Re: [CAD_CAM_EDM_DRO] Filters, and a ? for Mariss was Re: Rotor inertia.

ballendo wrote:

>Hello,
>
>Poles in a filter relate to the slope of the "filtered" output. A
>single pole filter has a 3db drop off for a halving (or doubling) of
>the input frequency from the filters' designed "center" freq.
>
>2 poles is 6 db slope, and 4pole is a 12db attenuation slope. In
>other words, the more poles a filter has, the more strongly
>frequencies outside the designed "center" frequency will be rejected
>from the output.
>
>As Jon wrote, one pro is that filters "slow things down". This means
>that fast changing things like noise are rejected, whilst the good
>stuff gets through. There was a recent thread on the xylotex yahoo
>group about using a simple r/c filter in the step line to prevent
>spindle motor noise from falsely stepping the drive. There was enough
>technical (simply worded, and presented) over the course of the
>thread to be worthwhile if you wish to understand a bit about how and
>why filters might be used...
>
>As to a filter improving a gecko; I'd bet Mariss has already covered
>that base!
>
>
>

At $114, I think output filters are totally out of the question on Gecko
drives.
They are the single most expensive set of components on my PWM
servo amps (4 $12 toroid inductors). I used 2 poles (and it is a
differential
filter, therefore 4 inductors) to get the 100 KHz PWM waveform down
below 1 V with an 80 V DC supply. This amount of phase shift
can actually make the servo loop unstable, so the amp has to have
the right loop compensation to work correctly. The opto-isolators
on the Gecko's inputs serve somewhat as a filter on the step inputs.
The encoder inputs on the Gecko 320 are NOT filtered, and in some
systems that can be a problem.

Jon