PID controllers (was RE: Holley ecu info)

[Kevin Vannorsdel]

Regarding PID...

Bob Pease also has a decent article on PID in the June 26 1995 Electron=
ic
Design magazine.  It was also published in the Aug 4th 1997 "Best of...=
"
supplement to Electronic Design.

Bob is a VERY interesting guy- anyone interested in Analog Design or ju=
st
wanting to read some interesting commentary should check him out...

He has a website too -   see National Semiconductor...

________________________________________________
  Kevin Vannorsdel     IBM Arm Electronics Development
    408-256-6492                Tie 276-6492     kv at us.ibm.com






owner-diy_efi at coulomb.eng.ohio-state.edu on 09/30/97 12:20:11 PM
Please respond to diy_efi at coulomb.eng.ohio-state.edu @ internet
To: diy_efi at coulomb.eng.ohio-state.edu @ internet
cc:
Subject: Re: PID controllers (was RE: Holley ecu info)


Tom Cloud wrote:
>
> >Hi All
> >
> >       Tom mentioned pid controllers in his last post. Would someone=
 please
info > >me on how they work? Have seen so minor descriptions, but nothi=
ng in
depth. > > > >Thanks Vance > > Vance .... I'll try to make it brief (to=
 keep
from showing my ignorance) > >  - all amplifiers have two inputs (even
"single-ended" op-amps, etc) > >  - one input is compared to the other =
and the
difference is amplified > >  - on single-ended inputs, the "reference" =
is
ground > >  - on a controller, one input is usually called the "referen=
ce" or
set- > point and the other the "error" or feedback > >  - a simple cont=
roller
just multiplies the difference between the two > inputs to yield an out=
put
"proportional" to that difference > >  - a simple proportional controll=
er can
only "see" -- or correct for -- > errors which are greater than the rec=
iprocal
of the gain (ex. with > a gain of 1000, errors of 1/1000 or less are no=
t "seen",
hence not > corrected) > >  - to "close the gap" in the proportional
controller, an integrator > is incorporated.  The small error not corre=
cted by
the proportional > function is continuously added back in over time (in=
tegrated)
until > it grows large enough that the gain of the system can correct i=
t. > >
- the differential part is to keep the system from over-reacting -- > i=
.e., fast
rising errors will cause the gain of the system to be > reduced to keep=
 the
output from going berserk

Ringing of a false signal or referance from a high voltage signal (ie
ignition coil) will drive any opamp up a wall

Steve
>
> That's my thumbnail sketch  ....  anyone care to correct and/
> or add to it ??
>
> Tom Cloud
>
>       Insanity is hereditary, you get it from your kids!



=