Certain frequencies liable to loosen screws?

[Infraviolet]

Are there certain vibration frequencies particularly likely to cause the loosening of screws?

I've been doing some PWM (for current limitng) driving of BLDC motors, and notice that the screws holding the motors in the mounts (M3, 8mm bolts going through a through hole in a printed PLA plastic mount in a tapped hole in the motor face) keep loosening themselves.

The motors do "sing" somewhat with the type of PWM being applied, so there are clearly plenty of mechanical vibrations present from the PWMing of relatively high current (10A peak, time averaged 1A) signals.

Is it likely I've got the bad luck to be PWMing at exactly the right frequency to loosen screws (I'm at 7.8KHz), or is screw loosening likely to be from much slower vibrations, such as those related to the time period of the motor's mechanical or electrical (7 poles so 7x mech freq)rotation which vary depending on the speed I'm turning it at.

If there are particular frequencies which cause loosening, is there a chart or equation linking the "frequency to avoid" to metric screw size?

Thanks

[Stray Electron]

    Absolutely!   I've had both cars and motorcycles that would vibrate the nuts off of the bolts when they were running if you didn't have them tight!  The worst offenders were the poorly made American cars from about 1965 to the early 1980s with the extremely loose fitting nuts and bolts. The nuts on some of those were loose enough fit that you could almost spin them off and when you ran the engine the nuts would vibrate and rotate and you could watch then work their way back and forth up and down the bolt.  My first Honda motorcycle would do the same thing, in particular on the bolts holding the carburetor to the intake manifold. I used epoxy those bolts on and then just take the entire manifold off if i needed to work on the carb.  On that MC the carburetor was hung out on the end of a rather long tube like intake manifold and an old man that was a long time MC rider taught me that you could add some weight to the carburetor or the intake and that would change the resonant frequency and would sometimes stop the bolts from vibrating out.  On both the cars and the MCs you could change the engine speed and watch how that caused the nuts and bolts to rotate faster or slower and sometimes even reverse direction.

   You might want to look up the Class fit of bolts and study  why poorly made, loose fitting nuts and bolts aren't used on aircraft or where reliability if important.

[thm_w]

Sure, but the solution would not be to change the frequency, it would be to use loctite or change the fastening method.

[Benta]

Sure, but the solution would not be to change the frequency, it would be to use loctite or change the fastening method.

Exactly. Your bolt torque is too low for due to the PLA mount.
Try using bolts with a flanged head (standard in automotive for the same reason).

[Infraviolet]

I've done some further tests with the motor, the screw don't easily start loosening, but once they've begun loosening even slightly vibrations take them out very fast. I deliberately loosened one by "that first bit", then put the motor in the worst vibrating state posiblr and watched the screw turning itself outward at a rate of about 0.3 rotations per second.

I've also found the worst loosening is when the motor encounters a "high speed stall", where the current limit set at a particular time gives insufficient current (in open loop mode) to turn the motor at a given high speed. So the rotor stands "still" (vibrating) while the magnetic fields from the stator coils race round at the speed which the rotot can;t keep up with when the current is low. This causes far more vibrations that just the 8KHz of PWM.

I've had a look at locking washers, is there actually a definite answer as to which type is best?

There's effectively 4 types, split ring, serrated faces, internal teeth and external teeth.

I've heard the split ring sprung ones often make loosening worse not better? Is this true, it is one of those things where anecdotes outnumber data. How about the other sorts, how do they compare?

Might a drip of superglue between the bolt head the the PLA also be a good idea? Its not a true threadlocking compound, but I'd rather avoid using one of those here as it would have to go in to the tapped holes (aluminium motor case, not steel) and would risk leaking towards to the motor windings. A tiny bit of adhesive between the screw head and the PLA though might make it just hard enough to start turning that vibrations can't make the initial loosening, while still being weak enough to be unscrewed when twisting the screw with a screwdriver.

[indeterminate]

Use Threadlock  & a flanged bolt
you can get a pre cote thredlock that is dry on the bolt before assembly " no drips"

https://www.kerbkonus.com/proddb/pdf/en.ds.60.precote30.pdf

[Doctorandus_P]

bolts vibrating loose is a common problem, but mostly when they are already loose. When bolts are properly torqued the probem is already a lot less. But for safety critical parts theare are other measures, such as:

* Nordlock washers.
* Nylock nuts.
* Locktite.
* Cotter pins.
* Safety / Locking wire (very common in aircraft industry).


But for a more direct answer to your question. I do not believe that the frequency on itself is a big factor. It may be a small factor, but there are a lot of other things to consider too. For example the exact motion of the the vibration. Take for example a vibratory feeding bowl. These are designed in such a way that the parts in the bowl always rotate in the same direction, They even move upwards the ramp against gravity.

Also, the thread itself is a factor. Metric thread generally has a finer pitch then imperial threads, and therefore they have less tendency to work themselves loose.

[jpanhalt]

On full scale airplanes there are various ways to secure fasteners.  Serrated washers and washer-head nuts and bolts can be used.  In critical areas, locking nuts and safety wires are used.   Depending on function, lock nuts can be nylon/fiber inserts or deformed thread.  The latter is required in high temperature areas.  Thread locking compounds, e.g., Loctite, can also be used as a substitute according to specifications/regulations.

I do not like serrated fasteners because they mar the surfaces to which they are attached.  They might be fine where periodic disassembly is not needed.  When such disassembly is needed, I prefer safety wiring or one of the other alternatives.

For any soft material, like wood or plastic, I would never use a serrated fastener.*  For your electric motor mounts, nylon insert, distorted thread, or thread locker is probably OK.  In cars, you see a lot of liquid thread locker and distorted thread used.  Be aware there are generally 2 types of liquid thread locker.  One type sets firm but can still be unfastened relatively easily.  The other sets hard and is more difficult to remove without heat.

For IC engine mounts against wood in my model airplanes, I use safety wire.  They may get loose, but they never come apart.

*A flanged, e.g, washer-headed,  but not serrated fastener increases the contact surface area and offers some protection.  I use them on wood too.

[Kleinstein]

Mechanical systems can have many vibrational / acoustic resonances. Quite often the resonances are quite sharp and can amplify vibrations / forces quite a bit. The vibration can this way also reach parts that one may not expect to be reached by the vibrations if they were static fores. The exact frequencies depend on the details and can be hard to predict. One may also hear a much stronger noise at the resonaces.

[thm_w]

Might a drip of superglue between the bolt head the the PLA also be a good idea? Its not a true threadlocking compound, but I'd rather avoid using one of those here as it would have to go in to the tapped holes (aluminium motor case, not steel) and would risk leaking towards to the motor windings. A tiny bit of adhesive between the screw head and the PLA though might make it just hard enough to start turning that vibrations can't make the initial loosening, while still being weak enough to be unscrewed when twisting the screw with a screwdriver.

You were told that loctite is the appropriate method and PLA is too soft to allow appropriate clamping force. A toothed washer or adhesive on PLA is going to be weak compared to loctite.

If there are tapped holes going inside the motor, there is already a risk that aluminum burrs will fall off and enter the windings. That is worse than loctite dripping in, and you can get non-drip loctite gel anyway.

Why don't these motors have appropriate mounting flanges (eg NEMA17)?
Or are you trying to mount something else like a encode to the motor itself?

[Ian.M]

Oversize the holes in the mounts to accommodate a metal sleeve, slightly shorter than the mount thickness.  The PLA will squish a bit but the bolts can still be torqued down properly against the sleeve.  It will probably also need washers or flange head bolts to spread the load on the PLA and make up for the loss of contact area due to the sleeve thickness.

[jpanhalt]

@Ian M
That doesn't help much.  The spool (attachment) you make will turn very easily as the plastic cold flows. 

[Ian.M]

The sleeve will be compressed between the bolt head and the surface the threaded hole is in, allowing the bolt to be torqued enough to resist loosening.  The plastic doesn't have to grip the sleeve, even a loose slip fit will be OK.   Of course it wont transmit torque to the plastic effectively but using a single threaded fastener to resist torque that can undo it is never a good idea - you always need another off axis fastener or other feature to resist rotation.

[Infraviolet]

I made a very ugly discovery. This situation is much worse than I originally thought it. The motor gets warm enough to seriously soften the PLA, that strongly contributes to the PLA not keeping the screw tensioned. I'm going to buy a PETG roll, I hear it is almost as easy to print with as PLA, no need for chambers the way ABS can require to stop drafts cooling the print, and try again with 100% infill density in that area (its a pretty small volume where the screw is pressing), plus wide headed flanged screws, plus serrated washers. I've been able to get M2.5 and M3 sized screws to give, seemingly, good tension when holding printed plastic parts together in the past, so I think the reason I had trouble here was when the PLA went soft.

[T3sl4co1l]

If PWM is low enough that you're seeing significant torque pulsation/ripple, and excess heating, raising frequency so that winding inductance dominates a bit more, is probably a good idea.  After that, if you're simply running it too hard (too much V for F, too much I/torque, too much V/F/RPM in general), or there's not enough cooling, well, just don't do that.

Note that prints are generally quite poor conductivity so if you're boxing up that motor, consider adding ventilation holes/slots, more air space, a fan, etc.

Tim

[Doctorandus_P]

Most plastics do not only deform when hot, but they deform always when under tension. It is called "creep". And it's effects can be mitigated in several ways. One way is to use big washers under your screws which distribute the force over a larger area.

It would help if you post a picture of your setup.

[Infraviolet]

This is not quite my setup, but I found these images online of someone who's using similarly shaped mountings for a BLDC of similar form factor. Not at home today, so can't photograph mine right now.

My shaft type is a bit different, and I've got a wider mount as I'll be fitting a printed gearbox around that shaft as my project progresses, but the images at the link below show a mount which fits on the to "face" of a BLDC the same way. As with mine, the BLDC is mostly exposed to the air except on that face plate side.

https://www.printables.com/model/761682-bldc-motor-mount

Also, the guy who designed that printed it so that when on the print bed the Z axis was in the direction aligned to the motor's shaft, whereas my print, due to other features urthr away from the motor itself, was done in the "other" orientation, with the Z axis being up away from what that design has as a "plinth" (pretty much the orientation it sits in in that guy's first image). So his screw holes for the "face plate" are more correctly circular than mine, his screws are compressing layers along the printed Z axis, my screws compress along the printed X axis.