Electrical motor noise affecting digital meter reading

[mmagliaro]

I have some LCD panel meters (bought off eBay), mounted in a little plastic housing
on my workbench to monitor voltage and current from a home-made power supply that I use for testing engines
and motors.   It works great.

I happened to be testing an old motor from a Trix engine, and the volt meter went absolutely nuts when I got
up to about 6 volts (readings jumping all over the place, reading utter nonsense, and so on).
The motor was running fine, and hooking a conventional handheld digital VOM on the output of the power supply,
I could see that the voltage was just fine.

I tested a bunch of other motors (Kato, Atlas, coreless, etc).  The display was fine.  I concluded that the little old
3-pole Trix motor is probably very electrically noisy, so it was radiating some RF that was somehow affecting the
panel meter.   In fact, clipping a .01uF cap across the motor leads got rid of the crazy meter readings, so I'm reasonably
sure that's the problem.

Question:  What is the solution?   Encase the meters in a metal grounded box?   I suppose that might help, but
they have a large plastic front display, which obviously can't be shielded.  Also, placing the cap on the power supply
terminals does not help.  It has to be placed across the motor itself.  So I can't put a cap on the supply outputs to
get rid of this problem "for all motors".

The meters are powered by 9v batteries.  I am wondering if the "floating" ground of the battery is the issue.
Perhaps connecting the negative battery terminal to ground would help?  I am a little worried there, because these
meters had instructions saying they MUST be isolated from the measured circuit and from each other.   Still, I don't see
how connecting the negative battery terminals both to ground should hurt.

Thanks for your suggestions.

[C855B]

DO NOT connect the meter's battery (-) to the measuring (-)! These circuits are designed to be isolated this way, for your and everybody else's safety, and accuracy of reading. They are most definitely not be used with a common-rail ground.

I would not assume RF interference, although stranger things have happened. My take is that back EMF is putting enough of an AC waveform on your measuring leads to overwhelm the meter's anti-aliasing compensation. I would want to put an oscilloscope across the meter leads to verify this. That the .01µF cap works at the motor but not at the meter terminals confirms the AC component, where the cap at the meter is far enough back in the circuit to become part of the impedance of what amounts to a low-frequency feedline, rather than simply shunting the spikes.

Short answer, however: bigger cap across the meter leads. Get the biggest value you can in a non-electrolytic.

[mmagliaro]

Thanks for the info, Mike.
After I wrote that, I realized that grounding the - battery terminal would do no good anyway.  After all,
in a standard hand-held DVM, which I also tried, and which does not have this noise problem,
the internal 9v battery is not grounded to anything.  Not to worry, I won't do it.

As for the capacitor, I actually already did try using a larger value across the meter leads (not at the motor).
I went all the way up to a 0.47 uF, but it did not abate the problem.   Are you thinking I need something really large
like 100 uF, (like a bipolar electrolytic)?

[C855B]

I've long forgotten the freq vs. value calculations to determine the right value for the capacitor(s). You might try going a little smaller on the cap (.001), and doing both a .47 and the .001 across the meter leads to trap both low- and high-frequency components. A choke in the (+) lead may also help. You've got some serious noise going on there.

[mmagliaro]

Thanks, Mike.   I don't see how putting the two different valued caps across the meter leads makes any sense, though, since they would just be in parallel and create a single capacitance value.

I tried an 0.47 again - no difference.
I put two 100 uF electrolytics back to back, so I end up with a 50uF bipolar cap.  That helped significantly.
I was at least able to read steady values, but they are a couple of volts off (i.e. reading about 10 when a separate
meter and an analog meter both tell me it's more like 12.5, and when I know that the LM317 circuit supplying the output
does indeed put out about a maximum of 12.5 under load.

This solution does not work as well as putting a small-valued cap, like .01uF, directly across the motor terminals, so I really think this
must be a case of RF.  If the cap isn't right near the brushes, then the short lengths of wire coming out of the motor will become an antenna for the motor noise and it will radiate into the room (and into my meter) before the cap can damp them out.

I am not an AC or RF guy by any means, but I suspect that this is what's going on.   The large value cap across the meter may be helping just because it is averaging the spikes and wild swings in the measured value.

Wow... this really tells me a lot about why my HO trains used to drive my parents nuts when they were watching TV back in the 1960s.
FWIW, a Kato, Atlas, LL, or other "decent quality" conventional motor does not cause this problem.  It is only the very
crude, old-school Trix motor that is doing this.

[C855B]

Two separate caps, yes, will parallel to an additive value under DC, but under AC or RF they will each show their own relative impedance in the circuit.

I just had an idea... these inexpensive digital panel meters typically don't have the robust filtering their big brother DVMs have. If you still suspect RF is going places it shouldn't, try putting a cap on the battery leads, which would also have the antenna effect you suggest.

[peteski]

Two separate caps, yes, will parallel to an additive value under DC, but under AC or RF they will each show their own relative impedance in the circuit.

I just had an idea... these inexpensive digital panel meters typically don't have the robust filtering their big brother DVMs have. If you still suspect RF is going places it shouldn't, try putting a cap on the battery leads, which would also have the antenna effect you suggest.

I'm not an electric motor expert, but I play one on TRW!    I agree with C855B.  Real multimeters most likely have some sort of filtering done on their DC voltage input.  The inexpensive panel meters most likely do not (as the manufacturer probably expects the user to add some whatever input conditioning is needed).

Electric motors are notoriously noisy (both electric and electromagnetic radio waves). That is why the European countries mandate the noise suppressing circuit to be installed in models using electric motors.  The noise is generated  mostly by the commutator/brushes (when they are arcing).

Like it was mentioned, a small value cap (something around 0.001 to 0.01 micro Farads should do a good job shunting all the noise spikes.  Larger value (electrolytic) caps aren't good for dealing with high frequency signals.  If you happen to have some of those European noise suppressing chokes (coils) you removed from some Bachmann models, you could also install them in series with the DC leads of your meter (before the the cap).

The other thing I like to mention is that there are digital panel meters where you are allowed to connect the negative input lead to the battery ground. These meters can measure their own battery voltage.  But many meter circuits will not work properly if the negative input is connected to the negative battery lead (ground).

I also agree that the noise picked up by the meter is most likely not due to lack of shielding - the noise is simply picked up by the input leads of the meter, directly from the motor.

[mmagliaro]

The little cap works great, but ONLY when connected directly across the motor brushes, or within about an inch of them.
If I put it a foot or so back on the wires, or back where the probes come out of the panel meter, it does not work.  That
is an indicator that this is an RF problem.  The noise is being radiated off the wires like a transmitter, so putting the little cap across the meter inputs is useless.  The noise has to be suppressed close to its source: the arcing brushes.

While that works, the thing I don't like is that it's not a general purpose solution.  I will have to connect a little cap across any noisy motor I measure.  While I can do that, I would rather find a way to make the meter impervious to the noise.
I don't think that's going to happen.

[peteski]

The little cap works great, but ONLY when connected directly across the motor brushes, or within about an inch of them.
If I put it a foot or so back on the wires, or back where the probes come out of the panel meter, it does not work.  That
is an indicator that this is an RF problem.  The noise is being radiated off the wires like a transmitter, so putting the little cap across the meter inputs is useless.  The noise has to be suppressed close to its source: the arcing brushes.

While that works, the thing I don't like is that it's not a general purpose solution.  I will have to connect a little cap across any noisy motor I measure.  While I can do that, I would rather find a way to make the meter impervious to the noise.
I don't think that's going to happen.

I suppose that the meter might not be shielded at all and it is in fact picking up some RF noise.  While you have the meter hooked up to the running motor, you could try to wrap the meter housing with aluminum foil (leave the display uncovered), then electrically connect the foil "shield" to either negative of the meter's battery or to the negative lead of the meter's input.  See how that works.  Or alternately, put the running motor in a grounded metal enclosure (even one made from aluminum foil) and see if the meter readings stabilize.

Or, for another test, try running that motor from a power supply while the meter is disconnected from it.  Have the motor the same distance from the meter as when you experience the problem. Then use the meter to measure something else not connected to the running motor (like a voltage of a 1.5 or 9V battery). Is the meter reading steady or is it jumping around?

[mmagliaro]

Good suggestions, Peteski.
I ran the motor off a 9v battery, and hooked a simple GOW to my power supply that has the meters on it.
The ammeter and voltmeter measure perfectly, rock-steady, no matter where I position the running motor.

So NOW, it appears that this is NOT RF.  The motor noise must be coming in through the leads to the meter.

But how does that reconcile with the fact that I can put a .01 uF cap across the motor leads, and it squelches the problem completely, yet if I put that cap across the same leads back by the meter, it does nothing?   If the noise were on the wires,
the cap should work anywhere, no?  I thought that putting it near the brushes was important to short out the high frequency
noise before it could travel  down the wires and use them as antennas to "broadcast" the noise as RF?

When I power the motor off the 9v, I still used 3 foot clip leads and no capacitor, so I would have expected lots of radiated noise.

[peteski]

Good suggestions, Peteski.
I ran the motor off a 9v battery, and hooked a simple GOW to my power supply that has the meters on it.
The ammeter and voltmeter measure perfectly, rock-steady, no matter where I position the running motor.

So NOW, it appears that this is NOT RF.  The motor noise must be coming in through the leads to the meter.

But how does that reconcile with the fact that I can put a .01 uF cap across the motor leads, and it squelches the problem completely, yet if I put that cap across the same leads back by the meter, it does nothing?   If the noise were on the wires,
the cap should work anywhere, no?  I thought that putting it near the brushes was important to short out the high frequency
noise before it could travel  down the wires and use them as antennas to "broadcast" the noise as RF?

When I power the motor off the 9v, I still used 3 foot clip leads and no capacitor, so I would have expected lots of radiated noise.

That is a puzzler Max.  Is it possible that when you put the cap across the meter's input (at the meter), the electrical connection wasn't good?

Even more puzzling to me is that only one of several motors you tried is creating the problem.  All the electric motors are basically the same.  I suppose that the design of the commutator (things like gap width between the segments) and the material used for brushes (and maybe even their shape and the force with which they are contacting the commutator) make a difference. But that much difference?

If you have any of the small RFI-suppression chokes Bachmann installs in their recent locos, I would say to try to use them to see if they (along with a cap) would help to suppress the noise.  Install a 0.01 uF or 0.001uF cap closest to the meter input across the leads, then add the chokes in series with the leads. 

[mmagliaro]

No, the capacitor connected at the meter as definitely making contact.  I ran these experiments multiple separate times, disconnecting and reconnecting things.

So, I tried a choke in series with the motor.   Near the motor or at the other end near the meter (makes no difference),
the inductor helped a good bit, but did not get rid of all the bouncing of the meter readings.     My chokes are all just in a little box, so for most of them, I have no idea what their value is in mH.

Still, nothing stops the problem as cleanly and perfectly as a little 0.1uF cap directly across the motor leads.  This leads me to conclude the following:

1. Old Trix 3-pole motors are really noisy (electrically)
2. The noise is most likely getting from the brushes into the wires, and traveling to the meter that way, but it is probably
ALSO radiating off the wires as RF, and some of it is getting in that way.

so...
3. Attaching the little cap right at the motor is the surest way to stop it, because it squelches the source.
Otherwise, I think I am dealing with noise coming into the meter from two sources: the wires, and as transmitted RF.

[victor miranda]

no one is looking....

I have a guess about the unsteady voltages.
lets call it brush float.  if the commutator is out of round...
the motor will act pretty much normal at slow speeds.
as the rotor goes faster one of the out of lobes will toss the brush off
the commutator.

after a lot of use, most 3 pole motors create a triangle shaped commutator
with the area near the gaps being eroded more than the centers.

what RPM the float happens has mostly to do with spring tension,
and then how bad the erosion.

I know slotcar motors do this.
I've not seen it in a 5 pole, it has to be possible.
I know my 0-6-0 did it, I just didn't know what it meant.

the solution is to true up the commutator.

[peteski]

I assumed that (unlike old fashioned rheostat-based throttles), Max's power supply is well regulated. So whether there is a load or not (or a fluctuating load) the voltage supplied by the power supply should remain relatively constant.

Max states: volt meter went absolutely nuts when I got up to about 6 volts (readings jumping all over the place, reading utter nonsense, and so on). The motor was running fine, and hooking a conventional handheld digital VOM on the output of the power supply, I could see that the voltage was just fine.

So only his panel meter shows "nutty" voltages. His digital VOM measuring the same circuit is reading steady voltages.

[victor miranda]

the power supply is not too important.
you can get electrical noise from any brushed dc motor.
arcing is a good way to generate electrical noise.

For the out of ordinary erratic behavior to happen....
the brushes have to come off the commutator.

The erratic behavior of the volt meter
has a lot do do with how fast it reacts to the voltage shift.

I can imagine a lot of things may be happening when
one of the brushes is tossed out of contact.
I am confident The line voltage will jump up.
If the motor is still spinning when the brush lands,
I would think the power from the coils could add to the voltage
or depending on the timing, could pull the voltage down.