Currently working on a maslow variant for the open source market. Going to work on keeping it as backward compatible as possible and will be building a new board for stepper control, support for a lathe axis and VFD. Anyone interested in contributing let me know. Most of my machine design experience is in the 3D printing space and came across the maslow as a possible tool for prototyping.
4 Likes
I thought stepper motors would not work because they become loose without power?
Steppers have higher resistance to forced rotation than comparable dc motors. The existing motors go through a gear box that increases the amount of torque delivered.
It is called detent torque and is an interesting phenomenon.
https://www.motioncontroltips.com/faq-whats-the-difference-between-detent-torque-and-holding-torque/
you would not want to use them for the main motors, but for the Z axis (where
you are going through a lead screw), unless you have a very high pitch on the
lead screw it’s not going to drift (and you end up re-zeroing the Z frequently
for bit changes anyway)
David Lang
You’re right, when power is removed the motor shaft would turn freely.
It might be possible to use a worm gear drive after the stepper, with an encoder between the two to keep accurate track of position, but I don’t think the stepper-encoder-wormgear lashup is a common one. Putting the encoder after the gearbox would greatly reduce the resolution.
It would, but it’s now pretty easy to get encoders with 16k positions/rev, which
is pleanty good enough for us (we currently get ~8k positions/rev with the
encoder before the gearbox)
David Lang
I am curious as to why you think they would turn freely, will use CNC steppers with 200 positions and a 3:1 gearbox ~66 positions to compare to the resolution achieved by the original motors. At ~8000 positions with a 291:1 reduction that’s only ~30 positions at the shaft you could send information on sources for encoders that would be great as I have a good source for strong DC motors for the lathe axis but need to find encoders for feedback.
From what I could find, Detent Torque is around 1% of a motor’s static energized torque. The Maslow setup puts a small sprocket on the output shaft with a load somewhere between 10 and 15 pounds of tension from the chain. Without a gearbox to reduce that force, it seems like it would take a pretty hefty motor to have enough Detent Torque to hold reliably against the tension.
If you’ve done the experiment, what kind of load could be supported?
I suppose that with an encoder and controller that stayed powered up after the motor was shut down, the change in position could be tracked as the sled dropped and the chains sagged after motor off.
cut-n-paste from another email
I was initially thinking of the AMT10, but the AMT 21 or AMT 23 could be
interesting
https://www.digikey.com/product-detail/en/cui-inc/AMT102-V/102-1307-ND/827015
~$23 each AMT10 series, 2048ppr with index pulse (8K resolution)
This matches the current maslow resolution (when put on the output shaft, where
it would detect movement due to backlash). The index pulse would need to be
wired/programmed, but would eliminate the “find 12’oclock” problem.
https://www.digikey.com/product-detail/en/cui-inc/AMT112S-V/102-3316-ND/4835227
~$37 each AMT11 series 4096ppr, but no index pulse, probably not a good fit for
the maslow.
ttps://www.digikey.com/product-detail/en/cui-inc/AMT232B-V/102-4485-ND/8031587
~$50 each, AMT23 series with either 4k or 16k absolute resolution.
This uses a 3-wire seriel interface and would eliminate the need for the
real-time pulse tracking, and at our speeds, would potentially make it so that
we could eliminate the arduino and do everything from a Pi. The 0 position ie
programmable so that you could mechanically index the sprocket after it’s
mounted and then set that position to 0.
https://www.digikey.com/product-detail/en/cui-inc/AMT212A-V/102-4525-ND/8537384
~$56 each, AMT 21 3.3v RS-485 serial interface, 14 bit (16k position) absolute
resolution, but also includes multi-turn counter, so it could track the entire
position, not just one turn of it
lookup back driving gearboxes. most 3:1 gearboxes are able to be backdiven,
which means that they will move under tension.
there are very few gear arrangement that cannot be backdriven, worm gears,
cyclodial gears, and strain gears are the ones I know of
some other gearboxes can be harder to backdrive at high gear ratios, but 3:1 is
not high.
Early on there was a fork of the maslow code to support steppers by someone with
the same thinking you have. He found that the position did move when power was
off and he had to re-build his machine to use a worm gear instead of his initial
planetary gearbox.
David Lang
Thank you for the input, will be using steppers with enough torque to drive plus a home position for test. I have the standard motors for the first build and seeing as availability of those motors will be a problem in future default to thinking of steppers due to work on 3d printers and CNC machines.
1 Like
when my 3d printer is turned off the stepper motors that drive the xy plate move pretty easily. maybe there is a way to always keep the stepper motors energized. Not really sure how that works, but good luck.
1 Like
I was very excited by those - it would be tremendous to have absolute position tracking like that over the entire length of chain! They lose the turns counter value when power is lost though, and it looks like the turns counter cannot be set on power-up, only zeroed. Unless some form of backup power is supplied to the encoder (how much current?), one would have to do the chain calibration at each power-up, or handle the issue in the firmware.
I haven’t found much support info about these beyond the commands for zeroing and reading them - have you found the RS485 connection pinouts?
We are looking at different gear/motor options, there are actually a lot out
there. The stock motors are (like the rest of the Maslow), just barely good
enough, with motor drivers that are just barely good enough.
trying to stick to the stock drivers will make things hard,the new shields are
rated at 3A per motor (with overload protection) instead of 2A per motor
(without protection), so that significantly increases the available motors.
If you put an encoder on the output shaft instead of requiring one built-in to
the motor, that significantly widens your motor options as well.
David Lang
There is a link to the full datasheet on the website.
There is a proprietary program to configure them, I don’t know if that can set
the zero or not
OK, I found the data sheet.
This command resets the device. Note that for multi-turn encoders this command will result in a zero’ing of the turns counter as the encoder always starts with the turns counter at zero.
I also found the pigtail cable ($55…) but not the USB-to-encoder cable mentioned in the proprietary program manual. Guess I’ll have to contact them to see if there’s more info.
We may want to have the discussion about motors here I believe this is where the quotes originated. I’d like to also open up the discussion on whether a hand router is cost effective for the ideal needs of the Maslow or if a CNC spindle with an open source VFD/ motor controller to run it.
https ://forums.maslowcnc.com/t/for-those-that-were-not-part-of-the-kickstarter-where-to-obtain-the-main-motor-worm-gear-encoder/120
I looked at 500w brushless spindles but they are just as expensive as a palm router and not quite as powerfull. Going up to 800W increases the cost to around $200 which is the cost of a pretty powerfull router.
didnt’ seem to make economic sense to me especially given the zero warranty with imported spindle vs 1 year warranty with domestic products. To be fair the Chinese spindles do have a warranty but when shipping costs more than the product is worth it becomes a useless warranty.
Will sit down and compare routers to the 110v spindle I have on my bench top router tonight.