 100% duty cycle or not and they begin 180 degrees out of phase. Look up interleaved buck regulators to see some theory and app notes. This method is commonly used in low voltage high current power supplies to reduce ripple. The chip I posted will do up to 4 phase.
BTW, In current mode feedback you need time for the current transformer or inductor to reset so I wouldn't want it to go to full 100% because the feedback would go unstable.
100% would only work in an open loop system.

[steve_qix]

Steve:

I am interested in your bi-phase PDM modulator more...

Why do you need two PDM generator chips to create the bi-phase switching pulses? I would think you could take a single PDM generator chip's output and then create an out of phase pulse by using an RC network and a comparator.

I do not have much time these days, but would like to do this someday.

Thanks,
Dan
W1DAN

hi Dan,

I don't know if an R-C network and a comparator will quite do it, although it might.  With this two-phase system, the pulse trains have to be IDENTICAL, or you will end up with the original switching frequency showing up in the filter.  And of course, any nonlinearity in triangle (or input to any comparator) will cause major distortion.

You don't actually need THAT many more components to do this.  You need an additional driver and associated components.  I'm using the floating charge pump system I came up with for the driver.

By using two PWM generators, you get, essentially, double the resolution that you would otherwise get at the ripple frequency.  If you ran a PWM generator at 240kHz,  the operational bandwidth of the entire PWM generator has to be very high (probably above 10mHz to resolve a 2% duty cycle pulse, without integration).  On the other hand, if you run two PWM generators at 120kHz, the whole system's bandwidth requirement is 1/2 that of the 240kHz system (or the resolution is significantly improved), and you still have the advantage of a 240kHz ripple frequency becuase you are combining the outputs (180 degrees out of phase with each other).

Interesting system, for sure!  I wouldn't bother with a small-ish transmitter.  The added complexity is not worth it.

Regards,

Steve

[W1DAN]

Hi Steve, Frank and all:

I did look at some of the info online and see they use two PDM generators locked together. The results seem to be similar to a three phase power supply rectifier and filter. So I guess there is some reason not to just delay the output of one generator to create a second out of phase pulse train.

Thanks for the info....

73,
Dan
W1DAN