Low current kills brush life? WTF?

Hi,

Just a guess, but maybe it has something to do with the arc vs time profile.
At a given speed the arc will extinguish in a certain amount of time t1. As
the speed is decreased via less current, the arc lasts longer and the current
isnt that much less than before so more wear results. Arc's are not linear at
all in nature, so there could be a big variation at some places on the
characteristic and much less at other places.
Another guess would be that the phenomenon is related to the gap length,
where certain arc times will cause different results with different
tangential speeds.

Just a little curious here too: did you ever try an air pressure stream to
help extinguish the arc much faster? Is this a possible design enhancement
that can be implemented on your system(s) ?

Appreciate the idea, but I'm not convinced the arc is more of an issue at low voltages than higher. That arc depends greatly on the total energy involved, which for the windings is proportional to the square of the current. So, reducing that current greatly reduces the energy in the arc.

Anyway, in this device the rotor has an integral MOV, like I said. This is a ring-shaped MOV with 3 taps in parallel with the 3 commutator segments. It extinguishes the arc by shunting the current when the voltage exceeds about 30V. At 3×10^6 V/m this corresponds to an arc length of .00039". An MOV supposedly has a response time in the tens of nanoseconds. The arc is not as significant an effect in these motors as it is in ones without it, and I have a fair amount of experience with them. In motors that lack it, the arc is very visible and MTBF is a fraction of the motors that have it.

The brush wear in this case would seem to be a combination of friction and some sort of plating effect, electrical action on the oxide film that exists on copper surfaces - something along these lines.

Try viewing the brush noise with a CRO when running at both the voltages you are testing.

You seem to be hypothesising a lot about what happens at each voltage but not testing much. What the 2 revs? the 2 voltages? the 2 average currents? the shape of the 2 current waveforms on the CRO? etc.

duffy said:

Appreciate the idea, but I'm not convinced the arc is more of an issue at low voltages than higher. That arc depends greatly on the total energy involved, which for the windings is proportional to the square of the current. So, reducing that current greatly reduces the energy in the arc.

Anyway, in this device the rotor has an integral MOV, like I said. This is a ring-shaped MOV with 3 taps in parallel with the 3 commutator segments. It extinguishes the arc by shunting the current when the voltage exceeds about 30V. At 3×10^6 V/m this corresponds to an arc length of .00039". An MOV supposedly has a response time in the tens of nanoseconds. The arc is not as significant an effect in these motors as it is in ones without it, and I have a fair amount of experience with them. In motors that lack it, the arc is very visible and MTBF is a fraction of the motors that have it.

The brush wear in this case would seem to be a combination of friction and some sort of plating effect, electrical action on the oxide film that exists on copper surfaces - something along these lines.

Click to expand...

Hi again,

You said that you had turned the voltage down by 0.2v and saw a big
change in wear. This is not a large current change. This is what i
was referring to.
You are also saying that there is a mechanism for reducing the arc
already in place and that reduces wear as compared to the more
unsophisticated devices...this would say that the arc is responsible
for at least some wear, and just because the mechanism helps it
doesnt necessarily do everything. Another, additional technique
may be required to see a longer brush life.
I dont have anything in front of me to test so i can only guess at this
point, but as MrRB had mentioned it is sometimes better to do a few
tests than to constantly come up with reasons why nothing will ever
work because that will always lead to no improvement
If it is the friction then you have to look into some method to reduce
that friction.
Sorry i dont have any more direct info on this problem at hand.
Im just throwing a few ideas out there.

Maybe make the motor mount snap in so it's fast and easy to replace

I ALREADY mentioned the 2 voltages AND the two currents in this thread. I looked at several waveforms on the scope, including the motor. I measured the change in cycle time at the lower voltage, which essentially gives me the change in revs, so that I could change my software routine that shuts the sevo off after the maximum travel span.

I also measured the effect of the change in temperature on the angle of the servo (as mentioned) and we measured the effect of elevated temperature on MTBF. We have both a static and dynamic life tests, as mentioned. I have clearly been doing a fair amount of testing!

Now, perhaps you would be good enough to tell me how to use that servo rev and scope data to increase MTBF? You didn't mention that part.

My question (yet again) is if anyone has any experience or knows anything about low brush current being detrimental to brush life.

duffy said:

<snip>
I lowered the servo voltage to 3.5V. Could lowering that voltage by .2V really cost us some 200k off the life? I'm looking for other explanations - we didn't chart humidity with our life tests, perhaps when we did the earlier tests in the winter, when the humidity was lower...

Doesn't sound like anyone has any information on this, (figured it was a long shot) but I appreciate the replies.

Click to expand...

Hi again,

That 0.2v statement was what i was referring to. You say that lowering
the voltage by 0.2v costs some 200k off of the life. I said that i didnt
think that the reduction in current should be that significant for a change
of only 0.2v, but maybe it is. I wasnt however referring to your other
posted numbers.
Yes, this is a very specific problem that will be hard to answer unless
someone has already had this experience. It may be that you are
looking for too much too, but then i can see why you are trying.
If you say the arc does not appear with the arc mechanism then
that's the end of that story.

The only other idea i can come up with at this time is that maybe it
has something to do with the 'riding' friction. The sticking friction
should be done with as the part starts moving and then the sliding
friction is the key factor, but at lower speeds it could be that the
sticking friction starts to (again) play a part in the total friction
even though it is not stopped completely.
Another idea might be that some sort of resonance is coming into
play. There may be a critical speed that causes (or destroys) this
resonance as the brush moves slightly in and out of the holder.
With the right speed the brush may ride up high enough to lower
the overall pressure, but with a lower speed the brush is able to
push down a tiny bit more because it now has time to do so.
I think changing the spring tension (if possible) might show a
difference.
You said you wanted to hear other things if no one had any experience,
so that's about it for now

BTW the tests i meant was with some change to the arc, if possible.
The tests have to be the right type as well as enough to draw any
conclusion from.

Just for the record, the phenomenon of copper brush wearing can be
detected using audio, but then again that doesnt help to stop the wear.

I think we can at least narrow down the factors that come in to play
in order to limit what has to be looked at to get an answer. It must
be either arc or friction, unless anyone else has any more ideas.
To test for arc wear you would have to find a way to eliminate the arc
completely, such as with air, unless of course you are already seeing this.
To test for friction wear, you would probably have to loosen the spring
tension or find some sort of lubricant which im not even sure is made
for this kind of thing.

You might be able to find a forum that deals with motors more than
anything else. This forum is mostly electrical and electronic.

One other thing i could do is ask around a little, if you have the
patience to wait a little while.

I like that resonance idea! Could be at one particular speed the brushes vibrate over the commutator gaps at some mechanical resonant frequency, and raising or lowering the voltage by just a fraction enhances or destroys that resonance. The brushes even resemble little tuning forks.

Here's a couple of pictures of the brushes themselves. These have some wear but are still good. A mechanical resonance mode at a lower speed that would knock 20% off the life makes more sense than some hinky surface film issue that only happens at low current.

duffy said:

I like that resonance idea! Could be at one particular speed the brushes vibrate over the commutator gaps at some mechanical resonant frequency, and raising or lowering the voltage by just a fraction enhances or destroys that resonance. The brushes even resemble little tuning forks.

Click to expand...

Yes, that's the theory. The brushes are acting like little spring/mass/damper
systems with resonance.

If this what is causing the spike in wear then a simple (well sort of)
experiment would be to add some mass to the system to lower the
resonant frequency, and perhaps push it down low enough to move
it out of the normal operating range.
If you could get a tiny dab of solder on the two brushes without
hurting the spring characteristic too much (hopefully not at all)
this would lower the frequency quite a bit because there
is very little mass there to begin with.
Unfortunately this would mean also testing the motor to check for
wear. This brings up the question, what kind of test equipment do
you have on hand to measure the brush wear?
I guess i should have also asked how did you do all the other
tests? I guess i also have to ask, how long would it take you
to do another test after this little adjustment in mass?

If it seems like soldering the brushes would be too harmful,
a dab of epoxy might do the trick too. I just thought that
solder would have the most mass for something relatively
easy to apply and would take the frequency down the
lowest per unit volume.

Can you get the commutators made physically larger and the brush assemblies made thicker or wider.
Its common to see industrial DC motors with brush and commutator designs that are two or three times bigger than a lower life expectancy standard issue motor.
I have a number of industrial DC servos that have commutator rings nearly as large in diameter as the rotor and with brushes that can carry 20x the rated motors running amps.
I have also found that design in some very small long life DC motors used in office equipment built before brushless DC type motors became common.

Just a thought.