Compact Logic PID Loop Deadband

bgtorque · Nov 30, 2013

Hi Guys, I have an existing PLC that has a PID that controls a valve that maintains a constant opposing thrust force (in order to keep bearings happy!) on a driveline when a test article generates thrust. Basically my issue is that the PID works really well except through the deadband, which is about 30-70% of the Cv range. Is there a way I can get the valve to paddle through the deadband nice and quickly (say between 35-65%)?

Thanks for any help.

Lemming · Nov 30, 2013

Hi, I assume its the valve controller that has the DB ?? and the PID lag getting from 30-70% and back is the issue??

If so, I would manipulate the CV of the PID to generate a working reference to the valve that is always actively controlling.. Ie working from your reference of 0-30 & 70 -100

PID output is 0-100%, 0-50% PID = Valve 0-30% & 50-100% PID = valve 70-100%

Hope this helps...

bgtorque · Dec 2, 2013

Lemming said:
Hi, I assume its the valve controller that has the DB ?? and the PID lag getting from 30-70% and back is the issue??

If so, I would manipulate the CV of the PID to generate a working reference to the valve that is always actively controlling.. Ie working from your reference of 0-30 & 70 -100

PID output is 0-100%, 0-50% PID = Valve 0-30% & 50-100% PID = valve 70-100%

Hope this helps...

Thanks for the comments Lemming. I'm not sure how I would achieve this? It seems that you are suggesting inputting two ranges for control, are you suggesting putting two PID's in series to achieve this? This is all pretty new to me, so i'm trying to find my feet.

Thanks again.

Lemming · Dec 2, 2013

Hi

Take the raw PID CV and use it to drive a second manipulated CV. Limit test the the PID CV 0-50%, and if true scale 0-50% = 0-30% for the analogue out... Same for PID 50-100%

May be a good idea to add a third band in the middle to act as a DB, but you can set its width more appropriately... You will have to play with the values to figure out what works best in practice.

Didn't mention it previously, but assume you cant narrow to DB in the valve driver ??

bgtorque · Jan 2, 2014

So if the PID CV is say 37% would the manipulated CV statement be something like this(in Excel speak):

CVmanipulated = if (CVraw>29%&<70%,70%,CVraw)

So rounding up or down to eliminate the deadband, I guess i'd need to expand the statement to take account of if the CV is increasing or decreasing before it gets to the deadband?

Lemming · Jan 2, 2014

No, that not what I was suggesting.... The statement clamps the output to 70% if the PID CV is in the DB. If my previous post didn't make sense to you, I am not sure how else to explain it... What PLC are you using ??

bgtorque · Jan 2, 2014

CompactLogix. Sorry if i've got the wrong end of the stick.

kamenges · Jan 2, 2014

What he is saying is:

Code:

if (PID.CV <= 50) then

   Output = PID.CV * Lower_Deadband_Limit / 50)
else

   Output = Upper_Deadband_Limit + ((PID.CV - 50) * (100 -         Upper_Deadband_Limit) / 50)
end_if

Lemming · Jan 2, 2014

Or if you prefer ladder..

If you do want a (smaller) DB, change the 50 to 48 & 52, and throw a limit test in the middle to force the output to neutral.

Probably worth scaling the PID to reflect direction as well, ie -100 to +100

bgtorque · Jan 3, 2014

Thanks Lemming for taking the time to clarify, that makes it much clearer. I appreciate your help. Happy New Year!

bgtorque · Jan 3, 2014

Lemming, is that Scale with parameters block your own Add-on instruction? I can't seem to find it in the standard tool set?

Lemming · Jan 4, 2014

Fairly sure is on the AB knowledge base..., though could make your own easily, or use a compute instruction..

Compact Logic PID Loop Deadband

bgtorque

Member

Lemming

Member

bgtorque

Member

Lemming

Member

bgtorque

Member

Lemming

Member

bgtorque

Member

kamenges

Member

Lemming

Member

bgtorque

Member

bgtorque

Member

Lemming

Member

Similar Topics