PWM motor controller interfers with strain gauges

[Joeydee]

Test apparatus includes a DC permanent magnet motor and strain gauges. The speed of the motor is controlled from a PC through a pulsed width modulation controller board. When this motor is running at anything other than full speed, there is a large offset in the amplified strain gauge output (up to 4V). The strain gauges are connected in a full bridge configuration and all standard screening techniques have been tried with no success. But when the PWM motor controller is bypassed, the offset is no longer present. Is it possible to filter the PWM signal or is there another method of speed control?

 

[itsmoked]

That's a tough one.

Do tell us more.

Voltage/current of the motor.
What is the PWM driver rated at, current wise?

PWM freq.

Do the sensor and the motor leads run together anywhere? How far?

What is the sampling freq?

What exactly have you done to try to mitigate this so far?

The PWM waveform is probably very steep as this reduces losses and heating in the switching elements. Conversely this greatly increases the number and height of the frequencies present in the drive cable. This increases the radiation out of the drive cable. If your drive is not heavily loaded you can sometimes slow the rise times and drop a large amount of noise radiation.

You could try a clamp on choke.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[Joeydee]

Motor: 24 V DC, 6 A

PWM: 15 A, 17 kHz (measured approximately from ocilliscope)

The stain gauges are attached to a metal support which is mounted on a metal tube, which houses the motor. The gauges are insulated from the support. The motor controller and strain gauge amplifier have separate power supplies. The amplifiers are located as close to the bridge as possible and cables are only about 6in long.

So far, I’ve tried connecting in a resistor bridge instead of the strain gauges and the problem disappears. This indicates that the strain gauges are picking up the signal, not the cabling. Capacitor across the strain gauge bridge supply and output do help, but not enough.

How would I slow the rise time of the PWM waveform? What’s a clamp on choke?

 

[itsmoked]

Joeydee; Nice description there. Thanks.
Your test is a nice one too.

Was your replacement resistor bridge about the same resistance values as your bridge?

Changing the PWM slope would require more work than I would bother with at this point.(partially premised on you being the designer)

Make sure your metal tube/structure is solidly grounded. Use a "star" topology. The system ground to a point. From that point a ground lead to your PWM Drive. And one to your bridge amplifiers.

You want to avoid ANY case where the drive power(capacitively coupled) can share a ground path with the bridge stuff.

Is your motor bi-directionally driven?

Try a 0.01uF cap across the motor terminals.

Are the motor leads twisted? Should be.


Clamp on choke.. You've seen them. They are on various cords that exit your computer gear. Likely your LCD or CRT monitor has one on the video cable. The lump.

Top of the page.

http://www.jameco.com/Jameco/catalogs/c281/P165.pdf

Likely the bridge is acting as an antenna to the magnetic fields pouring out of your motor.

If these are in close proximity you may need to actually physically shield the bridge.



Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[BobM3]

Do the four gages (wired together) form a big loop? If so, try rerouting the connecting wires to minimize the loop area.

 

[GonzaloEE]

These are my 2 cents:

-Check for ground loops.
-Keep separate ground planes for PC input signals, PWM board output and Motor driver.
-Shield PCB tracks, pouring GND plane under any loops.
-Prefer ceramic and tantalum caps for VCC de-coupling.
-Agree using chokes to isolate motor current pulses from logic.

Good luck!

 

[sreid]

Is there any conductive connection between the motor case or shaft and the metal the strain gages are mounted on? With PWM driving the motor windings, there are large capacitively coupled spikes from the motor windings to the motor case/shaft.

 

[BobM3]

Have you looked at the output of the bridge on a scope. How does it look compared to the PWM waveform?

 

[felixc]

Can a magnetic coupling from the motor to the casing induce such effect?

If you drive the motor from a DC source, do you also get an offset on your readings?

 

[itsmoked]

felixc;

When this motor is running at anything [blue]other than full speed[/blue], there is a large offset in the amplified strain gauge output

He gets NO offset with his present equipment if the motor is not in PWM.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[GonzaloEE]

I'd look around your PWM controller board, a kind of 'black box' to me. Just handling a little some ideas posted before:
- Try isolating power sources for Motor+PWM board, and strain gauges, to isolate noisy VCC or ground loops problems.
- If possible, try powering the PWM boards without a motor.
- Does your PWM board use some special switching-PS that could add noise to power rails? Get some 0.1ohm resistors, and check V and I waveforms at some supply points, with a scope.

Good Luck!

 

[GonzaloEE]

Try running the motor and PWM board with a battery, and drive it from a laptop. I mean completely isolated from the strain gauge supply.

Good Luck!