Linear motors: Day Dreaming about future printers.
-
@bot do you know how much they cost (in which price region?)
There was once a DIY project here: https://reprap.org/forum/read.php?160,813817,page=1 but it died somehow.
-
I have no clue about the price. Most listings I find for "linear motor" are AC induction motors, so totally not what I'm looking for, or they are stepper motors with leadscrew shafts embedded in the rotor. Not at all what I'm after.
This find, of what seems to be an economical brand of linear STEPPER motor is quite tantalizing. I bet they are still too costly, but the smallest of them seems to be sized right in terms of power, weight, and form factor.
It looks like the forcer that the user took apart (in that RepRap Forum thread) is the same one I am eyeballing. He says it is usually $900. Certainly more than I'd like. But, I wonder how easy it would be to convince chinese manufacturers to drive down the cost -- it's certainly not too hard to make these, especially if they are "hobbyist" grade where the tolerances can be looser to hit a price point.
I can't justify it for any real reason TODAY, but it's something I have in mind for my next printer.
The Dual-Axis one is so intriguing. The drawbacks of it probably outweigh the benefits, though. Which is a shame. Imagine just slapping a hot end on a couple (or more) forcers and just letting them all work at once. They could be working on different parts of the build simultaneously, as long as the toolpath kept the buffer zones in check.
Though, I am not sure how rigid the air bearing is. It's possible that supplementary support from a linear rail or something would be required, which would kinda kill that dream.
The single-axis ones are enticing enough on their own. They can also have multiple forcers per rail/platen. Like Idex, but simplified (no belts, no pulleys).
Interestingly, in that RepRap forum thread, somebody mentions that the ~8.9 N force output of the forcer is not much compared to typical Nema 17 motors. Is that true? It seems the size of the stepper has to grow significantly to drive the same load, when compared to a rotary stepper. Is this a fundamental drawback of the linear motor?
*not actually a robot
-
@arhi There's a magnet under the table that is moved by a corexy mechanism. I use a 1" cube N52 neodymium magnet and I think the ball is 13 mm diameter. The bottom of the table is 1/2" thick plywood- I overdid that- so the magnet is controlling the ball over >1/2" distance, even at those speeds and accelerations.
I used 16 tooth drive pulleys and have the motors set for 1000 steps per rev. That limits the top safe speed to 1600 mm/sec because of the back emf generated by the motors at 3k rpm. If I use larger drive pulleys it can easily go much faster. I shot a video of it running at 2k mm/sec (3750 rpm) on an erase pattern. The Duet was complaining about overvoltage the whole time, so I won't be running it that fast again without putting in larger drive pulleys.
The sand is sodium bicarbonate sand-blasting medium. At that speed the ball throws the sand around and the sand becomes "thick" so the movement isn't in the straight lines it should be. Everything behaves nicer at the "normal" speed of 500 mm/sec which is about 5-20x faster than people normally run these things. My real-time videos look like other peoples' time lapse videos!
The motors are iHSV-42-40-07-24, 75W, NEMA-17 size integrated servos with 1000 line encoders. More here: https://drmrehorst.blogspot.com/2020/04/the-spice-must-flow-gets-servo-motors.html
The motors are rated for 0.185 Nm torque which translates to 26 oz-in, but servomotor torque ratings aren't holding torque like steppers. Stepper torque drops as speed increases. Servo torque remains almost constant up to the back emf limited speed. I wasn't sure they would have enough grunt to move the mechanism at all, let alone drive it with acceleration of 20k, so I was pleasantly surprised when they did. -
@bot A friend at the makerspace has a pick and place (?) machine that he got at an auction that has some large linear motors set up in an XY arrangement (maybe H-bot)- not sure if they are servos or steppers. He's had the thing sitting in storage for about 3 years and every once in a while he reminds me it's there waiting to be turned into a 3D printer. The drivers and power supplies for the motors are in a box that is bigger than most 3D printers. The machine probably weighs 200-300 kg and the working area is maybe 400 mm square, IRIC. I think it requires 3 phase power. We played with it when he first got it and were able to make the mechanism move a bit using the control panel on the machine, but that's as far as it ever went. here's a picture
-
Wow! Those things are beasts. It seems like linear steppers need to be larger than rotary steppers.
I found this interesting article which describes some benefits and drawbacks of the different types of linear motor. There is a tubular layout that seems interesting.
*not actually a robot
-
@JoergS5 There may be hope... I contacted the author of that topic in March and they said the linear rails had been sent off to be EDM'd.
@bot I too have been dreaming about using linear motors in a build. I've heard they can sometimes be salvaged from old electric typewriters from the 80's like the Daisy Wheel or Brother EM series. But I don't know if those are suitable for a printer.
-
What I also like very much is the idea to combine air-beared sliding with a solid construction like taking hardstone: https://www.jfa.de/html/en/produkte/p/53/guidances-with-air-beared-slides-ll-s/
-
@bot said in Linear motors: Day Dreaming about future printers.:
Most listings I find for "linear motor" are AC induction motors, so totally not what I'm looking for
That's probably a more practical solution. It should mean that the bar can be a simple steel strip (possibly laminated) with an optical encoder strip mounted on it, instead of needing precisely-milled and epoxy-filled teeth. I think this could be done as a DIY project. Linear encoders can be bought for around $50, see for example https://www.rls.si/en/rlb-linear-component-magnetic-encoder-system.
If it was built as a 2-phase motor, there is a possibility that our upcoming closed loop stepper motor driver could drive it, depending on how much power it needed.
Duet WiFi hardware designer and firmware engineer
Please do not ask me for Duet support via PM or email, use the forum
http://www.escher3d.com, https://miscsolutions.wordpress.com -
@dc42 said in Linear motors: Day Dreaming about future printers.:
@bot said in Linear motors: Day Dreaming about future printers.:
Most listings I find for "linear motor" are AC induction motors, so totally not what I'm looking for
That's probably a more practical solution. It should mean that the bar can be a simple steel strip (possibly laminated) with an optical encoder strip mounted on it, instead of needing precisely-milled and epoxy-filled teeth. I think this could be done as a DIY project. Linear encoders can be bought for around $50, see for example https://www.rls.si/en/rlb-linear-component-magnetic-encoder-system.
If it was built as a 2-phase motor, there is a possibility that our upcoming closed loop stepper motor driver could drive it, depending on how much power it needed.
Very interesting. I will broaden my casual search. I had no idea this would be a viable option, let alone a superior option to the stepper types.
Typically, though, I'm finding very large AC inductor motors, whereas I want something compact with similar force to that generated by a typical motor/pulley/belt arrangement.
Hmm, here we go. Something like this might be suitable in conjunction with the closed-loop duet expansion: https://www.nipponpulse.com/products/motors/motors-linear-servo
-
I remember when I was a teenager, I saw Eric Laithwaite on playing with linear induction motors and magnetic levitation. I got magnetic levitation working - which uses a similar induction principle - but I never got round to building a linear induction motor. I seem to recall that he reckoned transverse flux motors were better than the original longitudinal type, but they were wider.
Thinking about it some more, the stator would be a steel bar faced with fairly thin copper sheet. I think all those years ago, I reckoned that the active bit could be made from laminated steel or ferrite E- or C-cores, which are readily available. You would need to maintain a small air gap between the faces of the cores and the copper face of the stator.
-
I looked into linear steppers for 3D printing about 10 years ago.
I was quoted about $1000 per axis, with a 300mm stroke IIRC. That was at volume too.
Most of the companies making them are aiming for 1 micron repeatability, for machines were $1000 is a small cost.I know of at least 1 person who was working on a low cost linear servo, but he was trying to raise money for it as part of a robotics startup.
You can build something very similar to those NPM linear servos for very little money, the electronics are not super complicated... the hardest part is the control.
With a bit of imagination, you could probably even build a linear stepper in the same style. Thrust output might not be great, but if you can make it a stepper at least control is simple. -
If it was built as a 2-phase motor, there is a possibility that our upcoming closed loop stepper motor driver could drive it, depending on how much power it needed.
Is it possible to drive DIY 2-phase closed loop linear motor with (new) Duet hardware directly (or with additional hardware)? Btw I'm not so sure that you can get enough force from a linear motor without using strong magnets instead of steel bar and a copper sheet. Some example demo videos on Youtube make such a linear motor seem quite weak.
3D-printer part collector || https://grabcad.com/eetu-4/models
-
@Visionary said in Linear motors: Day Dreaming about future printers.:
Is it possible to drive DIY 2-phase closed loop linear motor with (new) Duet hardware directly (or with additional hardware)?
That would depend on the interface t the closed loop motor.
I've looked at the possibility of using a DIY linear induction motor to drive axes of 3D printers - I have some knowledge of them from experiments many years ago. For the Z axis, I doubt they are practical or that they would offer any significant advantages. For X and Y, I don't see any arrangement other than a bed-slinger that wouldn't require one of the motors to be mounted on a moving axes, which is undesirable because then the moving mass would be considerable.
Duet WiFi hardware designer and firmware engineer
Please do not ask me for Duet support via PM or email, use the forum
http://www.escher3d.com, https://miscsolutions.wordpress.com -
Peopoly uses one linear motor on each X and Y axis and two linear rail blocks on each Y-axis rail. On their printer it seems to work even though there is some cantilever pulling/pushing going on. Thomas Sanladerer shows some details on the printer on his review video.
STōN-WoLF printer also has kinematic configuration that could be a bit more suitable for linear motors.
I've had (Chinese) ball screws on my Z-axes and they are ok, but they still have slight Z-wobble and they do have noticeable amount of backlash. Using linear motors also on Z-axis would definitely be expensive, but I'm curious if there would be better in every other way (speed, noise, precision, no moving parts, no grease). Preventing bed from dropping hard when unpowered is an issue too.
Many of the examples of DIY linear motors on YT and on Internet use 3-phase setup and drive the motor with VFD or what I assume is a BLDC drive. If it would be possible to use 2-phase setup instead and use directly say 6HC stepper drivers then that would very simple if possible. Or could ODrive Pro-board be used through CAN-connection to Duet to drive a 3-phase linear motor? I do realise that this would be expensive, but e.g. one Clearpath servo imported to Europe costs >400€ and that doesn't even include belt drive components.
3D-printer part collector || https://grabcad.com/eetu-4/models
-
@Visionary said in Linear motors: Day Dreaming about future printers.:
Or could ODrive Pro-board be used through CAN-connection to Duet to drive a 3-phase linear motor?
I don't know about O-drive (they are closed source now), but I've used simpleFOC with open source driver boards to drive sensored BLDC motors via step/dir interface.
If it would be easier to DIY a 3-phase linear motor, that would be a cheap solution. -
@Visionary said in Linear motors: Day Dreaming about future printers.:
I've had (Chinese) ball screws on my Z-axes and they are ok, but they still have slight Z-wobble and they do have noticeable amount of backlash. Using linear motors also on Z-axis would definitely be expensive, but I'm curious if there would be better in every other way (speed, noise, precision, no moving parts, no grease). Preventing bed from dropping hard when unpowered is an issue too.
A linear motor powering the Z axis doesn't fix anything. It would have to remain energized to stop the bed from falling. When power is cut, the bed will drop.
I don't know how ball screws in the Z axis could have backlash unless there's an alignment problem (screws not parallel to guide rails, guide rails not parallel, maybe both). Belts don't wobble, and a simple worm gear drive will prevent bed motion when power is cut. I use a 30:1 worm reducer in my printer with a 695 mm belt lifted Z axis. The motor is driven directly by the Duet2 Wifi board, and I don't need any special config for brakes, etc. It uses two linear guides and two belts tied to a common 8mm drive shaft. The two belts are just simple loops. There's no backlash. Full step distance is 20 um and maximum speed is about 15 mm/sec. It just works, every time.
-
The backlash in my case may have come from linear rails combined with a ball screw and a stepper motor. Stepper motors can't quite perfectly instantly reach and stay at commanded position. The torque is very low when stepper motor rotor is very close to position it is suppose to be and this causes some springiness, which shows as backlash. The particular linear rails that I used when I tested this were quite stiff and not very smooth, which caused more load on the system making the backlash more pronounced. This kind of backlash isn't big problem and doesn't show on prints because normally you are only going one way on Z-axis during printing. However, this could cause problems for Z-hopping and nonplanar printing.
I can't remember the exact backlash but I think it was around 0.01-0.03MM (using SFU1605 + Stepper + disc shaft coupling).
Also I had my endstops at max end and therefore I had to home to max on Z-axis, which took maybe like 20 seconds to do due to low speeds. My Z-motors would stall if I went any faster.
