Saturday, May 27, 2006
Integrating the Frankenmotors into Godzilla...
I mounted one of the new motors onto the Godzilla z-axis (horizontal). I had to widen the mounting hole, which I did with a wood rasp. Instead of bolting it down I used some of the ever-useful C-clamps to secure it to the bulkhead.
Speed tests
At 12v friction within the z-axis structure of Godzilla reduced a theoretical 40 mm/sec down to 30.5 mm/sec.
At 5.4 volts we timed the movement of the y-axis tower moving at 11.4 mm/sec.
Extrapolating from those two measurement points it would appear that it will require somewhere in the range of 1.5 to 1.9 volts input to achieve 4 mm/sec. I'm guessing the it will be a bit more than that, maybe 2-2.25 volts. I know that 12v friction accounts for 23% of the motor load. At 5.4v it acounts for 36%. One can reasonably assume that for 4 mm/sec the friction proportion of the motor load will be even higher.
Before anybody jumps to any false conclusions I am going to mimic voltages using pulse width modulation (PWM).
Vibration
Being able to run the z-axis at 30 mm/sec gave me some nice insights into the strengths and weaknesses of the design. At this velocity having the threaded power rod solidly attached to the bearings was essential. I used duct tape. Lock washers would be needed for a more permanent arrangement. Nuts tended to try to walk off of the threaded rod if they weren't solidly fixed and would do so very quickly. They also tended to take the bearings with them which caused all sorts of drama.
At 11 mm/sec the operation of the z-axis was much more manageable. Interestingly, the y-axis tower vibrated a bit but only to the touch, not visibly. I suspect that I can reduce that by applying a thin layer of felt to the support feet of the tower as I had originally planned.
Observations
At first blush it appears that Godzilla will be as designed a good test platform for extending the Mk II extruder performance up to 10 and perhaps as much as 15 mm/sec. My feeling is that the design will require a rebuild to achieve 20 mm/sec. Most of the problem of going much beyond about 15 mm/sec I think will happen in the x-axis bridge between the two y-axis towers. The Frankenmotors, because of their high torque, exhibit a considerable jolt on startup. For the z-axis motors this isn't an issue since they are solidly secured to the work table. Given the way I've designed the connection between the z-axis rails and the y-axis tower and the fact that the y-axis motors are not used while the extruder is in operation I suspect that a bit more felting will sort out the jolt of motor startup.
The x-axis, however, will be a problem. Startup jolt will tend to try to rotate the assembly in the yz plane. The y-axis tower's feet will tend to try to absorb this. At higher speeds, however, it could be a problem.
We discussed avoiding startup and stopping nonlinearities by not trying to extrude during these operations. There are going to be some very interesting control and firmware challenges in achieving these sort of translation speeds.
I am really looking forward to slapping a shaft encoder on one of these z-axis assemblies and building a control board for it. Getting two of the z-axis assemblies to work in concert should be even more exciting. It should be fun! :-D
Speed tests
At 12v friction within the z-axis structure of Godzilla reduced a theoretical 40 mm/sec down to 30.5 mm/sec.
At 5.4 volts we timed the movement of the y-axis tower moving at 11.4 mm/sec.
Extrapolating from those two measurement points it would appear that it will require somewhere in the range of 1.5 to 1.9 volts input to achieve 4 mm/sec. I'm guessing the it will be a bit more than that, maybe 2-2.25 volts. I know that 12v friction accounts for 23% of the motor load. At 5.4v it acounts for 36%. One can reasonably assume that for 4 mm/sec the friction proportion of the motor load will be even higher.
Before anybody jumps to any false conclusions I am going to mimic voltages using pulse width modulation (PWM).
Vibration
Being able to run the z-axis at 30 mm/sec gave me some nice insights into the strengths and weaknesses of the design. At this velocity having the threaded power rod solidly attached to the bearings was essential. I used duct tape. Lock washers would be needed for a more permanent arrangement. Nuts tended to try to walk off of the threaded rod if they weren't solidly fixed and would do so very quickly. They also tended to take the bearings with them which caused all sorts of drama.
At 11 mm/sec the operation of the z-axis was much more manageable. Interestingly, the y-axis tower vibrated a bit but only to the touch, not visibly. I suspect that I can reduce that by applying a thin layer of felt to the support feet of the tower as I had originally planned.
Observations
At first blush it appears that Godzilla will be as designed a good test platform for extending the Mk II extruder performance up to 10 and perhaps as much as 15 mm/sec. My feeling is that the design will require a rebuild to achieve 20 mm/sec. Most of the problem of going much beyond about 15 mm/sec I think will happen in the x-axis bridge between the two y-axis towers. The Frankenmotors, because of their high torque, exhibit a considerable jolt on startup. For the z-axis motors this isn't an issue since they are solidly secured to the work table. Given the way I've designed the connection between the z-axis rails and the y-axis tower and the fact that the y-axis motors are not used while the extruder is in operation I suspect that a bit more felting will sort out the jolt of motor startup.
The x-axis, however, will be a problem. Startup jolt will tend to try to rotate the assembly in the yz plane. The y-axis tower's feet will tend to try to absorb this. At higher speeds, however, it could be a problem.
We discussed avoiding startup and stopping nonlinearities by not trying to extrude during these operations. There are going to be some very interesting control and firmware challenges in achieving these sort of translation speeds.
I am really looking forward to slapping a shaft encoder on one of these z-axis assemblies and building a control board for it. Getting two of the z-axis assemblies to work in concert should be even more exciting. It should be fun! :-D
Comments:
<< Home
Those numbers look good. BTW, I use nuts with those nifty nylon insert/lock-washers built into them on threaded rod and bearing interfaces. Simple to adjust, and stand up well to vibration. The one that travels the long way along the threaded rod to met the bearing is usually only good once. (Or, if you can break in the new nylon ring, then manage to back it into position...) If it has to be removed a new one is then needed.
Where do you get those? I've seen them in devices already built but I've never seen them for sale. :-(
I've bought boxes of them here and there in hardware stores, but I had to track some big 12mm ones for the chassis on my truck through a local industrial fastener supply store. Most larger towns with an industrial park seems to have one of these little stores, and they can order in a mind boggling array of stuff!
If you're using PWM to adjust the motor's speed anyway, then it is easy to apply a soft-start by ramping the voltage up on startup. This would avoid the problem with the motor jolting the assembly around.
You might try these guys for weird fasteners - they're the UK's main supplier and they have online ordering.
Post a Comment
<< Home
