0
\$\begingroup\$

I have a fairly simple project coming up. My task is to design a system which powers and controls a Pan and Tilt positioner for a camera. It has some weird and unavoidable requirements/constraints:

  • No telemetry allowed in the Pan and Tilt positioner. Power must go straight from a control box, through a wire, directly to motors.
  • PWM or AC should be avoided. The Pan and Tilt conductors will have to be in the same Minikevlar cable assembly as some low-voltage analog signals, and PWM would likely be too noisy.
  • As the title suggests, speed control is not a requirement.

The Pan and Tilt will likely contain 24V brushed motors for both axes. My initial approach would be to use a switch joystick to open/close a circuit for each motor in forward and reverse, and supply it via a 24V+ DC power supply (I say 24V+ because I will be accounting for Vdrop). I would also need a current limiting device of some sort to protect the motor when it reaches end-stops.

My second idea would be to use PWM anyway, and if it causes interference in the sensitive low voltage signals in the cable, I could keep changing the PWM frequency until I find something that works (while driving the motor properly, of course). However, if I keep the PWM duty cycle at 100%, I feel like it defeats the purpose.

What are your thoughts, and how would you approach this? I'm assuming my solutions are missing something important because my experience with electronics is not quite as extensive as I want it to be.

\$\endgroup\$
7
  • \$\begingroup\$ This sounds like you are a victim of a "make it someone else's problem" approach to requirement writing. If you really don't need speed control, a power supply with overcurrent trip could be an option. You'll need to think about how you make it so that after the trip you're allowed to go in the other direction still. And you need to worry about how you're going to distinguish huge startup current draw from hitting the mechanical endstops. Also beware that a brushed motor makes its own electrical noise! \$\endgroup\$ – Chris Stratton Aug 11 '20 at 15:12
  • \$\begingroup\$ What are the motor specs? Is there a maximum and minimum length of cable that you can depend on? Do you care about wasted power? If not, consider analog voltage control for the motors. You will need current sensing to limit the current. You might want to signal the operator when current limiting is active. \$\endgroup\$ – Charles Cowie Aug 11 '20 at 15:32
  • \$\begingroup\$ As mentioned, a brushed motor creates significant electrical noise regardless if it's straight DC or PWM driven. My thought is forget about designing one and go source something pre-fabricated. I don't believe for a minute that you can't find a commercially available something that will do the trick for you. BTW, 100% PWM is not PWM, it's DC ;) \$\endgroup\$ – Kyle B Aug 11 '20 at 16:17
  • \$\begingroup\$ @CharlesCowie The Pan & Tilt hasn't been selected yet. The one I choose is going to depend on price, lead time, etc. But a majority of my choices use (or have the option of using) brushed 24VDC motors. The cable length is going to be ~150', and I do not care about wasted power. Although, massive heat sinks and fans would also be somewhat awkward (but not impossible) to integrate. \$\endgroup\$ – NuclearDude Aug 11 '20 at 16:19
  • \$\begingroup\$ @ChrisStratton I didn't think of startup current. I'm assuming its order of magnitude will be similar to stall current, which might make this tough. Perhaps if there is a time-delay in the current limiting feature? Startup current stays high for a very short time, but a stall condition would probably exist for much longer. Sounds like a job for a microprocessor at this point \$\endgroup\$ – NuclearDude Aug 11 '20 at 16:22
1
\$\begingroup\$

[edit: assume electric-field coupling between the 24 volt PWM wires, onto the low_level analog. Assume a HIGH_PASS FILTER: a series Cap into a shunting resistance (the R_sensor, or R_source of the analog signal) to Ground.

Edit:We compute the edge_speed of the PWM, to reduce the 24 volts down to 0.001 volts, using the High_Pass Filter to do attenuation.]

Let us model the trash injection from fast-edge PWM signals, into the sensitive low_level analog.

Assume 10pF per foot between two wires (coax is about 100pF/meter).

At 150 feet, there is 1,500pF of electric_field energy storage.

Assume the low_level analog has 100 ohm resistance.

What slewrate (rise and fall time is permissible, in 24 volt PWM lines?)

For 1 millivolt induced noise into the analog, with 100 ohms, the limit is I = V/R = 0.001/100 = 10 microAmps.

Now we just compute the edge_speed (you might use L+C filter to slow down the edge from a motor driver, to have a SLOW edge out on the cable).

dV/dT = I/C (which comes from Q = C * V, and differentiate to get I = C * dV/dT

  • dV/dT = 10uA/1.5nanoFarad = 1e-5/1.5e-9 = 0.7e-4 or 7,000 volts per second.

And the actual rise or fall time needs the 24 volts in the math

  • Trise or Tfall = 24 volts / 7,000 volts/second = 24/7,000 seconds

And Trise or Tfall, to induce only 1 millivolt on 100 OHM_resistive line

  • Trise, Tfall = 0.01 * 24/70 or about 3 milliSeconds

To have a PWM produce pulses that slow will be VERY INEFFICIENT (causing lots of heat).

But you can impose an L+C filter (series L, then shunting C, then the cable to the motor). You will need a Damping resistor in parallel with the inductor.

{edit: the suggested solution is

  • efficient PWM driver with perhaps 100 Hertz pulses of variable duty cycle

  • a L+C filter, to reduce the edge rate to 3 milliSeconds, located between the Driver and the 150 foot wiring

  • the unavoidable large capacitance between the PWM wires and the low_level analog wires

  • the sensors producing the low_level analog signals, with *ASSUMED output resistance of 100 ohms or less

\$\endgroup\$
2
  • \$\begingroup\$ I was following the logic until the very end; if we want to find the time it takes to induce 1 milliVolt on the 100Ohm line, wouldn't we just do the following? Trise, Tfall = 1milliVolt/ 7kV/s = 1.43E-7 s. Then, to get the PWM frequency which would cause this, we can just divide 1 by the rise/fall time to get 7MHz. \$\endgroup\$ – NuclearDude Aug 11 '20 at 18:19
  • \$\begingroup\$ Also, wouldn't increasing the PWM frequency and decreasing duty cycle reduce the induced voltage? By having the PWM pulse be "on" for a very short time, you reduce the amount by which the voltage can grow (assuming that 7000 V/s is constant). \$\endgroup\$ – NuclearDude Aug 11 '20 at 18:59

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.