Reducing friction in transmissions

[katou]

I am curious to see if anyone out there can suggest how to decrease the friction in a shumaker-type transmission.

Not that I'm all worried about the friction, I'm just curious to know how the real engineers would do it if they had to.

Higher ABEC ball bearings?
Roller bearings instead of ball?
Test to get lowest belt tension that still keeps skipping to a minimum?

I have looked into ABEC bearings at McMaster-Carr, and I can't seem to find flanged bearings that are even rated. And flanged roller bearings? Forget it.

Katou

 

[Miles]

I certainly think it would make more sense to give your attention to minimising losses in the belt/pulleys or chain/sprockets than worrying about bearings types....

Do some tests with no load, to start with?

 

[Lock]

I am curious to see if anyone out there can suggest how to decrease the friction in a shumaker-type transmission.

I couldn't find a "shumaker-type transmission" at a quick google... D'you mean one of these?

... chain, instead of belt... maybe. I have read that high ABEC numbers are not as important as a good manufacturer. I wonder why he left the belt in the system at all, versus gears that mesh directly.
tks
LlokC

 

[boostjuice]

... chain, instead of belt... maybe. I have read that high ABEC numbers are not as important as a good manufacturer. I wonder why he left the belt in the system at all, versus gears that mesh directly.

Because metal gears that mesh directly need to be bathed within an oil containing enclosure to last any reasonable length of time. A bit of grease chucked on is not good enough for metal gears in a dirty environment. An oil filled enclosure adds extra materials, complexity, machining costs, weight and bulk. Replacing pulleys with straight cut gears on the first stage of reduction would also make a stupidly loud amount of noise. They are noisy enough at lower speeds let alone at 7000RPM+. Helical gears would not be viable due to their extreme cost. Unlubed Nylon/Acetyl gears might be an option despite their lower strength for a given tooth form compared to steel, however nothing appropriate has been found available off the shelf like 5M HTD pulleys/belts.
Matt S has managed to keep his single stage reduction units at <$300 which is a great achievement. All other gearing approaches would undoubtedly be more expensive and rejected by the market as being too expensive.

1st stage belt/pulley transmission is quiet, long wearing, efficient at high RPM and quite inexpensive. I can't think of anything more ideal for early stage RC reduction.

 

[katou]

Ah Miles, ever the pragmatist. You are right of course, but how would that go, lets see...

Change reduction ratios and test no load amps
Change types of chain and test no load amps
Change width of belt and test no load amps

I foresee some problems with cost of development. There must be a way to get a feel for the losses and their magnitude, before building.

Usually, the way that takes the longest, is the best. I wish there was another way. How do people normally do this optimisation? Do they just build 5 different versions and test them all?

Katou

 

[Miles]

There must be a way to get a feel for the losses and their magnitude, before building.

For chains: Avoid small sprocket sizes (below 13t).

For belts: Avoid small pulley sizes; use the minimum width of belt for the torque requirements at the recommended tension; align the shafts accurately .

 

[mud2005]

as far as bearings go, I know for skateboarding ceramic bearings are used. steel ball bearings expand when hot and cause greater friction and even more heat while ceramic do not expand.
they aren't cheap though, check out VXB

 

[Jeremy Harris]

I've been experimenting with reducing transmission losses and can give some loss figures based on the tests I've done. These were aimed at finding the most efficient transmission for a low power application, my outrunner powered, solar charged, boat, where even a watt or two of wasted power was worth getting rid of.

5mm pitch, 15mm wide HTD belt drives with around 3:1 reduction ratio and decent pulley sizes absorbed about 7 watts

3mm pitch, 15mm wide HTD belt drives were better, only losing about 5 watts. I think the main cause of the power loss is belt flex, the 3mm pitch belts bend more easily than the 5mm pitch ones.

Bearing power losses were negligible,even for cheap bearings. I wasn't able to detect a power difference by swapping out cheap Chinese bearings for good Japanese ones, although the noise level improved. Bearing seal losses were significant though. Swapping out a 10mm double sealed bearing for a double shielded one saved around 2 to 3 watts.

Lip seals are even worse than sealed bearings, a single lip-type oil seal can easily soak up 3 to 4 watts.

Small pitch chain (the only type I tested was 6mm pitch) was the most efficient drive. Even running very small sprockets, the chain only wasted about 3 watts.

To put all these losses in perspective, my project runs with an output mechanical power of around 50 to 70 watts, so losing 10% or more of this in the transmission was worth chasing down. In the end I've gone for direct drive, as the reduction in motor efficiency from this configuration is less than the transmission loss from using a reduction drive. This wouldn't be the case for a higher power application though, as the transmission losses from belts and chains don't seem to change that much with speed or power; the greatest proportion of the loss is at low speed/power and the increase in power loss with speed is quite modest.

You can get a reasonable estimate of transmission loss by running the motor at different speeds without the transmission connected, noting the power used at each rpm. Next, fit the transmission and repeat the test and measurement. The difference between the two sets of readings will give an estimate of the transmission loss, although losses will increase once loads are applied. Nevertheless, this is a reasonable way of determining which transmission is likely to be most efficient in practice.

Jeremy

 

[liveforphysics]

Excellent info Jeremy! Thank you for taking the time to share your results with us. I've also never found bearings to make a bit of difference in anything outside of noise and durability.

When driving gears that mesh directly, they have a fixed loss from windage/oil friction, but the gear/gear friction varies directly with power. At higher power levels, the flex/windage/oil losses can roughly be neglected, and you can see loss measured as a percentage of the power, meaning loss is something like fixed% (torque * RPM). An unfortunately very noisy profile of straight cut stub-nose gear is always the highest efficiency. Chains are often next in line, depending on a few factors, then helical gears, then cog-belts, then bevel gears, then V-groove pulleys, then worm drives.

To minimize loss, make the gears/sprockets/pulleys as big and as narrow as you can fit in your application. Belts have the big advantage of staying pretty quiet at high speeds.

 

[Miles]

Yes, thanks Jeremy.

Here's the data from a test I did on my bike, last year.

No load data

For WOT (43V)

Motor only:
0.72 A

Drive system total:
Low belt tension, direct drive.......0.98A
Low belt tension, top gear...........1.01A
Nominal belt tension, direct drive..1.11A
Nominal belt tension, top gear......1.13A

Kv 59rpm/volt
Gates 8M PowerGrip belt 16t to 100t pulleys
SRAM Dual-Drive

 

[katou]

This is exactly what I was hoping for! Thank you all so much.

I wonder if the high efficiency of the chain drives hold when very small sprockets are used?

Data logging rocks. I still have to cast a couple of concrete countertops before I get busy with the ebike, but man, I'm getting seriously excited.

What is a lip seal? The only sealed bearings I'm familiar with have a snap-in plastic shield that is visible from the side. Like this: