Honeywell UDC 3500

[JDHammond 0]

I am having an issue with one of my UDC controllers. My controller is swinging too much and oscillating way too much. It isn’t being used as a controller, just as a pass through device. I’m taking a 4 to 20 mA signal to the UDC as a PV and reading the slide wire as a position indicator on an air damper. My problem is, say my PV is at 50% my slide wire value moves to 52% and switches the opposite direction. Then it moves to say 48% and so on. This is putting a lot of wear on our slide wires, dampers, and motor starters. Is there a parameter that I can adjust to say if you’re within say 5% of your PV to not move? I have played with different parameters and am getting now where. I have written in the logic of the PLC to minimize this swinging with timers, but I would like to do it with the controller instead. I would appreciate anyone’s help! Thanks

[kaiser_will 65]

First, was this system working reliably or is this a startup? If it was working reliably, something may have changed in the system that has uncovered the instability of the PID loop. Do a search for PID Loop Literature to get a better understanding of the parts of a PID loop and how their values can affect the outcome. Making PLC code changes to deal with an overactive PID loop only adds complexity to the problem. Can you put the slidewire on a meter and exercise it full span? I am guessing the slidewire has a deadspot (or two) where the impedance is changing non-linearly, which leads to the PID controller wanting a large change. I have run into this very problem before. A typical averaging multimeter may not fully reflect a deadspot, but an oscilloscope will. Honeywell typically has local support engineers (if you have a support contract) that are of great support in dialing a controller's application. What model of UDC are you using?

[JDHammond 0]

This system had been in place for a while but we added these controllers about 8 months ago. I think we have had the swinging problem for awhile. As to my knowledge there is no PID loop in this application. This application is using hot oil to heat our presses and our treaters. The OEM installed a KFM controller that was sending a 4 to 20mA to two devices that controlled 2 dampers that were oposite of each other for flow reasons. We removed the 2 devices that controled the dampers due to being obselete, no liteture, and not working with the UDC 3500 controllers. The KFM controller is where the PID loop is and not in the new controllers. Like I said they are receiving the 4 to 20mA from the KFM controller as in one damper is 100% open at 4mA and the other damper is at 0% open and vise versa. We did this by just doing a reverse calibration on the opposite controller. The system seems to work ok. Im thinking we have a problem with the damper motor itself that is causing this problem in the first place. I called the contractor that installed these new controllers and he said try the deadband and see if it works, which in this configuration is .5 to 5. I went to both extremes with no luck. The logic I wrote in the PLC seems to work fine without issues, but again I wanted to do it in the controller instead of the PLC. We ordered new slidewires and I will install them when they arrive and see if that helps at all! Thanks for the reply!

[Mickey 71]

If you are not using the Honeywell controller to control with then why is it there? Bypass it and see what happens.

[kaiser_will 65]

Mickey has a point. The UDC controllers may be adjusting the signal. Gather data to verify. The Honeywell UDC3500 controllers are acting solely as signal converters and damper controllers, and the KFM controller is the master with the PID control. Does the KFM controller support dual/multiple PID loops with separate outputs? If the 2 dampers are to run independantly of each other, then they should probably have isolated PID control loops (i.e., be driven by separate controllers - such as the UDC3500). I have no experience with the KFM controller. What steps have you gone through to tune the master PID control loop? If you have the time, it may be beneficial to gather some empirical field data with the goal of tuning the loop. In the past, I have worked through similar scenarios by temporarily rigging a PLC in place of the master controller, and using a PLC data collection package (WebDock is a freebie) to build a spreadsheet of data (date, time, PID overshoot, PID error, output values, etc.). If a PLC swap is not in the cards, you may be able to program additional controller outputs to give you what you want and use a PLC to act as the data gateway.

[JDHammond 0]

"The Honeywell UDC3500 controllers are acting solely as signal converters and damper controllers, and the KFM controller is the master with the PID control. Does the KFM controller support dual/multiple PID loops with separate outputs? If the 2 dampers are to run independantly of each other, then they should probably have isolated PID control loops (i.e., be driven by separate controllers - such as the UDC3500). I have no experience with the KFM controller." Yes the first part is true! The KFM doesn't support multiple PID loops. They dampers dont run independantly, the have to equal 100% flow between the both of them ex 70% 30%. The KFM controller is obselete also, and our plant was built by germans with all german equipment. The attachment is our RSView screen that gives a idea of what im dealing with. The KFM controls the oil temps by the dampers GS114 and GS115. The Oil is heated by the hot air from our incinerator. GS 115 is the one giving me problems. Doc1.doc Edited 22 Mar 2010 by JDHammond

[DanW 41]

Is this still an issue, or is it resolved?

[paulengr 44]

Although I doubt the OP is still reading, the problem is obvious at least to me. Probably there are two different controllers which are acting independently to try to run two different dampers. This is a recipe for disaster since their loop bandwidths are identical...every move by one gets picked up by the other (with a delay), which then feeds back into the first one. The delay doesn't have to be very different before you get a great oscillator (more than 180 degrees of delay anywhere in terms of frequency with close to or exceeding 1.0 for gain results in an oscillator which may or may not be damped depending on how close you are to 1.0). There are two ways out of this. First, you can give them different controller bands. Say adjusting one to operate very slowly and the other very fast. With a BODE plot of the controller loops, this is fairly easy to do and visualize. Alternatively, you can run them in a dependent/independent manner. For instance, run the PID loop to control only ONE damper. The other damper position is determined by using information about the first damper only, perhaps controlled by a PID loop where the set point is derived from information about the first damper. You can still get oscillations depending on if you use the output signal or feedback on position. The other alternative is to set up a cascaded controller. The first (master) controller determines overall damper position. The second PID loop (actually two of them in this case) simply responds to the data from the master loop. These are positioning loops. You may also have to solve this as a ratio controller, which is a common problem in mixing applications. For instance if you have 2 valves controlling hot and cold water, the mistake is to try to drive them from a single PID and a single feedback (temperature) as this never works. Instead you have to drive one to control flow rate, and the second to control temperature. Even though the second doesn't account for temperature, even if it begins to open up more, then the first valve will close up to account for the overall flow and vice versa.