Query Regarding RMS to DC converter

[sabu31]

Can AD637 RMS to DC converter be used with a single supply operation. Basically I want to calculate the average of a square wave. Are there any simpler solutions available. The square wave varies from 0 to 3.3V.

 

[KlausST]

Hi,

"RMS" and "average" are two different things with different results.

Example:
Lets say you have a 50% duty cycle 0V / 1V square wave as input..
* the output of an "RMS including DC" gives 0.707V
* the output of an "average" gives 0.500V
* the output of an "RMS without DC" gives 0.500V (this often what you see on true RMS DVMs)

Lets say you have a 50% duty cycle -1V / 1V square wave as input, then...
* the output of an "RMS including DC" gives 1.000V
* the output of an "average" gives 0.000V
* the output of an "RMS without DC" gives 1.000V (this often what you see on true RMS DVMs)

--> Decide what output value you expect.

Klaus

 

[sabu31]

Thanks Klaus for the reply.
I need to have a DC signal proportional to the width of the square wave. The input waveform varies from 0 to 3.3V.
I have two queries.

a)With Respect to AD637, can we use it with unipolar supply (ie ground and -Vs connected to ground itself).
b) Since I don't need any exact calculation, are there any simple circuits which can give DC (average) equivalent value of an input square wave signal. The frequency of signal is 50Hz.

 

[betwixt]

Seems like you are measuring the pulse width ratio rather then RMS. If it is a 0V to 3.3V 'logic level' signal, treating it as a digital signal and using a simple MCU would probably be easier. You can use a timer to measure frequency (if needed) and the width of the pulses within each cycle. This method has the advantage of being usable with any reasonable pulse rate. If you do it with an analog averaging circuit you have to choose different values for each frequency or tolerate ripple in the measurement. Using an RMS converter works but is more expensive and still frequency dependent.

Brian.

 

[KlausST]

Hi,

Thanks Klaus for the reply.
I need to have a DC signal proportional to the width of the square wave. The input waveform varies from 0 to 3.3V.
I have two queries.

a)With Respect to AD637, can we use it with unipolar supply (ie ground and -Vs connected to ground itself).
b) Since I don't need any exact calculation, are there any simple circuits which can give DC (average) equivalent value of an input square wave signal. The frequency of signal is 50Hz.

In correct terminology you want (correct me if I´m wrong)
* to measure the duty cycle of a 0V/3.3V pulse wave with a frequency of 50Hz.

If so: "RMS" is a "no go", because it is not linear with duty cycle.
--> so, forget about AD537.

If you need an analog value: then there are some things to consider
* a low pass filter will get the average of the input signal and is linear with duty cycle. However, it relies on accurate and stable 0V and 3.3V levels (which usually depend on VCC and load current and thus are not very precise). And a low pass filter has two other known problems: 1) The output will always carry some amount of remaining voltage ripple.
2) the output takes time to settle. and since it approximates the target voltage it never will be perfect.

So you need to consider
* how much ripple voltage is allowed
* how much time is it allowed to take until the voltage error is below your defined level.

****
If you use a microcontroller, I recommend (likeBrian) not to use any analog circuit at all, just use the "input capture" feature to measure timings. And do the duty cycle calculation (and filtering) in software.

Klaus

 

[Easy peasy]

An RC filter ( ideally 2 pole ) will give you DC level proportional to the PWM for fixed voltage height of PWM input

0V - 0 PWM, 5V = 100% PWM for 5V high PWM.

now you need to talk about required response times