Ok the only crashed I get now are when a prop falls off the quad will in flight. It has cost me two cameras and a good amount of embarrassment. It only happens once every 15 flights but I hate it. So has anyone ever super glued the collet prop ad pater to there motor? It seems even if the motor goes bad at some point a new motor comes with a new collet adapter.
I love your geo-fencing technique... "who woulda thunk it"... using *real fences*! ;) Makes for some interesting control if you don't have FPV and descend on the wrong side of a fence! ;)
On the collet slipping, it was quite obvious in the audio track is was coming... which leads me to an interesting thought... I used to do research in predicting epileptic siezures using EEG data... I suspect I could whip up some code to predict losing a prop from analysing the audio stream... it's not likely to be a deployable solution, but it would be a fun challenge (yes, I'm that kind of sick person! ;) ).
Why not use the APM itself to detect the vibration?
There has to be some transmission of the vibes into the airframe and thus into the accelerometers. Yes, if the APM is sufficiently damped/cushioned it should be deep in the noise but in some form it is there.
I work on MRI systems and understand about weak signal stuff (do ham radio too, EME work is *very* weak signal..oh yeah.. radio astronomy too..).
While putting together a cross section graphic and searching for these jewels, it occured to me that I have found no data or even tests on how much it takes to pull off one of these prop adapters. The test jig could be similar to the tensile strength test jig... (from a GA Tech online presentation HERE )
Of course to do it right, several motors, adapters, props and various combos would have to be sacrificed in the name of science. The entire system could be tested to failure, then the piece parts. Integrating a batch with torque values of prop nut/spinner data would tell the story of how much is enough torque to lock the collet.
Much engineering data...looks like someone's Mas.Sci degree research to me...
Vibration in the FMU is a mixed signal... i.e., the accelerometer is measuring all vibration sources at once. That would require a blind source separation solution. It could be used to augment or validate analysis derived from an data channel (or channels). I simply proposed audio because I thought it would be fun (and not so easy)... it's not something I'm likely to proceed with (no time, too many other paid research projects at the moment)... more of a gedanken experiment.
THIS fellow used a program on a cell phone to measure the vibrations of his efforts to balance props and propellor adapters.
He went throught various quality adapters, clipped off the motor shafts, and spent more time testing than flying it seems.
He was just spinning up the motor and adapter in these tests. One problem is knowing if the rpm was the same between tests. I have noticed resonance effects at different rpms. Again, 'the sound' is where it tells on the system.
No expectations Tim. There are so many places to spend time in this hobby when we should just be flying and, of course, trying not to crash.
I agree that you could probably detect a prop problem by sound alone, but since prop failure is pretty sudden, you wouldn't likely detect any anomalies before flight.
I wouldn't expect a collet to suddenly let go (i.e., to transition from secure to flying off). I'm not game to set up a test to validate this claim, but I believe you'd have several seconds (perhaps 10s of seconds) in which you'd hear the collet either slipping against the motor shaft (friction effects would generate some lovely sounds) or youd hear the propellors flapping irregularly as the collet was not travelling at a constant angular speed for a given constant motor speed.
Of course, this is intuition mixed with speculation, without validation, so take it as unvalidated hypothesis.
I have never had a prop come off until today when it happened twice on two different props. On one, the nut came off leaving the collet on the shaft. Recovered the prop but not the nut. The other the entire assembly came off teh shaft and was recovered. I suspect the cold caused differential expansion and contraction of the shaft/collet/collar (or whatever it is called) around the collet and the nut. Luckily no real damage.
I have never cared for the aluminum clamp type of prop nut. When you tighten them really well, you may be stripping the aluminum threads. Then add to the problem by using the prop on a motor turning in a direction which will cause the nut to turn in the loosen direction. THINK OF THIS: a table saw uses a reverse thread nut. And the nut is NOT aluminum ! Every turn of the saw blade will tighten the nut. They don't want that saw blade to come off. Multi copters should ONLY use prop screw shafts and real nuts! (IMHO)
I think has been my whole point from the begining. The design is simply flawed, because there are many ways for it to fail.
Yes, double checking each prop before you fly will help. Yes making sure you do not over tighten the collets will help. Yes listening to the quad as it flies might give a glue to something being wrong, but all of this could be avoided by designing a system that has the collets on the motors in a fail safe manner.
I now go through a long check list before I leave the ground yet things can still go wrong. Clearly the level of response posting this comment shows that there are a lot of people who have had trouble with the current system.
Which simply suggests that if there is a need for a better solution, go ahead and design one (and presumably make some money from it!) ;)
The collet-type adapter is not exactly flawed but probably not the best choice for a multirotor. All too often we put nickle-and-dime parts on our very expensive aircraft and want to blame the people who make the parts. Again, cheap ain't better, or best in this case.
Is there a better device? Absolutely.
As previously mentioned, multirotors are totally unforgiving at tossing a prop, by any means. The numbers alone will increase the risk of a prop adapter failure. Add to that the variance in quality between the motor shafts and adapters and the results are what we are experiencing.
The simple way to make sure our props are secure is to use another adapter. There are several makes now available that attach directly to the motor housing (rotational housing) IF you can find one for your motor.
Of course this type adapter, to be totally immune to rotational forces on the nut, would have to be in two types: CW and CCW threads.
Invariably someone will mix the two up, with predictable results OR they will mix up their ESC/motor wiring and mix up motor direction with directional adapter types.
I would hate to see any 'regulation' on this subject. It could be that clubs set (or have already) local rules for prop adapters to cover insurance risks. It could be that large organizations (AMA, IRCHA, etc) petition the manufacturers for a better solution. I certainly wouldn't want the Federal Government to get involved. None of us wants anyone to be injured, much less hang an expensive DSLR camera under an aircraft that could toss a prop at any time.
+1 with Scott on the probable cause. The aluminum will change size faster than the steel of the motor shaft. If the fit was 'good enough' before, it certainly will be less that that after getting cold.
Could the collet-type be made cheap enough if out of a steel alloy? Would the slight increase in weight be a 'no sale' for the weight-concious hobbyist?
Stay tuned dear viewers. These and more questions will be asked in the next episode of "As the Propellor Turns"!
think My Anders has the key here, it's not the nut that that is often the problem, but the "squeeze" adapter on the shaft. Red loctite is good for "gluing" an adpater to a shaft, and it easy to remove once the squeeze is off.