Actual Speed

The Actual Speed is the calculated speed at which the axis is currently moving, as computed from the change in the Transducer Counts status register. The Actual Speed is in position units per second.

This value is filtered to reduce the effect of quantizing errors. In firmware dating 19991216 or newer, the Actual Speed is filtered by applying a low pass Infinite Impulse Response (IIR) filter with a 7 ms time constant. In earlier firmware, the filter consists of summing the change in position over the last four time periods and dividing by four time periods.

What is Quantizing Error?

Before the position of an axis can be used by a digital controller such as the RMC, its real-world, infinite-resolution, analog value must be converted through a transducer into an integer number of transducer counts. The loss of position resolution that occurs when truncating to integers is referred to as quantizing error.

Quantizing error affects the accuracy of positions, but it also affects the accuracy of speeds even more significantly. The following example demonstrates the problem quantizing errors pose to computing the Actual Speed.

The simplest way to compute the Actual Speed would be to divide the change in position units since the last sample by the sample time period, as shown below:

ActSpd =	x(n) - x(n-1)
	T

where:	x(n) is the position sampled this control loop
	x(n-1) is the position sampled last control loop
	T is the sample period (control loop time)

Suppose that the sample period (control loop time) is 1 millisecond, the transducer returns one count for every thousandth of an inch, and the axis is moving at 4.8 inches per second (4800 counts/s).

As the controller samples the position each control loop, it will see the position change by either 4 or 5 position units, but never the actual rate of 4.8, because the transducer returns a fixed number of counts. Therefore, when these values are divided by 0.001 seconds, the computed Actual Speed will jump back and forth between 4000 and 5000, but will never show any value between.

The situation improves with higher speeds and higher-resolution transducers, but worsens with lower speeds and lower-resolution transducers. Filtering can be applied to smooth out the Actual Speed, as the RMC does, but the filtering must be limited to avoid smoothing out real changes in the Actual Speed.