RE: ePID deadband on output

["Mooney, Tim M." <mooney@aps.anl.gov>]

By the way, this is Mark Rivers' record, in the synApps std module.

I was playing with the algorithm using calcouts and the StripTool, and I think it migt be frustrating for end users, because the effect of AIIR changes with the rate at which the algorithm is run.  For example, the phase lag in FERR doubles when you halve the rate at which new readings are processed.  Maybe this is desirable from some viewpoints, but it will confuse people.  The epid record includes the time between readings in its current algorithm (see http://www.aps.anl.gov/bcda/synApps/std/epidRecord.html#Feedback%20Parameters), and I think many end users would like this also in the proposed filter.

Tim Mooney (mooney@aps.anl.gov) (630)252-5417
Software Services Group (www.aps.anl.gov)
Advanced Photon Source, Argonne National Lab


________________________________________
From: tech-talk-bounces@aps.anl.gov [tech-talk-bounces@aps.anl.gov] on behalf of Bruce Hill [bhill@slac.stanford.edu]
Sent: Tuesday, August 26, 2014 2:38 AM
To: Dalesio, Leo; fong@triumf.ca
Cc: Core-Talk; tech-talk@aps.anl.gov
Subject: Re: ePID deadband on output

Not my call, but we have been happy with the approach we used as it
gives independent control over the auto-on threshold, the auto-off threshold,
and the error filter factor which determines how sensitive the auto on/off is to
noise and transient conditions such as moving through the target point.

If people like this approach, the fields could look something like this:
AIIR - The auto on/off IIR filter factor
AON - The auto on error limit
AOFF - The auto off error limit
AUTO - Auto mode on/off
FERR - Filtered error

FERR = ERR * (1.0 - AIIR) + FERR * AIIR
if (AUTO && ABS(FERR) <= AOFF)   FBON = 0
if (AUTO && ABS(FERR) >   AON)    FBON = 1

Cheers,
- Bruce

On 8/25/2014 4:43 PM, Dalesio, Leo wrote:
> Did the conversation converge on a solution?