JeffM wrote:
What you need
to turn a pulse train into a something resembling DC
is an integrator. TV vertical sections use those gizmos.

Dave Plowman (News) wrote:
Thanks, Jeff. Can you clarify what's needed spec wise

As stated:
You need to put some numbers to this before proceeding.

Got an oscope?

You need an RC time constant that will integrate this a bit.
In a TV, the integrator must combine the closely-spaced stuff
during the vertical sync pulse interval to make one fat pulse.
Your task is similar.

If you make the time constant *too* long,
the intervals between cylinders will be joined up;
**that** is what you need to figure out how to avoid.

Surely you've seen pictures of how a power supply filter
slurs one (sinusoidal) peak into the next.
Similar deal there.

You need to characterize what you have
and/or do some trial-and-error runs.
http://google.com/search?q=filter+ca...+63-percent%22
http://google.com/search?q=filter+ca...3.2-percent%22
http://google.com/images?q=filter+ca...3.2-percent%22

- I'm more of an audio sort of person. ;-)

RC filters are common in audio as well.
I already mentioned power supply filters
and their similarity to this application.