Motor Shaft Currents

Originally Posted by pilko »
"I am currently replacing bearings every 3 months on my furnace fan motor".
Reply by cachehiker >>
"I'd be embarking on a quest to find a better fan motor.

There is simply such a thing as motors that have been optimized (cost reduced) for strictly pure DC or 50/60Hz AC applications. Bearings in such motors will simply not stand up to use with inverter/PWM drives when the motor frame is grounded. There are times when even isolated mounts and shaft grounding brushes aren't enough.

The biggest thing to remember about rectified line voltages is both supply rails end up hot. With 230VAC, the positive one will settle in at +160VDC relative to ground and the negative one at -160VDC. The processor will likely share its negative supply rail with the rectified line. Pushbuttons, displays, potentiometers, and such that are connected to the processor will be shock hazards unless they're optically or otherwise isolated. Disconnecting the processor from the mains for a while (depends on how fast it discharges) before connecting it to the USB port or your computer for programming is MANDATORY unless you have and abundance of computers at your disposal. Connecting a grounded computer to a hot rail will let the magic smoke out. You might be able to salvage some drives or some memory, but the motherboard will likely be roached. This makes debugging a major pain so concentrate on writing robust or easily testable code the first time".

@ cachehiker,
What do you think about filtering out the PWM frequency with a low pass filter?

pilko

I see where you are coming from now, but let me explain the source of the confusion. In post #4, cachehiker said:

If I wanted to keep the noise down, keep the efficiency up, and control speed, I'd be modulating a DC supply with a 16-20KHz PWM signal. An electrically isolated mount or some other means to limit current flowing through the bearings might be needed though. Better efficiency implies more torque before you stall or overheat.

Click to expand...

bigal_scorpio then asked,

DC seemed ideal but now you have me worried! What is the current flowing through the bearings this is new to me. Now I am concerned about it, is it dangerous? Can you elaborate on the isolation you mention?

Click to expand...

I responded that I wasn't aware of a design that had currents going through the bearings. My intent was indicated by the word "design." Motors aren't designed to have currents used for power to be conducted through the bearings. In retrospect, it is obvious that you were talking about the induced currents from using high frequency switching. I was aware of that effect, as I am sure was Nigel. I was, and apparently so was Nigel, simply talking about design, not the use of non-inverter rated motors with inverters. The follow-up comment about using a cheap motor bears that out.

Now on the point of inverter-rated motors and non-rated motors, it is well known that a non-rated motor may have problems when used with an inverter. Here is a DOE statement about the effects:

https://www.electro-tech-online.com/custompdfs/2011/12/motor_tip_sheet14-4.pdf

Additionally, here is a very interesting discussion of induced motor bearing currents:
https://www.electro-tech-online.com...oltages_Bearing_Currents_Report2_2nd_ed-4.pdf

Sorry for the misunderstanding.

John

That second article is a good one, about as good as I've seen.

I have no choice but to be a little suspicious of motors scavenged from equipment. You look at stuff sold a big box stores on the cheap and you know the manufacturers have bent over backwards to shave that last 10 cents out of the design. I currently work with what would be best catagorized as consumer appliances and the decision makers insist on bearings barely big enough to get through the warranty.

Pull some motor out of what is advertised as a "Heavy Duty" Harbor Freight or Walmart bench grinder or whatnot and it's expected life is all but guaranteed to suffer with an inverter drive. How much is difficult to say. If you're on the steep part of the curve, a relatively small increase in bearing currents can conceivably bring the MTBF down by a factor of 10. If you're at either end, the effect is much less dramatic. I have a scanned copy of the curve at work. I'll see if I can find it and post it.

Thanks jpanhalt and cachehiker.
The motor is a new Leeson 1HP, PM shunt, U-Frame, Class F ins. I am controlling it with a KB, PWM drive package.
I have been simulating a low pass filter and I am currently bench testing.

Regards
pilko

I found it! I don't remember which book the attachment came out of but it was circulated here at work while we were trying to work out what was suddenly happening to our bearings.

With the PWM generator load "upgrade" on our brush life fixture, we were probably pushing 1.5 A/mm∙mm given that our bearing life had dropped to about 400 hours. Isolating motors in the end product except for a single grounding wire and measuring the current flowing in it with a current clamp pointed to further fixture modifications to match this measurement which was probably equivalent to 0.9A/mm∙mm. We haven't had a bearing failure since which represents an improvement by a factor of more than 10. The brushes now wear out long before the bearings do.

I realize that the attached graph is probably little more than a quality approximation, but 40% less bearing current density equalling 10 times the life is what I'd call the steep part. In fact, it doesn't differ that much in appearance from the fatigue curves plotting peak strain vs life for aluminum that's being pushed too hard.

Of course if you're doing nothing in particular out of the ordinary (we definitely were) and you're already pushing 1.5A/mm∙mm through the bearings, then your bearings are probably way undersized and aren't going to last to terribly long under the mechanical load imposed upon them either.

@cachehiker,
Thanks for the info.
What was your final salution? --was it complete isolation of the motor from ground except 100k resistor to ground, or was it another salution?

pilko

The temporary solution was to connect the motor to ground through a 2.7KΩ resistor so the bearing leakage on the fixture was the same as that in the end product. It was initially much higher because we essentially had two motors with their frames connected and both of them were being pulse width modulated. The two motor/PWM combinations had a push pull effect going on which nearly doubled the bearing current. The 2.7KΩ resistor effectively halved that.

The final solution will be to get rid of the second PWM signal altogether. A plain old automotive alternator with a custom linear regulator controlling its field current will take the place of the old, hard to find, heavy duty, PMDC generators connected to pulse width modulated loads. As an added bonus, there will be no old, hard to find, heavy duty, PMDC generators any more. We'll be able to pick up spares at the auto parts store instead.

@ cachehiker,
With the motor "floating",I am reading 45V ac, on DMM from motor frame to ground, and a short circuit current of 30mA ac from motor frame to ground.

pilko

Bump --------------

From what I've seen, 45 VAC is about typical for a PWM control driving an isolated motor from a rectified 120 VAC bus. 20mA is a bit high if it's rms. What are you using to obtain these measurements? How big are the bearings in the motor? Are both of the outside bearing races grounded through metal end bells/caps?

I'll have to give the motor(s) we were having trouble with a second look as I don't remember the OD of the bearing that was taking the brunt of the abuse. IIRC, the ID was 15mm and they were 10mm wide. We also had 20mA flowing through just one of these bearings. The rear end bell/cap was plastic and effectively isolated the other bearing. If it had been 10mA through each of two bearings it's unlikely that we would've seen bearings failing before the brushes wore out.

The A/mm∙mm will be approximately proportional to the inverse area of the outer race. It will also be approximately proportional to the inverse switching times of your fet/igbt. Substituting 300Ω gate resistors for 100Ω gate resistors increases switching times and lowers our bearing currents but also raises the operating temperatures of the fet/igbt's. Based on this, I'll bet actual filtering of the output would have an even greater effect on bearing currents. I never experimented with that. Even if it was successful, it would've been too expensive to implement in the end product.

You don't need to do much, less than 50% reduction, to go from 3 months of bearing life to 3 years of bearing life.