[chbot] Filter

[Charles Manning]

If you have data with spikes in it then a median filter is a very good way
to get rid of those spikes. It is better than an IIR or FIR which "cook in"
the error.

basic algorithm for a 5 point median filter:

Take the five points: x0..x4.
Sort them into y0..y4
Select y2 (the median value).

You can then apply an IIR/FIR to the result.

It is pointless going the over way (ie. IIR/FIR) then median filtering
since the FIR/IIR will cook in the noise and the median filter can't take
it out.


On Tue, Oct 23, 2018 at 8:47 AM Hartley, Robin (Nokia - NZ/Wellington) <
robin.hartley at nokia.com> wrote:

> Hi Wallace,
>
> Something to keep in mind when measuring cam profiles is the shape of the
> cam follower. Flat bottom followers result in a very different effective
> profile compared with roller followers. Some followers that appear flat
> actually have a profile to them.
>
> When doing this the old fashion way with a DTI we would do it in situ
> using the followers, and we used to vibrate the bench at each measurement
> point to get the pointer to settle. When running the DTI directly on the
> cam drag on the end of the pointer causes stiction in the bearing that can
> give this semi cyclic error. A bit of vibration may help
>
> For selecting between a set of cams I'd just use Excels curve fitting to
> smooth the curve. And then do the velocity calcs on the resultant curve.
>
> Regards,
> Robin
>
> -----Original Message-----
> From: Chchrobotics <chchrobotics-bounces at lists.ourshack.com> On Behalf Of
> Marshland Engineering
> Sent: Saturday, October 20, 2018 10:30 AM
> To: chchrobotics at lists.ourshack.com
> Subject: Re: [chbot] Filter
>
> After quite a few chats and emails, the problem has become clearer. In
> some way it may not really a filter issue. For a 4 stroke engine cam, there
> are basically 4 components, base circle, up ramp, over the nose and down
> ramp back to the base circle.
>
> So we have 0 velocity, acceleration to the up ramp, constant velocity,
> deceleration, stop (over the nose) acceleration, down ramp, deceleration
> and back to 0.
>
> The figures to look for are the acceleration. If it is too high then the
> valve/valve strings etc loose contact with the cam.
>
> The figure for acceleration (from memory) is 0.008" per degree or .2
> mm/degree.  This is what I want to measure/compare.
>
> Here is some info on it.
> http://www.profblairandassociates.com/GPB_4stHEAD_ValveLiftDesign_hmb.html
>
> Basically I have several different cams and want to measure them all and
> choose the one with the most gentle profile.
>
> Cheers Wallace.
>
>
> _______________________________________________
> Chchrobotics mailing list Chchrobotics at lists.ourshack.com
> https://lists.ourshack.com/mailman/listinfo/chchrobotics
> Mail Archives: http://lists.ourshack.com/pipermail/chchrobotics/
> Meetings usually 3rd Monday each month. See http://kiwibots.org for
> venue, directions and dates.
> When replying, please edit your Subject line to reflect new subjects.
> _______________________________________________
> Chchrobotics mailing list Chchrobotics at lists.ourshack.com
> https://lists.ourshack.com/mailman/listinfo/chchrobotics
> Mail Archives: http://lists.ourshack.com/pipermail/chchrobotics/
> Meetings usually 3rd Monday each month. See http://kiwibots.org for
> venue, directions and dates.
> When replying, please edit your Subject line to reflect new subjects.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ourshack.com/pipermail/chchrobotics/attachments/20181023/59dc1a5d/attachment.html>