Re: ePID deadband on output

[Bruce Hill <bhill@slac.stanford.edu>]

Hi Matt,
It seems a bit much to have 2 auto on/off methods in one record, but it does provide more options for the end-user.
As Tim and Ken pointed out, the IIR does introduce a phase shift that depends on the sample rate,
so that might be confusing for some.   The PID itself has the same dependency on sample rate, however.
Re the IIR phase shift, since it's not used as part of the feedback loop, but just to check for stability and drift,
it hasn't been an issue for us.

Cheers,
- Bruce

On 8/26/2014 12:41 PM, Pearson, Matthew R. wrote:
> Hi,
>
> Would it make sense to implement both modes, one IIR filter and one time-bound deadband? The IIR approach makes sense: If the error gets small enough, turn off the control, and only turn it on again until the error increases too much. The time-bound approach would deal with the case where the user wants control turned off after N seconds, as long as it's been within the deadband that long (which might be more deterministic in some cases).
>
> Cheers,
> Matt
>
>
> On Aug 26, 2014, at 3:17 PM, Bruce Hill <bhill@slac.stanford.edu> wrote:
>
> > Hi Matt,
> > I used discrete records for this instead of modifying the epid record.
> > If you have time to look into implementing this in epid, that would be great.
> > Thanks!
> > - Bruce
> >
> > On 8/26/2014 7:38 AM, Pearson, Matthew R. wrote:
> > > Hi,
> > >
> > > It sounds like it would work for my application.
> > > If there is no ePID code, I would be willing to implement it in the ePID.
> > >
> > > Cheers,
> > > Matt
> > >
> > >
> > > On Aug 26, 2014, at 8:07 AM, "Dalesio, Leo" <dalesio@bnl.gov> wrote:
> > >
> > > > Hi Bruce.
> > > > Did you make this part of the ePID record? I have not looked at that code for a while. Did you modify the record? Could you send the code? I think it would be a good modification. There are a number of situations where motors should not be dithering.
> > > > Bob