Re: PID tuning for air roasters

Hi Andy,

I don't have any background in this stuff and am reasoning from a 10
page explanation of PID tuning contained in a manual for an Omega
controller that I am not using. Not because it's definitive but
because it's the easiest to understand (at least for me) that I've
found. If you're curious the doc is available here:
http://www.omega.com/manuals/manualpdf/M2828.pdf

The discussion I'm looking at starts around page 58 of the *.pdf

Anyway, it describes "Integral Time" as follows:

"With the proportional band alone, the process tends to reach
equilibrium at some point away from the main setpoint. This offset is
due to the difference between the output needed to maintain setpoint
and the output of the porportional band at setpoint. In the case of
the controller where the proportional band is equidistant from the
main setpoint, the output is around 50%. If anything more or less than
50% output is required to maintain setpoint, an offset error will
occur. Integral action eliminates this offset."

The thing I am having trouble with is that when roasting beans on an
ever increasing ramp there is no process that ever needs to reach
"equilibrium". And so I think I just need to set the proportional band
so that it maintains the ramp without overshooting. Presumably
different proportional bands for different ramp rates and at different
stages of roast, but that is manageable. If the controller overshoots
or undershoots for a second the optimal proportional band should
compensate IMHO, and it seems to me that the Integral calculation
would come into play too slowly to make a difference.

So, questions:


Assuming P is optimal (and I really mean optimal) for each ramp
segment will the integral calculation make any difference when there
is no setpoint that has to reach "equilibrium" over multiple seconds?
I see how Integral can make a big difference in the big picture, but
what if there is no big picture, only a succession of small pictures?

Ian's results with only proportional control look very nice, could
they be improved with Integral control?

I may be completely wrong about this of course, but if I am would
really appreciate understanding why.

Thanks!
Paul



On Tue, 14 Mar 2006 03:28:44 GMT, Andy Schecter
<schecter@xxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:



One of the aims of PID roasting is to make your roasts more reproducible.
Through proper tuning and design, you should be able to make the inlet
temperature follow the same temperature curve time after time.

The Integral ('I') setting gives you the extra oomph to reduce the temperature
error and bring the process as close as possible to setpoint. Whether your
setpoint is constant or changing, I can't see any reason why you'd choose NOT
to use it.
.

Relevant Pages

  • Re: PID tuning for air roasters
    ... Here is a basic math interpretation of how each portion works. ... If the tc and setpoint are nearly the same, ... Does the controller have ... entirely I got much faster response during the initial ramp. ...
    (alt.coffee)
  • Re: PID tuning for air roasters
    ... During a fast heatup, the derivative would be subtracting from the output ... If the tc and setpoint are nearly the same, ... Does the controller have ... entirely I got much faster response during the initial ramp. ...
    (alt.coffee)
  • Re: I-Roast 2 - whats with all the fuss?
    ... Then the heat controller, a PID ... actual temperature from the setpoint. ... and far the temp is from setpoint and will continue to build if the temp ... do true programmability. ...
    (alt.coffee)
  • Re: PI: Integrator implementation
    ... the 'I' part of the following controller. ... system reaches the setpoint. ... This is because the integrator remembers ... inside a conditional statement which tests for the absoulte magnitude ...
    (sci.electronics.design)
  • Re: PID question
    ... using this value to provide current limiting, via a PID algorithm ... controlling the set voltage. ... One way to accomplish this is to have two PID controls, a fast one that controls current, and a slow one than controls voltage, by modifying the setpoint for the current loop. ... This is called a cascade controller (a single output controlled by two measured variables, ...
    (sci.electronics.design)
Loading