What is PID Feedback?


What Is PID Feedback? Understanding the Measurement Behind Pump Control

PID feedback is the measured process information returned to a controller. In a constant-pressure pumping system, it is usually the line pressure reported by a transmitter. The controller compares that measurement with the pressure target and adjusts the pump’s speed demand. Feedback tells the pump control system what is actually happening.

Feedback and PID are different parts of the loop. Feedback supplies the measurement; the PID algorithm calculates a corrective output from it. A well-tuned controller cannot reliably maintain the true pressure if the measurement it receives is wrong.

1. Setpoint, Feedback and Output Are Different Signals

The setpoint, or SP, is the required target. The process value, or PV, is the measured condition. The controller output is the command used to influence the process. A system might have a 4 bar target, a measured pressure of 3.5 bar and a current drive command of 42 Hz. Those values describe three different things.

A common error definition is error = SP − PV, giving +0.5 bar in this example. Where increasing pump speed raises discharge pressure, the controller should increase its speed demand within the permitted limits. The VFD output frequency is not a substitute for pressure feedback: the same frequency can produce different pressures as demand changes.

2. What Can Provide Pump Feedback?

Pressure regulation uses a suitable pressure sensor or transmitter. Differential-pressure regulation uses the pressure difference between two locations, often measured by a differential pressure transmitter. Pressure at one point and pressure difference across a circuit are not interchangeable measurements.

Flow control needs an appropriate output from a flow-measuring instrument. Continuous level control needs a suitable level measurement. An ultrasonic level sensor may be appropriate, for example, provided its installation accounts for the liquid, foam, vapour and measurement dead zone.

A float switch or pressure switch normally reports whether a threshold has been crossed. “Level low” or “pressure sufficient” is different from a continuously available numerical measurement. The distinction between analog and digital inputs helps explain why a run-permission signal is not necessarily the PID process value.

3. Feedback Through 4–20 mA or 0–10 V

Industrial transmitters often represent their measurements using 4–20 mA or 0–10 V signals. The electrical signal represents pressure; it is not itself a pressure value. An analog input reads that signal and converts it into information the controller can use.

In a typical 4–20 mA arrangement, 4 mA represents the lower measurement limit. This live zero allows a suitably configured input to recognise very low current as a possible broken wire or loss of transmitter power. It does not guarantee detection of every sensor failure.

With 0–10 V, zero volts can be a valid measurement, making it harder to distinguish a genuine zero from a connection fault using the signal level alone. Input type, measurement range and fault handling must be configured separately for either interface.

4. Scaling the Signal into Pressure

Scaling converts the electrical input into engineering units such as bar, metres or m³/h. For a linear 0–10 bar transmitter with a 4–20 mA output, the conversion is:

Pressure (bar) = [(Current (mA) − 4) / 16] × 10

Ideal signal values for a 0–10 bar transmitter
Signal current Measured pressure Percentage of span
4 mA 0 bar 0%
10.4 mA 4 bar 40%
12 mA 5 bar 50%
20 mA 10 bar 100%

Incorrect scaling can make a controller appear successful while the actual pressure is wrong. If this transmitter is mistakenly configured as 0–16 bar, an actual pressure of 4 bar will display as 6.4 bar. When the controller settles at an indicated 4 bar, the actual pressure will ideally be only 2.5 bar. The controller sees zero error, but the real target has not been met.

This example converts an analog signal directly into bar. Controllers using percentages or raw counts must place setpoint and feedback on compatible scales. Do not apply square-root extraction to an ordinary linear pressure measurement unless the measurement principle specifically requires that conversion.

5. Check the Electrical Interface

A two-wire 4–20 mA transmitter commonly receives power and transmits its measurement through the same current loop. Three- and four-wire devices may have different arrangements for power and signal. The wire count alone is not a wiring diagram, and not every two-wire sensor is a 4–20 mA transmitter.

Check whether the input supplies loop power, the transmitter output arrangement, polarity and allowable load resistance. The supply must provide enough voltage for the transmitter and the voltage drops across the input and wiring. Use a compatible connection scheme from the equipment documentation.

Appropriate separation from motor power cables, cable shielding and the grounding arrangement all affect measurement quality. There is no universal shield-termination rule for every installation; follow the equipment’s EMC instructions and account for the site’s equipotential bonding.

6. Sensor Location, Filtering and Delay

The controller regulates what the sensor can see. Pipe losses can create a difference between pump-discharge pressure and pressure at the point of use. A blocked pressure connection or trapped air in a measurement line can delay or distort feedback even when the electrical signal appears valid.

Filtering can reduce short-term fluctuations, but excessive filtering makes the controller respond to older information. Measurement and process delays must be considered when tuning the loop. A calm display does not necessarily indicate accurate measurement or good control.

7. What Happens When Feedback Fails?

A broken cable, lost transmitter supply or communication failure can remove the controller’s view of the process. A PID loop running in a VFD may increase its speed demand if it incorrectly interprets the failure as very low pressure. Signal-loss detection, the detection delay and the resulting action therefore belong in the control configuration.

Depending on the application, the response may be a controlled stop, alarm, restricted operation or transfer to a validated backup sensor. Holding the last value indefinitely or automatically commanding maximum speed is not a universal solution. A sensor frozen at a plausible value may also escape simple high- and low-limit checks.

For measurements delivered digitally through a PLC, monitor communication status, data validity and freshness as well as the number itself. A value remaining on screen does not prove that new measurements are arriving.

8. Separate Measurement Faults from Tuning Problems

Start with the measurement chain. Compare an independent reference, the transmitter signal and the value displayed by the drive under the same conditions. Agreement at one point may not be sufficient; comparison at several suitable points helps distinguish zero errors from span errors.

Then examine the response. If the measurement is correct but pressure and speed oscillate regularly, assess the PID settings and delays. If the measurement jumps or freezes, investigate the sensor, connections and data transfer first.

SCADA trends can record the setpoint, process value, raw input, drive frequency and sensor fault status together. This gives a firmer basis for diagnosing control problems than repeatedly changing proportional or integral settings without checking the measurement.

9. Does Feedback Make a PID Controller Adaptive?

No. A standard PID controller can respond to changing error while retaining fixed tuning parameters. That does not mean it automatically learns new gains. Autotuning and adaptive control are separate capabilities that the controller must specifically support. The same distinction applies to PLC-based PID systems.

Proportional action responds to present error, integral action to accumulated error, and derivative action to the rate of change. Increasing the gains does not always improve speed or stability. A stable control loop requires reliable measurement, the correct action direction, consistent scaling and tuning suited to the process.

Good feedback is the foundation of good control. It does not automatically replace independent overpressure or dry-run protection where the application requires those functions.