# Generating 500 kHz sine wave using modified PWM

I am trying to generate 500 kHz sine wave of varying amplitude using full bridge transformer isolated DC/AC converter.

I decided to use a sine wave table to change the PWM duty cycle. More the values per cycle better the sine wave output. For 256 values per half sine wave, we need to update once every 3.9 ns which means MCU needs to run at 256 MHz rate at least.

Another constraint is that Mosfet driver needs to turn on/off the Mosfet really fast under 3.9 ns. Mosfet driver tc427 need at least 30 ns to turn on/off the MOSFET.

I am really confused on how to go about overcoming the constraints. Any suggestions please?

• What power output are you hoping to deliver to the load? What is the load? Have you considered that an LC filter at the output can offer a lot in turning a rough sinewave into a much smoother one? Commented Feb 16, 2018 at 13:49
• Power output delivered to load is 500 khz sine wave AC output. There would be LC filter at the output. What should be the filter cutoff frequency to produce sine wave output? Load would be some resistance value up to 4000 ohm. Commented Feb 16, 2018 at 14:05
• What power out in watts? How will the load of 4000 ohm change? Could it be much smaller? What voltage is your DC bus? Commented Feb 16, 2018 at 14:10
• To repeat what Andy asked twice: Max. Power in Watt? It's not hard to calculate for you, but we don't have the insight into your problem to do it for you. Commented Feb 16, 2018 at 14:23
• 200 watts load with a sinewave peaking at 100 volts means an RMS voltage of 70.7 volts and this, in turn means a current of 2.828 amps - you have to be clearer about what you want. At the moment you appear to be implying that the load could be a simple (but high power) 25 ohm resistor and nothing like 4000 ohms. Commented Feb 16, 2018 at 14:30

From a simple two stage LC filter you could get a response like this with just a basic square wave input at 500 kHz and no PWM and no sine look-up table: -

If you had some minimal attempt at PWM (changing duty every quarter of a cycle or 500 ns) you would get an even cleaner output and/or require less filtering. But you have to define load impedances across the whole range for this to be successful. At the moment I have tailored it to a 4000 ohm load but if you require loads down at 1000 ohm or less then the filters change.

• Now, I finally understood why you were very much interested in load impedance. Yes, my load changes. I guess the other answer of using Class C amplifier would be the best way to go. Commented Feb 16, 2018 at 14:40
• Class C amplifiers use LC filters to shape the frequency response and these are just as suceptible to load changes as any other LC filter. No, if you have a H bridge output stick with that because you can at least experiment with PWM schemes more effectively. I've used half bridge drivers for doing the same but at 600 kHz producing up to 200 volt p-p output so I think my experience might be relevant! Commented Feb 16, 2018 at 14:42
• Do you think if I take output voltage feedback and change the DC bus voltage according to the changes in load ?. DC bus voltage is from a SMPS whose output voltage could be varied. Commented Feb 16, 2018 at 14:52
• This is functionality that only you know about. I don't know why you need to do this because I don't know what your overall aims are. I only see your question. If you require feedback you have to be cautious becayse the phase change produced by filtering could make badly designed feedback systems unstable. But, yes, you can do this sort of thing. Commented Feb 16, 2018 at 15:01
• Are you worried the load value changes impact filter Q?. Is there any other way to negate the changes in load values?. Commented Feb 16, 2018 at 15:01

Trying to get 500 kHz sine from PWM seems quite a stretch.

If the frequency is fixed, then look into something resonant, like a class C output stage. Basically, you start with pulses or a square wave, then use resonance and passive filtering to get rid of the harmonics to make a sine. This is how radio transmitters work, for example. They have to produce even higher frequencies at high power levels, and are allowed very little harmonic content.

• Yes, my frequency is fixed. Any links/resources to design the same?. Commented Feb 16, 2018 at 14:10
• @ren: Look into "class C" amplifier or output stage. Commented Feb 16, 2018 at 14:22