Test Theory: Incremental Encoder Phase Angle Testing

This section will describe why tests are performed the way they are, and what information is gained by the test. In some cases this is obvious, but in other cases the reasons may not be clear.

INCREMENTAL ENCODER PHASE ANGLE TESTING 

The question is often asked as to how important it is to perform the phase angle test. The standard answer is that it is more or less an auxiliary test ranking below the Count Test in importance. The Count Test verifies that the encoder is producing the correct number of counts per revolution. Accurate counting is essential, and it must be verified above all else. However, the phase angle measurement can expose problems with the encoder such as improper assembly, etc. Out of tolerance phase angles may even provide clues to counting problems. Large phase angle errors reduce the rotational speeds at which the encoder can produce acceptable signals for the electronics with which it is used. We can discuss the phase angles with the help of picture of the A and B pulses like you would see on an oscilloscope.

As with an oscilloscope, the first events are on the left, and the last events are on the right of the picture. We will give the first rising edge of A an angle of 0, and measure everything else relative to it. The falling edge of A is halfway between the two rising edges of A, so it must be at the 180 degree point. With this information we can determine the A symmetry angle. This measurement tells whether the A pulse is symmetrical. In simple terms, this just means is it HI for the same amount of time that it is LO? We determine this by subtracting the angle for the rising edge from the angle for the falling edge (180 - 0 = 180). In this case, it is 180 degrees which is perfect (perfect because the pulse is HI for the same amount of time that it is LO). The A symmetry angle is simply how many degrees the A pulse is HI.

We can apply a similar analysis to the B pulse which has a rising edge at 90 degrees and a falling edge at 270 degrees. Again, we determine this by subtracting the angle for the rising edge from the angle for the falling edge (270 - 90 = 180). We get another perfect 180 degree symmetry angle.

This tells us that these two signals are perfect by themselves, but we are not quite done. We need to see how they relate to each other, and we do that by measuring the phase angle. This angle is the angle from the rising edge of A to the rising edge of B, and we determine it by subtracting the angle for the rising edge of A from the angle for the rising edge of B (90 - 0 = 90). We have found that the phase angle is also perfect for our perfect encoder. If the encoder is rotated in the opposite direction, we would see that the rising edge of B occurs at 270 degrees, and our phase angle would be 270 degrees. This is also a perfect answer. These are perfect because the edges are all equally spaced, 90 degrees apart from each others (90+90+90+90 = 360).

Now that we have determined these angles, let's step back a minute and see what they mean. We also need to understand why 180 degrees is a perfect symmetry angle, and 90 or 270 is a perfect phase angle. Looking back at the picture, we see that there are 4 edges for each period (each complete 360 degrees) of encoder rotation. Our ideal encoder will evenly space these edges apart in the period. Any other angles will space some of the edges further apart and at the same time crowd others closer together. Spreading apart is fine, but crowding together is a problem. The electronics connected to the encoder will count these edges, and it will have a maximum rate at which it can receive the edges. This maximum rate really amounts to a minimum time separation between edges. As you know, when the encoder rotational speed is increased, the edges will move closer together. Simply put, if incorrect phase angles cause the edges to move closer together, it has the same effect as speeding up the encoder. Incorrect phase angles can cause minimum edge separation times to be reached at significantly lower rotational speeds that for the ideal phase angles. When the time separation between edges reaches the minimum, then the counting circuitry in the receiving electronics will begin missing counts. Of course, the flip side to this situation is that, if the rotational speed is slow enough, the phase angles matter very little because there is still adequate time spacing between edges.

Typical encoder specifications for phase angles are ± 22 degrees. This means that the 90 degree phase angle could range between 68 and 112 degrees, and it would meet that manufacturer's specifications.

The last topic to consider is how to make good phase angle measurements. The primary requirement is to rotate the encoder at a constant rotational speed. The TI-5000EX tester times the intervals shown in the picture and converts those times to angles. If the encoder speed is accelerating (speed ramping up), then the edges on the right hand end of the picture will be closer together than the edges on the left side, due simply to the increase in speed. This speed increase would cause phase angles and symmetry angles to look incorrect even with our ideal encoder. Since the tolerance on the phase angles is fairly broad, normally it is possible to get a constant enough speed rotating an encoder by hand. However, the best situation would be to rotate it with a constant speed motor.