The top end of the hobby sized servo market seems to peak out at a torque of around 180 Kg-cm, or 13 ft-lb. The Servo50 will be designed to deliver around 50 ft-lb.
It would be really cool to see someone's complex servo-based project scaled up; I know that if it were me I'd be frustrated by what's available today.
So how big will this thing be? I'm only hoping for the cover illustration at the moment, which is 2X on all axes. There are a couple of sub-projects to be conquered before any reality sets in:
- Motor control: looking to make a 3-phase, as in brushless, motor controller with motor axis as well as output drive position feedback. First type of control will be PWM. The board will be as small as possible - which is???
- Gearbox: I forsee either a simple planetary or eipcyclic gearbox, depending on the level of torque that can be squeezed out of the motor.
So it might look like the illustration, and what it *really* looks like is yet to be seen
Files
3PhaseDriver02_20180419a.zip
x-zip-compressed -
163.62 kB -
04/19/2018 at 20:21
Here's a timeline of what needs to happen to get the Servo50 into the hands of users. Note that the dates are totally bogus, as things take as long as they take - I'm not going to make any promises at this point. :)
Started using zones to pour large copper for the high current areas - after a while it got a bit easier. Not sure KiCad unconnected net tool is too happy about that sort of behavior. :)
The large(ish) inductor is for the builtin buck converter of the phase controller, supplying 3.3V. This will supply power to the CPU in the next revision of the board.
Out to fab is next, as well as ordering the parts.
The latest version of the board design files are in the 'files' section.
I've been trying to find the inspiration for this style of gearbox - I think it was somewhere here on Hackaday, but I can't find it any more. If you check out the Wikipedia article on Epicyclic Gearing, there's a small note on a version of what I'm doing that was done during WWII for the drive box for a portable radar unit. It might not be exactly what I've been doing, which is:
Start with a normal planetary gear set, with the sun, planets and ring gears.
Extend the planet gears, and meld in a gear with more teeth. They call this a stepped planet.
Make a ring gear that mates with the larger gears.
This results, in my current design, in around a 150:1 gear reduction in something a bit larger than my fist. It could be far smaller, but my 3D printer doesn't do well at anything smaller and the amount of output torque scales down as well.
Getting all the gear ratios to match up is a bit of a challenge; not every desired gear ratio is possible. :)
Okay, so the email said that I have to put a flow chart up, but doesn't say where to include it, so it's a JPG stuck in the files section. It doesn't really show the scale of the software challenge; I'll be describing this in a bit.
I've selected most of what I want to use for the first version of the motor controller. Once the board's out for fab I'll upload the KiCad files. In the meantime here's the working schematic.
It's a JPG, which is not optimal - so if anyone knows how to get a PDF up on one of these log files, please let me know. :)
So I've been noodling this sort of idea for a long time now, and I figure that designing an extra large motor controller/gearbox combo for a large hexapod isn't going so fast; maybe a side trip by way of this contest will help me in my eventual developments.
The Servo50 is envisioned as:
50 ft-lb target torque
120 RPM maximum slew rate
12-24V supply
Brushless motor
Motor controller with:
Documentation (KiCAD and source code)
PWM control
final drive feedback with index
motor axis feedback for good commutation behavior
Planetary or epicyclic gearbox
I have a few good 'from scratch' gearboxes that I've designed and fabricated. The electronics are inspired by the ODrive project, however the board for the Servo50 will only control one motor.
That's it for now, time to start on that motor controller.