Tachometry, the measurement of the speed of rotating objects, is a common enough application that has many practical but few noteworthy solutions. Some of these jobs, however, have quirky aspects that make them interesting. One such category includes remote, outdoor, noncontact sensing of large, rapidly moving, and potentially hazardous objects like windmills, waterwheels, and aircraft propellers. The tachometer illustrated here is specifically optimized for performance monitoring of aviation engines, but is adaptable to other applications with simple changes of RC time-constants.
Safe sensing of large rotating objects can only be done from a distance; optical methods are generally the obvious choice for accomplishing this task. Unless elaborate telescopic optical systems are used in front of the detector, the optical signal is apt to have a relatively low-amplitude. That’s due to the tendency of the rotating object (propeller blade, etc.) to fill only a small fraction of the typically wide field of view of simple detectors.
This tachometer (see the figure) makes do with an uncomplicated detector (phototransistor Q1 with a tubular light shield) by following the detector with an adaptive, low-contrast threshold circuit previously published in two Ideas For Design (ELECTRONIC DESIGN, “Chronometer Settles in One Cycle,” May 28, 1996, p. 98, and “Build Your Own Optical Heart-Rate Sensor,” December 15, 1997, p. 104).
Q1’s photocurrent produces an ac signal across Q2 and Q3 of ~500 mV pp for every 1% change in incident light. This logarithmic relationship is constant over many orders of magnitude of photocurrent. Therefore, it’s able to provide reliable circuit operation despite wide variation in outdoor light level. A1 and the surrounding discrete components comprise a highgain adaptive filter that rejects ambient optical and electrical noise and presents a cleaned-up signal to cascaded comparators A2 and A3. This outputs a clean 5-V p-p square wave to the C1…C4, Q6...Q9 charge pump.
This charge pump is borrowed from another Idea For Design (ELECTRONIC DESIGN, “Nanopower VFC Includes Self-Compensating Charge Pump,” June 22, 1998, p. 131). Analysis reveals that, if we assume C1 = C2 = C3 = C4, equal stray capacitances around the transistor emitter nodes and equal transistor bias voltages, then each complete cycle of A3’s squarewave output will inject a net charge onto U5 given by:
−5 V * C1 = 28 nCb
with the amplitude of the charge pulse well-compensated against transistor junction tempcos. Transistors are used here instead of the more usual diodes to eliminate some undesirable effects on the accuracy of charge transfer caused by the ripple on C5.
Thus, a frequency-proportional average current of −28 nA/Hz is injected onto C5 and thereby into the A4 inverting two-pole low-pass-filter. This results in an output voltage of VOUT = 28nA/Hz * RCAL. For RCAL = 1.071M, VOUT = 30 mV/Hz for an output scale factor of 1 V per 33.3 Hz.
Such a combination works well for two-bladed aircraft propeller timing because it results in a convenient conversion factor of 1 V = 16.7 rps = 1000 rpm. Other applications with different speed ranges may call for other scale factors. This is easily accommodated by maintaining an inverse proportionality between all circuit capacitances (C1...C6) and the desired conversion factor. For example, if an output relationship of 1 V per 100 rpm, in the same context of 2 pulses/revolution, were needed, all of the capacitors would be increased by a factor of 10.