Simplest PWM

[Stretchyman]

https://imgur.com/a/MEWH4wa

https://imgur.com/uITBDpz


Sure there are simpler curcuits but have been using this and minor variations (far better hald bridge drivers but the IR is fine) for over 10 years with no issues.

200V FETs used here, I do like the 4227's, Low Qg, Low Rds(on) and you can push 1KW+ thru' a pair. If you're only modding 50W then use the T0220 version (vs TO247's)

Nifty circuit, not my design but have added a few features over the years.


                                                                Explanation of Operation;

Audio (1V p-p max) is converted into a PWM stream by U2 LTC6992, Q2 acts as invertor as U3 needs + & - PWM signals.

R1 supplies DC to the MOD I/P of Q2 enabling the PWM duty cycle to be set (50%).

Low Power (active High) pull down R3 which reduces the duty cycle enabling low power.
(This is best used with a latch or micro to detect this at startup and ONLY enabling the PWM if it starts up in low power).

Pulling the enable pin low activates U3 driving the half bridge FETs, Q4 & Q5 & also activates the RELAY O/P via Q3. This will connect the 48V to the main supply and isolate it when not enabled.
FUSE ‘48V from PIC’ with 100mA and the main supply (via relay) with something sensible.
The DC-DC U4 is 600mA O/P, Vin 18-72V for 12V O/P, saves the need for a separate supply for the FET drivers on the RF and any other cicuitry that needs 12V.

Butterworth filter values for L1, L2 etc must be calculated for the impedance of the circuit it’s driving and with a sensible LPF action. This just needs to remove the PWM clock (150K - 500K) and has nothing to do with restricting the audio B/W. This MUST be carried out before U2.

[Charlie_Dont_Surf]

You are using, of course, a classical push-pull Voltage-mode Class-D source. Here's a simulation model for something similar from a few years ago.

I use 13.5 V high voltage here because at the time I was trying to avoid the use of a separate HV inverter, hence the load represented as a 0.75 Ohm resistor. I used a TI UCC5350 dual gate driver, simply because there was already a model and symbol for it in the schematic capture tool, I use and Nexperia PSMN019-100YL transistors but many N-channel FETs will work here. C46/R60, C47/R61 and C48/R62 are snubbers to get rid of the usual ringing.

I use a Bessel filter here because I want a smoother group delay (the rate of phase change with frequency) than your Chebychev filter, though I fully admit that probably no one is going to notice the difference between a Bessel, a Chebyshev or a Butterworth filter at HF/shortwave. The filter response is probably too restrictive like this as there is some rounding of the edges with 1 KHz pulses. (See below.) The snubbers probably don't help in this regard, as their corner frequency is pretty low.

I should get rid of C45 and rejigger this since I would not want that there in the real application. The intent is that the last two elements in the output filter (one C and one L) would physically be on the RF board, where the L would be the RF choke. The remainder of the elements would be on the separate PWM board.

My notes say that I had a lot of problems getting this going in simulation with a 1 KHz source. Not completely surprising to me. I think that I tweaked this later; I just can't find the revised circuit model right now.

[pattymorr]

Great Share,

I would like to reproduce your Simplest PWM and end up with a few questions. The schematic is missing Q1 reference (guess is 2n7002?)  and in general all component values. The PCB layout is for another version of the schematic as well as  it shows a 3906 transistor.

In terms of DC loading impedance, I am looking to have some 3.4Ohm , What would be the Range of DC Load to this PWM?

Thanks in advance for any information

Sincerely,

Patty Morr

[Stretchyman]

Hi Patty, I can send you a PCB, have plenty.

2907/3906 same thing, just general purpose PNP.

The circuit will drive any load just adjust the filters value to match 1R i/p and 3R4  o/p.

Please PM me.

Regards

Stretchy