Simple square waveform to DC converter

[dunya]

Any ideas on how to convert +/- 10V quasi-square waveform to DC (for pwm control) without using RMS to DC chips. 5% accuracy will do.
Thanks

 

[BrianG]

Please give more details on what you actually have as the waveform. Does the +/-10V quasi-square waveform mean it varies in amplitude or what?

 

[dunya]

The quasi-weaveform has a fixed frequency of 60hz but it is going to be a duty cycle controlled based on the amplitude.

 

[BrianG]

Ok, I understand that the frequency is fixed at 60Hz, however, you say "but it is going to be a duty cycle controlled based on the amplitude" . Clearly variable duty cycle is the objective of producing a PWM signal, but what is the "control analogue" which will vary it?

 

[BrianG]

To clarify my last reply - is the square wave just a squared-off sine wave which has equal plus and minus symmetrical components which both vary in amplitude together, or does the square wave vary linearly from minus 10V to plus 10V at 60Hz?

 

[dunya]

It is going to be a reference DC signal that will be used by the PWM to set the duty cycle.

 

[dunya]

I Hope this will clarify things better. The square wave is really symmetrical on the + and - side with slight dead time applied between the + and - portion. The amplitude will change between +/- 11 and +/- 15V based on other factors and the DC will be altered to keep a fixed rms value of the waveform (i.e. 10 VRMS).

 

[BrianG]

I'm still not quite there yet! Do I understand correctly that you are designing a closed loop system where
1) a d.c. control "demand" signal drives the PWM system
2) the resultant 11-15V quasi square wave is processed by the circuit in question to produce an "rms" signal - the whole point of your original question. This is then fed into
3) a comparator system to adjust the PWM to keep
4) the demanded "rms" level constant with load and power fluctuations?

 

[dunya]

Yes, that is what happening.

 

[BrianG]

First thoughts if you are trying to build this closed-loop controller. Do you actaully need it to be referred to a near-rms value? For a closed-loop controller, since you are comparing the amplitude of the square wave against a demand signal, surely you could just use an integrator, or perhaps a sample/hold circuit to grab say the positive-going half of the sqaure wave?. Or have I missed something?