# Interfacing Of Velocity Transducers

**Interfacing Of Velocity Transducers**

The tachogenerator interfacing is quite straightforward since it requires only AID conversion, though it may, in some applications, require low-pass filtering to remove commutator spikes andlor output voltage attenuation to match it to the AID working range. An example of interfacing a tachogenerator to a digital system is shown in Figure 1.

Since the transducer output is analogue (i.e. continuous) the velocity measurement resolution depends only on the number of bits used by the AID converter; an 8-bit converter will therefore provide just under 0.4% measurement resolution, that is (1/256) x 100 = 0.39%.

Optical incremental encoder interfacing, on the other hand, requires a voltage amplifier as well as a counting circuit in order to interface it to a computer. The optical encoder output is a low-voltage pulse train whose frequency is proportional to the shaft angular speed. To measure this speed the interfacing hardware needs to count the number of shaft revolutions per unit time. This can be achieved in two ways:

- By counting the number of lines detected by the photodiode in a unit time, a technique suitable for measuring medium to high speeds .
- By measuring the time taken by the photo diode to detect a single line.

The latter is a useful technique when measuring low speeds with high resolution and is based on gating a high-frequency clock with the photodiode output pulse corresponding to a single line transition (Figure 2).

The former technique is, however, more useful because in most cases robot shaft actuators rotate at medium to high speeds (that is in excess of 1 rev Is) and will therefore be described in more detail.

To measure speed by the former technique, one pulse per revolution (i.e. a single line or slot on the encoder disk) would be sufficient. But digital counters have an inherent count resolution of ± one clock pulse so, in order to allow the accurate measurement of low angular velocities, more than one pulse per revolution needs to be counted per unit time and therefore several dark lines (or slots) are required on the encoder disk. The velocity measurement time can thus be reduced at the expense of having more lines on the disk and vice versa.

The three main components of the processing hardware are therefore an amplifier, a digital counter and a clock circuit. A real-time clock is not strictly necessary since the measurement can be calibrated on any time interval to give the output in revolutions per minute as required. However, with the advent of very stable and highly accurate quartz crystal oscillators a real-time clock is quite easily built. A circuit is illustrated in Figure 3.

where the output waveform period T can be made a multiple or a subdivision of a second by choosing a suitable crystal frequency and divider factor. Examples of such circuits are found in modem quartz watches.

A much less complex and cheaper alternative to a real-time clock is provided by the mains supply frequency. This is maintained stably in Europe at 50 Hz and in USA at 60 Hz. Synchronization to the mains frequency for timing purposes is therefore very convenient and easily attained, as shown in Figure 4.

Using the mains frequency to generate the required time intervals, we can therefore design a suitable circuit for interfacing an optical incremental encoder to a robot control computer. An example of such a circuit is shown in Figure 5.

The output from the incremental encoder is a sequence of pulses which are counted to provide the velocity measurement, whose resolution there fore depends on the minimum number of pulses counted in the given measurement interval; that is on the minimum number of lines counted at the slowest shaft speed. The measurement time interval itself depends on the computer handshaking, that is on how often the computer needs a velocity measurement. This time interval can be exceeded if the measurand has a longer mechanical time constant since the shaft speed would not change significantly between computer updates due to the shaft mechanical inertia.

If the computer requires a velocity measurement every 100 μS for instance, and the measurand mechanical time constant is 1 second (e.g. a shaft driving a large inertial load), then the speed will not change significantly in 100 μS and the measurement interval can be much longer, say 100 ms in this example, and the computer will simply be given ‘old’ measurement values during the 100 μs access times until the next measure ment is carried out on the 100 ms mark. Given a measurement interval time, however, the measurement resolution depends on the number of pulses counted in that interval, hence:

*N = L . f*_{min }*. **t*_{min}

where

N = required number of pulses = 100/required measurement resolution

L = number of lines on the encoder disk

f_{min} = minimum frequency = minimum shaft speed in rev/s

t_{min} = measurement interval

For instance, for a 5% resolution (i.e. a maximum of 1 ‘wrong’ unit count in 20) we need a minimum of 20 pulses counted during the measurement interval. Using a disk with 60 lines on it we can calculate the minimum shaft speed for which we can guarantee the required resolution, that is 0.3 rev/s which is 18 rev/min :