I a mechanical engineer working at a mission hospital where we have a relatively new (2013) hydroelectric power plant. The turbine has a speed sensor (DC Tachogenerator) that provides a voltage to a control system to maintain constant RPM.

This Tachogenerator has been going through brushes rather quickly, and even with new brushes there is some fluctuation in the voltage (even though I know from the generator frequency that the system is running close to steady state). The frequency readout that receives input from the Tachogenerator will fluctuate around 3 Hz even though the generator frequency fluctuates under 0.5 Hz. With old brushes the reading is so poor the plant will overcompensate itself outside of the power quality parameters and shut down.

Our maintenance staff will clean the contacts of the rotor whenever they need to change the brushes. There is usually lots of carbon inside the housing.

I would like to know if we can replace this Tachogenerator with a more precise unit, preferably brushless. Long brush life is also important. We can't be shutting down several times per year to change brushes.

This resource is leading me to believe that a DC Tachogenerator is superior for our application. From my understanding an AC Tachogenerator would need a rectifier and capacitor to give a DC voltage with minimal ripple. However, the brushless nature of an alternator seems better to me. http://www.polytechnichub.com/difference-ac-tachogenerator-dc-tachogenerator/

Please see the picture for details on the tachogenerator currently in use. enter image description here

  • \$\begingroup\$ "High signal quality due to patented LongLife technology" What does Baumer recommend? What is your load R? \$\endgroup\$ – Tony Stewart EE75 Sep 8 '17 at 1:33
  • \$\begingroup\$ tachogenerators are ancient as generators in cars made obsolete in the early 70's with replacement with alternators, but there are far more accurate electronic frequency tach instruments, but question is how much current is need to drive regulator. normally TG's are only supposed to drive low power X watts \$\endgroup\$ – Tony Stewart EE75 Sep 8 '17 at 1:52
  • \$\begingroup\$ It is possible the armature is damaged causing excessive brush wear with arccing between rotating contactors from carbon contamination and erosion. They ought to last many years, voltage output is dependent on calibration and magnet aging but rated here for 20mV/RPM. Need more specs on application to suggest better part but certainly brushless. \$\endgroup\$ – Tony Stewart EE75 Sep 8 '17 at 1:56
  • \$\begingroup\$ What's the nominal speed? You will hardly find an industrial device that outputs more than +/-10V. \$\endgroup\$ – Marko Buršič Sep 8 '17 at 7:41
  • \$\begingroup\$ Nominal speed is 100rpm \$\endgroup\$ – Lukas Woltjer Sep 8 '17 at 8:23

Daft thought, but this is a synchronous alternator based plant right?

Why not take the feedback from the main generator frequency? Frequency to voltage is a fairly standard bit of circuitry and you should be able to make the time constants small compared with the mechanical inertia so the control loop will still be stable.

Regards, Dan.

  • \$\begingroup\$ How, if the generator is connected to the mains? \$\endgroup\$ – Marko Buršič Sep 8 '17 at 8:11
  • \$\begingroup\$ If the machine is connected to a much larger grid then there is no real need for speed control except to trip the thing off line if there is a problem because it will rotate at grid sync speed simply by the nature of the thing (The rotor angle is determined by the torque and hence the input power, but steady state shaft speed always tracks the grid). I get the impression however from the fact that the output frequency varies that this is at least substantially a stand alone plant. \$\endgroup\$ – Dan Mills Sep 8 '17 at 8:15
  • \$\begingroup\$ This is a synchronous alternator based plant. It runs in "island" mode. If frequency to voltage is rather standard that seems like the best option to me. I'm hoping to simulate the tach output in order to avoid having to replace further parts. The key will be having a very steady DC output with minimal ripple. If you can recommend a commercial solution I will suggest this to the team. \$\endgroup\$ – Lukas Woltjer Sep 8 '17 at 8:28
  • \$\begingroup\$ Plenty of 24V Din rail kit out there that looks suitable, you will need a small transformer to take the generator voltage down to something reasonable to feed the frequency to voltage converters input, and most of them top out at 10V DC on the output, so depending on what the tacho actually produces you may need to do some tweaking but it should be simple enough. \$\endgroup\$ – Dan Mills Sep 8 '17 at 9:25
  • \$\begingroup\$ It may only need an electrolytic capacitor load shunting the output which probably drives and inductive load to prevent EMF spikes that short brushes during commutation. \$\endgroup\$ – Tony Stewart EE75 Sep 8 '17 at 13:41

The best speed detector technology uses an inexpensive incremental encoder. There is a wide range of applications for precision frequency and phase control on a rotor position ranging from 1 to thousands of pulses per revolution. Then using a simple pulse voltage averaging method it is converted to DC . But explicit voltage, current load , tolerance, ripple and accuracy must be defined.


This method of system power regulation seems to be very poor as described. If a brush tachogenerator is used to regulate an inductive actuator for power or speed control, then the commutated RPM voltage will certain erode brushes without a suitable RC capactive load to act as a flywheeel to each pulse. Otherwise arccing will erode the brushes quickly.

Some AC and DC measurements of output would help and a scope photo even better.

  • \$\begingroup\$ Thank you Tony for your answer and comments. I will have to wait until we shut down for regular maintenance to get more information on what the specifications are on the regulator. I can, however, get voltage and current produced by the tachogenerator during operation. I'll post those values when I get them. \$\endgroup\$ – Lukas Woltjer Sep 8 '17 at 2:58
  • \$\begingroup\$ While I would say upgrading to an encoder is a good idea, I would suggest even more strongly upgrading to a sin/cosine resolver. Resolvers are just about bullet proof, and extremely accurate. There are many signal converters to take a resolver output and convert it to an analog signal or an incremental (encoder) signal. \$\endgroup\$ – R Drast Sep 8 '17 at 10:01
  • \$\begingroup\$ sin cos is good for phase frequency and good for tight servos but to convert to V/RPM for what appears an "island" power sysyem with 0.5Hz acceptable tolerance is overkill \$\endgroup\$ – Tony Stewart EE75 Sep 8 '17 at 13:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.