Under-loaded and cold ball-bearing extremely high friction?

[alsostefan]

We're trying to find the cause of very high losses in a custom high-speed brushless motor. The losses occur at light/no load with a cold motor. As soon as the motor heats up to normal operating temperature (approx 80 deg case) all is good. The motor's rated speed is 12000 RPM (at 1kW).

Temperature measurements seem to point at the oversized front bearing we're using to make these motors survive the massive amount of vibration from the (unavoidably unbalanced) tool it's swinging around. The front bearing is the first thing that heats up according to our sensors, so either the bearing or the shaft is getting hot quickly.

The bearing is an NSK 6304 and it is under-loaded running the machine at working speed but unloaded. The static friction of the bearings we verified to be very close to the ideal as calculated by SKF's (web) prediction tool even when cold. The hot running friction seems pretty ok too, SKF's tool predicts about 11-45W of losses depending on load at given the power consumption that looks not too far off. But a 'cold start' we're losing about 300-400W somewhere for the first minute or so... Can a skidding / under-loaded bearing have 10 to 40 times the friction torque of it's cold 'starting' torque?

 

[JJPellin]

How are the bearings set up for axial thrust? Normally, one of the bearings would be fixed axially and the other one floating. If they were both fixed, you could be placing the rotor in tension and loading up the bearings in thrust until the rotor heats up. Some motors use wave springs to place a light axial load on the bearings even when cold. Does your design incorporate any spring loading mechanism for axial loading? How is the bearing lubricated? Is this a single speed motor or is it variable speed? What speed are you running at start-up when the high load condition is noted?

Johnny Pellin

 

[alsostefan]

> How are the bearings set up for axial thrust?

Both bearings are press-fit into an aluminium flange. The much smaller rear bearing fixes the shaft axially (nut on shaft clamping the inner ring). The front bearing is free floating on the shaft (can move it by hand when cold, but not enough to spin the shaft in the bearing).

> Some motors use wave springs to place a light axial load on the bearings even when cold.

This motor doesn't use any wave springs (in an attempt to reduce running friction and save battery on extended runs).

> How is the bearing lubricated?

NSK standard grease.

> Is this a single speed motor or is it variable speed? What speed are you running at start-up when the high load condition is noted?

It's meant to run at a fixed speed (with varying load). The exact speed is a little dependent on tool specifics, but say 12000-15000 RPM. The controller is set to run that regardless of load, including at startup.

 

[JJPellin]

Interference fit on the outer race with a stationary housing is not common. I guess the massive imbalance makes this necessary. I suspect that you have excessive interference on the bearing OD to the aluminum housing and that the bearing heats up at these high speeds before the aluminum heats up enough to relax the radial overload condition. What fit criteria are you using? A standard ISO H6 fit could have as much as 0.0013 inch interference. I would prefer to stay below 0.0005" interference unless you know that the growth of the aluminum is going to take away all interference at running temperature.

I think your bearing is experiencing radial overload from a combination of the outer race interference, the dilation of the inner race at the high speeds and the rapid thermal expansion of the bearing balls at high start-up speed. It becomes a runaway process. The overload generates heat which causes more overload which generates more heat. I would reduce the interference fit on the OD as much as possible and use the controller to ramp up to full speed gradually.

Johnny Pellin

 

[TugboatEng]

12000 RPM is quite a high speed for a 6304 bearing. I must ask what lubricant you are using?

 

[alsostefan]

So far the NSK standard grease. In a future revision we might reduce the bearing size a bit, but right now we aimed to have these motors run reliably for a good amount of time (ideally 20k hours). All regular motors tested died in a matter of months (approx 100 hours), wrecking both bearings and their housings in most cases. Besides the high cold losses the motor does seem to run beautifully once loaded up. It's not at all certain those cold losses are due to the bearing (alone), hence the original question if it's even possible.

 

[TugboatEng]

I do suggest cruising through this bearing selection guide. We seem to have jumped to step 6.

https://www.skf.com/us/products/rol.../bearing-selection-process/bearing-interfaces

There are a few things to consider based on application. If high vibration and load direction reversal are requiring an interference fit of the outer race you really should consider making the housing material out of steel or other material with similar CoTE so that your interference doesn't change so much over the operating temperature range. Also consider using a retaining compound instead of an interference fit.

Angular contact bearings can operate with preload and will be more tolerant in your application.

Use heaters to preheat the motor prior to operation.

 

[electricpete]

If skidding is the problem, one way that cold temperature plays into that is the viscosity of the base oil within grease. The viscosity is highest at low temperature. Skidding occurs when friction overcomes traction. Loading helps traction to avoid skidding. Low temperature causes high viscosity and high friction making skidding more likely.

If internal clearance is a problem, maybe one option to explore is larger internal clearance class bearing. (C3 instead of C0 or C4 instead of C3).

=====================================
(2B)+(2B)' ?

 

[romke]

NSK standard grease is not a clearly defined term. Could you more specifically indicate what type of NSK grease is used in the bearings? NSK supplies quite a few sophisticated greases that do differ quite a bit in thickener type and in fluid type, fluid viscosity and viscosity index of the fluid used and therefore are meant to be used in quite different applications.