2

Is it a good idea to connect a galvanometer (100 µA panel meter) directly to a PWM output of a microcontroller (or any other digital device for that matter)? I set PWM frequency to approximately 10 kHz and duty cycle varies from 0 to 6% for full scale.

enter image description here

My worry is if the push pull output has a dead time, then the inductor will induce spikes on the driver pin. Does it require extra Schottky diodes to the power supply or are the internal protection diodes sufficient?

I have the diodes lying around, it is not a matter of cost, it is a matter of what is best / necessary practice?

Another option would be a 27k resistor in series and use full 0 - 100% duty cycle, but the question remains as the inductor acts as a current source and "doesn't care" about a series resistor.

®Stellaris LM4F120H5QR Microcontroller

Nick Alexeev
  • 37,739
  • 17
  • 97
  • 230
jippie
  • 33,033
  • 16
  • 93
  • 160
  • What are you trying to do? BTW, galvanometer is not a digital device, but the very opposite - analog..) – Eugene Sh. Jan 21 '15 at 19:35
  • @EugeneSh. The "usual" reason is to use it as an analog "display" of some digital data. – vicatcu Jan 21 '15 at 19:43
  • @EugeneSh. I want an analog read out of a digital value. I works just fine, I just want to know if it is a bad idea to do it without flyback diodes. The difference between need for a PCB with a couple headers and diodes, or two straight wires. – jippie Jan 21 '15 at 19:47
  • @jippie You could add a resistor in series with the galvanometer. That could do two things. (1) You could use a wider range of PWM duty cycles, which could improve resolution. (2) Series resistor would provide current limiting for the μC output stage. (disclaimer: The above is based on general understanding. I haven't worked with galvanometers before.) – Nick Alexeev Jan 21 '15 at 20:02
  • @jippie Have a look also at [this schematic](http://www.eleccircuit.com/the-cheap-frequency-meter-using-ca3130-and-lm555/) where a galvanometer is driven with a PWM output from a 555 timer. – Nick Alexeev Jan 21 '15 at 20:04
  • @NickAlexeev (1) Don't need the resistor for resolution, I have a 500 counts full scale on a 4 cm span ;o) (2) The impedance is about 1.8k, so maximum current would be in the order of 2mA at 100% PWM, but averages at 100 µA as that is the full scale. – jippie Jan 21 '15 at 20:05
  • @NickAlexeev the diode in that circuit is used to clamp the voltage to a given maximum. If you over-power a galvanometer, the needle smashes itself into one of the corners which is suboptimal. As the PWM frequency is high enough, the needle moves really nice and steady. Only thing is not to set PWM over 6%. – jippie Jan 21 '15 at 20:07
  • I like the series resistor... how else are you getting voltage to current conversion? You could even add a cap for a low pass. (Meters are pretty slow after all.) – George Herold Jan 21 '15 at 20:16

4 Answers4

3

Use a series resistor, scaled such that 100% PWM output (i.e. 3.3V or 5V depending on the MCU) gives 100uA or a bit over (up to 50% over).

Direct connection would only allow something like 10% duty cycle before full-scale deflection, not very useful.

And the series resistor will attenuate any inductive spikes from the meter coil well enough to protect the MCU.

It is possible but unlikely that a 2mA overload, e.g. from a software error (integer overflow?) could cause damage to the meter movement. Unlikely to burn the coil out, but it's a fairly violent mechanical movement. 100uA movements should be pretty tough but more sensitive ones used to have delicate jewelled bearings, like watch movements.

2

A 100uA meter will be no problem- the outputs are push-pull and do not have a dead time.

I suggest you use a series 1% resistor to get a stable reading. Otherwise the scale factor will be determined by the output's current limit and by the resistance of the meter, both of which are variable and change with temperature. Use a resistor that's just low enough to get a full-scale deflection at maximum duty cycle (or slightly more and throttle it back with a firmware calibration).

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
1

The meter's inductance will cause spikes. However the coil in a moving coil meter is usually wound around an aluminum former which acts as a shorted turn (to damp the movement). When combined with the relatively low inductance and high resistance of the coil, the spikes should be quite weak.

Putting a resistor in series will reduce spike intensity even further, and protect the meter from overload if the PWM ratio goes over 6%. Most microcontrollers have internal protection diodes on their I/O pins, so the only thing to be worried about is whether they can handle the spike current.

With a 27k resistor the spikes would have to exceed 27V to push 1mA into the I/O pin. I tried driving 10KHz 50% PWM into a 50uA meter with a BS107 MOSFET and 27k series resistance. At the meter terminals the spike had a peak amplitude of 1.1V. At the FET it was less than 0.6V.

Bruce Abbott
  • 55,540
  • 1
  • 47
  • 89
0

AT 10 kHz, your meter will not move at all and integrate the current quite well.

The inductor in the meter will filter the current spikes from the controller and will generate a small over load, but this will be absorbed by the controller without any damage (the spike intensity is smaller than the injected current intensity and the controller is in low impedance when it stops driving the output, so there is no power and therefore no heat and thus no issue).

The real question is the range, but a factor 10 of momentary overload should not be an issue as the average value remains within the specification of the galvanometer.

MAC
  • 361
  • 1
  • 5