Newbie question: filtering full wave rectified sine waves?

[userx2]

Hi folks

I have not done any real DSP stuff apart from doing some spreadsheet simulations with all sorts of filters and their coefficients.

I now have a possible application for a small dsp.

The input to the chip will be a full wave rectified sinewave (positive going half sines).
The frequency is ~50 Hz (100 if counting the half sines).

These need to be samples with A/D at a rate of at least 5k samples per second.
I then need to low pass filter each half sine but I really need to preserve the sharp edges at the bottom where they join.

In addtion, I need either 0 or 180 or 360 degree phase shift.

Does this sound like a plasible application for a small dsp?

How can I go about implementing / evaluating something like this?

Best regards
X

 

[FvM]

I then need to low pass filter each half sine but I really need to preserve the sharp edges at the bottom where they join.

You better describe what you actually want to achieve.

After low-pass filtering the rectified sine you either get a pure DC voltage, or a DC voltage with superimposed 100 Hz sine, or some 100 Hz harmonics in addition, depending on the filter characteristic.

In any case, the filter will cause a group delay respectively phase shift and in no way "preserve the sharp edges" of the rectified waveform.

 

[userx2]

You better describe what you actually want to achieve.

After low-pass filtering the rectified sine you either get a pure DC voltage, or a DC voltage with superimposed 100 Hz sine, or some 100 Hz harmonics in addition, depending on the filter characteristic.

In any case, the filter will cause a group delay respectively phase shift and in no way "preserve the sharp edges" of the rectified waveform.

I would require the filter to perform as in the attached sketch.
I can tolerate a small phase shift up to 12 degrees.

Perhaps the waves can be re-assembled into a sinewave before / during filtering.

Regards
X

 

[FvM]

The intended operation isn't feasible for a linear filter. Why don't you apply the full wave rectification after filtering?

This can be easily done with a band pass filter, which has zero phase shift at the center frequency.

 

[userx2]

The intended operation isn't feasible for a linear filter. Why don't you apply the full wave rectification after filtering?
QUOTE]

The wave comes from the mains side. (See my other thread in Analog) and I have no solution (yet) to get it through an opto isolator unrectified.

Regards
X

 

[FvM]

You didn't tell much about the application, so I can only guess about the exact requirements. I understand that you want the phase rather exact, which is essentially a digital information (zero crossing pulses). The 50 Hz magnitude is an analog information, but the required accuracy hasn't been told yet.

 

[userx2]

You didn't tell much about the application, so I can only guess about the exact requirements. I understand that you want the phase rather exact, which is essentially a digital information (zero crossing pulses). The 50 Hz magnitude is an analog information, but the required accuracy hasn't been told yet.

Hi Fvm

I have not explained it all in detail but you are correct.
The final output is zero cross pulses. Accuracy is 50us.

Regards
X