PID Control Theory


Thread Starter

Satish Doshi

I would like to know about the velocity and position control algorithm for PID controller. What is the difference between these two algorithm? How bumpless transition is achieved in velocity algorithm?

Brian Prigge Electromatics, Inc.

I don't know exactly where you are coming from, but it sounds as though your questions are applicable to two different types of PID implementation.

First the velocity loop and the position loop are used in motion control. They both have similar function and operation, but the application is very different. The basic servo uses two PID loops. The first loop, velocity, is "inside" of the second loop, position. The purpose of the velocity loop is to ensure that the load is moving at the commanded rate. It gets it's feedback as a speed signal. The second loop is the position loop. Once control of the load has been established by the velocity loop, the position loop can operate with the load. The purpose of the position loop is as the name implies, to keep the load in position while it is in motion. This is where following error comes in. Following error represents how far the load is either ahead or behind the command signal. These signals usually take the form of positions as indicated by a counter value.

As to bumpless transfer, the only time I have ever run across that is in process applications, pressure, temperature, flow, etc. Quite often in the past, tuning a PID loop was not possible due to lack of training, lack of proper hardware or both.

In some cases the loop was run in manual for some reason. Then at some point in time it was placed in automatic. When this happened, and if the process was not on target with the setpoint, the control began to immediately try to correct, hence, the "bump".

This is all very basic and over simplistic. If you want to know more, there are some good books on the subject.


Hans H. Eder

1) For velocity algo's (also called incremental algo's) the result of the PID equation is a CHANGE. Example: P-controller. The change in the output is = proportional to the change in input (here the controller's output algo has to convert this into the full position). In case of a positional algo the PID calculates the full value. For the P-controller again: The output = proportional to the input + a base value.

2) Bumpless transfer means to avoid bumps (coming from the P-action) when the SP and the PV are far away from each other at the moment the controller is put on automatic or the controller is recovering from any situation that forced it into
initialization. Typically the SP is then set equal to the current PV and also the (internal) historical values are reset.

Hans Eder, ACT