PWM motor controller interfers with strain gauges

I have not used commercial off the shelf instruments, but have rolled my own strain gauged posts and amplifiers.

There are only two ways that the interference can occur. One is radiated. The other is conducted. If you have shielding in place, then I suspect that the interference is induced by conducted means.

Start with using two different power sources to see where the pathway is. You need to look at all the different possible ways that interference is getting into the strain gauge amps. The solution might be simply filtering the amp power supply system.

The strain gauge amplifier and the motor controller are running on separate power supplies and I have been unable to find any common points on both circuits where conduction could occur. Looks like the inference is induced through radiation of PWM signal as it passes through the motor.

The strain gauge amplifier and the motor controller are running on separate power supplies and I have been unable to find any common points on both circuits where conduction could occur. Looks like the inference is induced through radiation of PWM signal as it passes through the motor.

This (separate power supplies) may, in fact, be part of the problem. Try using a common connection on the power supplies. If the PWM amplifier is unipolar, then the strain gage common and the PWM common will be at the same potential. Of course, proper shielding (only one end of each shield connected at the common point) is also in order.

Is this home-made, old gear, or new stuff?

If new stuff, the motor & drive should be covered by regs which prohibit emission of noise which would interfere with the strain gauge setup.

Also, the strain gauge stuff shouldn't be affected by radiated or mains-borne pick-up. (This is a bit of nonsense, really, as virtually any sensitive measuring system is going to be prone to pick-up from all sorts of things - so it's back to the motor drive).

If you've followed the PWM board suppliers installation instructions (to the letter), they should be held responsible for emitted noise.

Can you give answers to my first question? It'll give us a clue which way to go.

The motor controller was bought in new from RS-Components.

http://uk.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=3639601

Have you got the PWM controller in in an EMI/RFI screened enclosure? You'll probably also need a filter on the power input, and the motor lead will have to be screened, with 360° connection between the screen & the controller enclosure. A good earth connection to the motor body is essential. The screen should also cover all wiring up to the motor body. A ferrite split ring on the motor lead may help.

Have a chat with the chaps at CJ Controls - they should be able to advise on these things.

I've been through these loops using steppers with chopper drives. Getting rid of the noise is really a black art!

Hi,

I'm not very expert on electric motors, just a bit expert on strain gages measurements, but according to your posts, motor and controller use only DC. If you have different supply for motor and strain gages amplifier I would think on induction. You've mentioned strain gages and full bridge, I suppose that are a DIY installation and not a standard sensor based on strain gages (as a load/torque cell for example).

If this is the case, you surely have screening gages connecting cables, but strain gages itself aren't shielded and therefore not protected from induction. This could be the reason, mostly if the strain gages system is near the controller.

I would try to be sure this induction isn't present. Can you shield the controller? I think it's easier than shielding the strain gages bridge.

You can try to disconnect the bridge from its amplifier and supply it with a cell battery (4,5 or 9 VDC) and measure the output voltage with a multimeter when controller is on and when it's off and see what happen. If the output offset appears, the problem is the induction in the gages system. Otherwise, the problem must be in the amplifier.

Cheers

These DC drives are PWM - that's pulse width modulated. The DC is chopped up at about 20kHz, and the ON and OFF times are varied to get the average current/voltage needed to drive the motor at the demanded speed. The ON & OFF switching times are very fast (to reduce the time the power transistors spend in the linear region, & so reduce power loss through heating).

The resulting asymmetric square wave has noise properties worse than any sine-wave type AC. It's horrible stuff to deal with!

If the controller has a facility to vary the PWM frequency try altering it to see if it makes a difference to the offset.

Otherwise it would be a simpler exercise to use a limier power supply the voltage and power requirements are not excessive, use an LM317 V reg with a couple of Mosfet buffers. Also your voltage and current measurements will be more straight forward.

Assuming that you used a good grounding practices, it is possible that the switching noises gets rectified by the gage amplifier, especially if you use a DC model. If it is a DC model, you could add filtering on the sensor (at the amplifier inputs). It will reduce the speed of response but unless you are looking for mS response time it won't be an issue. Make sure you use good quality film capacitors to minimize the leakage.

If no one of given recommendations works I would suggest you to check insulation level of strain gage contacts relatively metal surfuce. Properly attached strain gage should posses isolation level between its contacts and metal measured in range 4000-5000 MOhm or so. Crucial level for properly working gage is about 1000 MOhm. 600 MOhm means that strain gauge should be thrown out as broken.

For measuring you need to get MegaOhm meter which provides voltage measuring level about 100V. Strain gage cable should be unplugged, i.e. you have to conduct measurements between every terminal contact of bridge circuit and metal body. I had a cases in my practice when cable isolation had been broken as well. But it's hardly. I do not think so.

Raise the shields Scotty er Ludsey!

Okay, back to Earth.

I've worked with strain gauges but don't like them very much. They require a lot more care with installation than load cells. In fact, I've replaced every strain gauge system in my factory with load cells. My difficulties with them don't involve interference, however.

Load cells are actually pieces of metal with strain gauges built into them. One thing that differenciates them is the fact that the strain gauges are inside and shielded from the outside environment. Plus, the cables are sealed and shielded as well.

With regards to your PWM controller, make sure that your chassis is grounded. I would also try using shielded cable for the motor's cabling as well. Careful routing of the cables away from the strain gauge's cabling might also help.

Vulcan's suggestion of shielded cable for the motor leads is excellent. Let me add that the motor ground (bonding) connection should be one of the insulated leads within the shielded cable. Then connect the motor ground and the shield together as close to the PWM speed control as is physically possible--preferably on a ground terminal it may have immediately next to its output terminals.

That was the only way I was able to solve a similar problem with a 4-20mA transmitter signal and a VFD. There will be noise on the ground, and the only way to keep it from interfering with other systems is to return it as close as possible to its source.

Regards--JMM