My resin cast gear did last a few minutes, but then it failed.
.
I am getting a new gear, thanks @bar. I am going to replace the old grease that probably contains small peaces of metal, what type of grease is recommended?
PS. I wonder if a resin cast mixed with metal powder might be stronger?
They are less than a dollar each from the manufacturer. I got 200 extra coming in November
Does that mean youāre planning on selling these separately?
If I take my installed motor gear box cover off will things fall out? Iād like to check the grease but donāt want to lose my calibration or measurements. Iām hoping I can mark the tooth thatās up with my chain, take the cover off, grease the gears and reverse to put it back together. Will this work? Humā¦ Even on the flat of the shaft it will be hard to get the sprocket in the exact same positionā¦
Just add carbon fiber! If you do btw be really careful CF dust I would use as an additive is deadly to breath. It might help but @Metalmaslow is going to be able to hook ya up so wait and be safe.
Thank you
Take it open carefully, gently and you should be ok.
There is a shim washer that tends to fall out but other than that pretty easy to take apart. We will carry extra gears for anyone that needs them
you donāt need to worry about getting the sprocket to the same position, you
just rotate the sprocket to get one tooth at 12 oāclock
is the gear thatās failing the one that meshes with the worm drive?
My guess is that due to selection bias, that number is optimistic, that is, in reality more than 50% of kits sold havenāt been built. The people who have started building are more likely to be following and responding to the messages on the FB group.
How are you certain that we need to measure the maximally stretched length at full power? Put aside the need to test rigidity for a moment.
I understand that the pull between motors test stretches the chain, taking up all slop, minimizing distance between motors.
But could it be the case that during normal cutting operation the load is light enough that not all that slop is used? If so, the full power test gives inaccurate results.
Has anyone put a spring scale in the chain during a full power pull test to see how much chain tension is developed compared to the max achieved during cutting? Are these gearboxes even rated for a full power stall?
For calibration purposes, we really donāt need to exceed the maximum in use tension by very much.
In lieu of full power, take worst case chain tension 34lb, add a fudge factor, call it 40lb.
Only 1 person needs to do this 1 time. Ramp power till the spring scale reads 40lb. Note the PWM duty cycle at that point. Now add that duty cycle to the code as a global maximum. This should give us all the power we need and limit stress on the machine. If we stop seeing gear failures, great. If not, that is when we start looking for alternate gearboxen.
Even if only 600 kits have been assembled (20% of orders) , 6 failed motors is 1% failure rate. With such different construction methods itās hard to be positive where the issue lies.
I have never done a chain pull for calibration and my machine is accurate enough for my current needs. I have only really worked in the center (sweet spot) and havenāt done any intricate fits. Mechanically that just makes me nervous to stretch the chains like that. Just as accurate as possible manual measurements has been enough for me.
Then take into account that not all users of built machines are registered on the forum. Of those that are registered, not all would report a failure here. Being a DIY machine, users are more likely to think failures are their own fault and not hardware fault. Of those that would report, not all will be able to link the symptoms to gearbox. That requires intuition to suspect the source, and disassembly to confirm. Reports of unexplained inaccuracy could be gearbox failures undiagnosed.
Given the consistent failure mode (failed gears have the same unique wear pattern), and no one off alternate explanations for it, I suspect a systemic cause.
Our sprockets have a pitch radius of 1.02743cm. If 50kg cm is the rated torque, then chain tension should never exceed 48.6kg or 107lbs. But our application is not applying pure torque. We are hanging our load on a cantilevered shaft, and that requires derating. Constant load in a single direction requires further derating. I donāt know how much.
It is not uncommon in the gearmotor market for motor stall torque multiplied by gear ratio to exceed the gearbox breaking torque. Iām not sure if that is the case here. I wont do this to my motors, but to test it you could hang 107lbs on your maslow chain. Apply power to lift it. If it lifts at all or breaks, then we know the pull test exceeds the motor rating and is unsafe.
This makes perfect sense in principle - I am also planning to figure out how & at what level to limit torque on my Maslow. In that vein, I plan to mount a load cell to dynamically measure to torque on the output shafts.
However, I suspect that the PWM duty cycle would not indicate load - only speed. I think we would need a current sense to actually measure the motor load, unless we can count on a current controlled power supply. Iāve not looked at it yet, but I would assume that the h-bridge motor controller is providing a voltage controlled supply, with current limitting, but not current control.
Can someone who knows more about the supplies we are using provide some off-the-cuff insight?
Thereās also the point of trying for max tension to minimize the sag between the
motors (which throws off the measurement)
The chain is not elastic, and the max force between the motors is still FAR
below the chain yeild limit.
I do think itās a little odd that the left motor seems to be what fails. It
seems like some fail with almost no use, while other people are doing a lot of
cutting without issues.
the fact that someone reported that some motors have no lubrication in them,
while others have more seems like it is a significant factor.
David Lang
No. It is the input(driven) side of the final compound gear prior to the output gear. If you look at its shaft, the one without a gear in the pics, it is substantially smaller than the output shaft. I wonder if that shaft is flexing causing causing shallow tooth engagement?
Iām going to say I know nothing of what is going on in the code however, here is my brain just thinking this through, is the code for the TLE driven differently? If not previously there was a load share between 2 motors and 1 driver. What motor was shared with Z? Left or Right please?
Thank you
The power supply powers the entire system, so it does not do any current
control.
There is no current sensor on the controller boards, the only current limiting
(and only on the new smoke herder boards) is the self-protection in the
controller chips that sense they are overheating and cut the current down.
The PWM doesnāt indicate speed, it indicates the percentage of power (on
average) being applied to the motor, itās a different amount of total power
depending on the voltage being applied.
David Lang
While cutting, the PWM duty cycle would not correlate to load or speed since the load is variable based on position. However, in the unique case of the pull test with spring scale, the motor speed is fixed at 0, and the load is fixed at 40lb. The one variable we need to find is the associated motor control input, duty cycle. As 40lb was specifically chosen to be higher than calculated static tension at any point on the board, we should never need to exceed that input duty cycle in normal operation.