Bearing friction torque
Posted by cristian
|
Re: Bearing friction torque October 16, 2015 02:58PM |
Okay, I did my homework. I set up a system like this
and I tried to measure static friction torque. It turned out to be quite a difficult thing to measure "at home" on 608 ball bearings, because of all the possible alignment errors between the bearings and the line as well as the "noise" on the measure, given by the non homogeneous static friction in relation to the rotation angle of the bearings.
The measurements I made are enough to get an order of magnitude of bearings friction torque, but unfortunately they are not enough to extrapolate a full graph in relation to the load: after spending an afternoon playing with bearings, line, cans of water and a pipette, my girlfriend started to think that I went insane so I had to stop.
The experiment was simple: two weights M1 and M2 (water cans) connected with the line (0.5mm diameter) over two bearings (608zz, 22mm diameter), initially with M1 == M2 and then M1 was increased drop of water by drop of water until the system started to move. Reality is of course not as simple as that, since the line is partially elastic and not exactly at the center of the bearings, two bearings are involved, weights oscillate, etc. In brief, these are the measurements (in gram-force):
Given the error in the measurement and the intrinsic noise of the system, standard deviation will be considerable (the measurements are so few that standard deviation cannot be calculated for most loads). The relation between friction torque and load is probably not even linear in such a wide range. Excluding the first measurement where the friction was too low to be measured "accurately" and the load was not interesting for practical purposes, the worst ratio friction/load case seems given by the second measurement, with L =~ 1500 and the friction torque (per bearing) being T ~= (47 gram-force / 2) * (22 mm / 2) ~= 2.5 N * mm. If the bearings are used "naked", without changing their diameter (and the thickness of the line or belts is considered as negligible), the resulting friction is approximately 1.5% of the applied load. Looking at the measurements, it seems that friction increases with load (as it should be) but the ratio friction/load improves overall.
Given that I paid those bearings about 0.30€ each, this is a great result. However in a system with a lot of bearings (for example a corexy where you have at least 8 bearings) and long belts, this is probably not so negligible, because the load on the bearings depends on the tension of the belts which may be considerable.
What do you think?
Next step is measuring belt and line elongation as a function of load.
and I tried to measure static friction torque. It turned out to be quite a difficult thing to measure "at home" on 608 ball bearings, because of all the possible alignment errors between the bearings and the line as well as the "noise" on the measure, given by the non homogeneous static friction in relation to the rotation angle of the bearings.
The measurements I made are enough to get an order of magnitude of bearings friction torque, but unfortunately they are not enough to extrapolate a full graph in relation to the load: after spending an afternoon playing with bearings, line, cans of water and a pipette, my girlfriend started to think that I went insane so I had to stop.
The experiment was simple: two weights M1 and M2 (water cans) connected with the line (0.5mm diameter) over two bearings (608zz, 22mm diameter), initially with M1 == M2 and then M1 was increased drop of water by drop of water until the system started to move. Reality is of course not as simple as that, since the line is partially elastic and not exactly at the center of the bearings, two bearings are involved, weights oscillate, etc. In brief, these are the measurements (in gram-force):
Initial mass for M1 and M2 Final mass for M1 Difference 59 65 < 6 1062 1109 47 2063 2103 40 2063 2115 52 4064 4137 73 4064 4122 58 4064 4132 68The load on the bearings should be approximately given by L = (M1 + M2) / sqrt(2). Since M1 ~= M2, we have L ~= M2 * sqrt(2).
Given the error in the measurement and the intrinsic noise of the system, standard deviation will be considerable (the measurements are so few that standard deviation cannot be calculated for most loads). The relation between friction torque and load is probably not even linear in such a wide range. Excluding the first measurement where the friction was too low to be measured "accurately" and the load was not interesting for practical purposes, the worst ratio friction/load case seems given by the second measurement, with L =~ 1500 and the friction torque (per bearing) being T ~= (47 gram-force / 2) * (22 mm / 2) ~= 2.5 N * mm. If the bearings are used "naked", without changing their diameter (and the thickness of the line or belts is considered as negligible), the resulting friction is approximately 1.5% of the applied load. Looking at the measurements, it seems that friction increases with load (as it should be) but the ratio friction/load improves overall.
Given that I paid those bearings about 0.30€ each, this is a great result. However in a system with a lot of bearings (for example a corexy where you have at least 8 bearings) and long belts, this is probably not so negligible, because the load on the bearings depends on the tension of the belts which may be considerable.
What do you think?
Next step is measuring belt and line elongation as a function of load.
Sorry, only registered users may post in this forum.