"Integral Deviation" Value?

I'm working with the Delta temperature controller DTB969RRE-D.

It has four PID Settings: PB, Ti, Td and "Integral Deviation Value".

Does anyone know what that last value does? Can't find any documentation on their site, google and this site.

Well, the only thing I did find on google was in an Omega manual. It says that the integral deviation value

"If set to zero, the output will be zero when the process value is equal to the set point value. If the integral time parameter is used only to eliminate steady state error, it may take a long time to reach the point because it needs time to accumulated error. This parameter defines the default output level on start up."

Click to expand...

I'm not sure how to interpret that.

Any help appreciated.

The explanation has two different definitions. The latter, "This parameter defines the default output level on start up" is used by other brand/models to provide a given % output at start-up. However those devices use terms like 'Auto Out' or 'Man Out' that doesn't refer to the integral/reset term.

The former definition, "If set to zero, the output will be zero when the process value is equal to the set point value" could be either a poorly worded explanation of reset wind-up or a really bad interpretation of what PID controllers do.

The poorly worded reset-windup explanation: A PI or PID controller that is wound-up (with the integral factor driving the output to 100%) will typically re-zero/reset the integral term only when the PV crosses the SP. So it could have meant, "If set to zero, the integral factor will rezero or reset itself when the PV is equal to the SP".

The really bad interpretation of PID: In normal operation, when the PV = SP, the error is zero, so the output should hold where it is, in order to maintain a steady process value. If the output were to go to 0% output when PV = SP, then there would be no energy or mass movement into the process and the PV would drop (assuming reverse action) due to lack of energy or mass input. This operation is highly unlikely in a commercial device.

I think you'll have to dink with it and see which of these contradictory definitions apply.

Set it at 40 and see if the output is 40% when you turn it on. If it is, you know what the function is and you can decide where the output should be when you turn it on. If the output isn't 40% on start-up, then leave it at zero and see if operation is normal enough to tolerate.

I never heard of the term "Integral Deviation Value".
Resetting the integrator to 0 when the PV reaches the SP will not work in velocity or temperature control applications. These are non-iintegrating applications. If the PID is implemented correctly there shouldn't be a need for weird terms or parameters.

Sometimes I think it is best to write your own PID. Then you know how it works.

It looks like B&B Electric sells this device. Their tech support is generally very good. Give them a call and see if they know what it is. If you find something useful, please report back here..

Originally posted by Peter Nachtwey:

Sometimes I think it is best to write your own PID. Then you know how it works.

Click to expand...

That's the only reason I write my own. It certainly isn't because it performs any better than what is available already. But I can implement it in about half the time because it only contains what I really need and if it isn't doing what I expect it to I can look into it and see why.

I've never heard of that term either. The wording of it makes it very difficult to even hazard a guess as to what is meant. I know some controllers have an option not to implement integral correction until the PV is within some value of the SP. It may be something like that also. But I think ndzie d1 has the best plan.

Keith

Thank you all very much for the input. It's very appreciated.

I'm working with technical support at Delta to see if I can get a firm answer since it appears this isn't a well known industry term. I'll report back with updates.

Ok, I finally made contact with Delta and the guy said that the "integration default" value is the integral contribution when there is no error (SP = PV).

I guess there are times when you want to have a constant output at steady state.