are 20% and 100% respectively, the output per- centage will never be below 20%, even if the P, I, and D Mode adjustments call for the output to be below 20%. The output will remain in saturation at 20% until a higher output percentage is called for.

In short, PID works as it normally does, except the out- put never goes below Output at Minimum or above Output at Maximum.

Minimum Accumulated Error

The Minimum Accumulated Error setting disables error accumulation in the “I” Mode when the current error is equal to or less than a certain amount. For example, if the PID setpoint is 30 and the Minimum Accumulated Error is 1, the “I” Mode will not adjust the PID output per- centage as long as the input is between 29 and 31.

Because “I” Mode does not accumulate error within the Minimum Accumulated Error range, the control input is allowed to settle on a value other than the setpoint. It is possible in the example given above for the system to achieve stability at any value between 29 and 31 without the “I” Mode trying to bring the error to zero.

Application

The practical use of Minimum Accumulated Error is to set up a “close enough” zone around your PID setpoint. When the control input is within the zone, PID will not worry about making any “I” adjustments to attempt to equalize the input and the setpoint. Proportional Mode will handle any minor input changes that occur within this zone.

Filtering

Filtering is a feature commonly used by CC-100s in valve control (and it thus sometimes called “valve filter- ing”). The primary purpose of filtering is to dampen the rate of change of the PID percentage in an effort to make the PID control loop less reactive.

The filtering feature takes samples of the PID output percentage at regular intervals (which are determined by a parameter called the filter period). Every time a new filter period interval occurs, the PID percentage sampled during the last period interval is subtracted from the current PID percentage from the current period interval.

The result of this subtraction is multiplied by a param- eter called the filter percentage (0 - 100%) to yield the actual amount the PID percentage will change.

Over time, the application of the filter percentage to the change in PID position will result in a PID control loop with a smaller amount of reaction to changes in the input.

Example: A stepper EEV valve on a CC-100 is con-

trolled by PID control. Valve filtering is active in this CC- 100, with the filter period set to six seconds and the filter percentage set to 75%.

During one sample taken during a period interval, the CC-100 calls for a valve position of 50%. One period (six seconds) later, the CC-100 asks for a 58% valve opening.

The total difference between the current sample and the previous sample is +8% (58 - 50%). To determine the actual amount the valve will change, the CC-100 multi- plies the filter percentage (75%) with the total amount of valve position change (8%). As a final result, the new PID output value for the CC-100 will be 56%.

Note that filtering only slows down the reaction of the PID loop. When the control input is stabilized, the PID loop will eventually achieve the output percentage it is calling for.

To demonstrate this, suppose in the example above the CC-100 continues to call for a 58% output during the period immediately after the 6% adjustment. Since the total difference between the asked-for percentage and the current actual percentage is 2% (58 - 56), valve filtering will make the new adjustment for that period 1.5% (75% of 2). As a result, the new valve output would be 57.5%. Future filter periods will bring the actual output even closer to the asked-for output.

Application

PID filtering is used for systems that appear to be over- reacting to changes in the control input. If filtering is to be used at all, it is recommended to use caution, since even a small amount of filtering may cause the PID loop to become underreactive.

D-6 E2 RX/BX/CX I&O Manual

026-1610 Rev 13 14-SEP-2011

Page 242
Image 242
Emerson E2 operation manual Minimum Accumulated Error, Filtering