Rapid Tool-Head changer (very light multi-extruder)

Someone I work with told me of his idea some time ago: an extruder that can be attached to the carriage of the Mendel90 with magnets. At that time, I was rather suspicious about how well the whole thing would hold up during fast travel moves. But now, Makerbot released their new printers with magnetically held extruders in vertical configuration even, which got me thinking that this idea might not be as crazy as I thought first.
On thingiverse, someone posted a dual-carriage bukobot that parks the carriage with the unused extruder on one side so that you 1. reduce moving mass 2. reduce oozing over the print

Combining both ideas, it should be possible to build a CoreXY or H-Bot based printer that can easily deposit extruder units and / or bowden-fed hot ends on one side of the frame and pick up modules as needed.
You probably don't even need to alter anything in the firmware (apart of enabling servo control - which is already supported AFAIK) - the support start- and end-gcodes should suffice to make the printer swap tool-heads. I am not sure yet how attaching and detaching work best - if the magnets are too strong, the motors may lose steps before they can even pull a module off the receptacle. So I guess using just the right magnets and a little barrier actuated by a servo to lock the tool-head in place might be doable. Or maybe someone can come up with something better? After all, if the magnets are equally strong both on the parking position side and on the carriage side, how except of with a locking mechanism can we ensure that the module is attached to the right side every time? Maybe with a directional parking receptacle as in: Tool-head is deposited by going to Y min and slowly going in the direction of X max until the parking magnets engage and a limiting wall doesn't allow for the head to continue moving with the carriage; picking up a head works exactly the other way round - going to X min where no wall blocks the head from coming off the parking position.

Making the parts mate in a defined manner shouldn't be too difficult. One could for example use three spherical magnets that mate with the heads of hex screws. I don't think that - besides parts that the screws and magnets are attached to deforming in any way - this allows for enough play or variance to be noticeable in the print.

This way, an arbitrary amount of extruders could be supported without making the carriage too big and heavy while keeping oozing and object deterioration due to several nozzles scraping across it to a minimum.

Do you guys think this might be possible to do?

I think it is possible. You would probably want to use some type of "kinematic coupling", which can allow precise repeatability even when part tolerances are far off. These couplings are often used in optical hardware. This page has some examples. In the classic example, one part has three balls, and the other part has three grooves. This would be especially well suited for 3D printed parts, because the balls could be off-the-shelf metal ball bearings, and the grooved piece could be 3D printed.

You could probably come up with some kind of stationary wedge that would separate the two magnetically coupled pieces when the printer drove them into the wedge.

And you might want to avoid using magnets as both the attractive force and the kinematic couplings (but I know that has worked very well in the magnetic joints for Delta machines).

This is the first thing that I thought of when I read about switchable permanent magnets. I'm not certain how well it would work but its out there.

[code.google.com]
[nicadrone.com]

This Printers Tool Changer might give you some food for thought.

Thank you guys for the links, I will have a look at them...

hehehe, food for thought... This method might work best for deltas, but if the mating parts are oriented in Z direction like on the Makerbot, the magnets have to be stronger to be 100% reliable, I suspect...
That being said, I am also considering captive systems like this, but don't know yet whether they are practical at all in a CoreXY configuration.

@MattMoses: Can you elaborate why magnets as attractive force and kinematic couplings may not be such a good idea?
The kinematic coupling with 3 balls and grooves sounds very much like what I had in mind with 3 ball magnets and 3 hex screw heads, but the printed grooves sound like a better idea now that I think about it. In theory, if I was to use ball magnets as couplings anyway, one could install a steel plate behind the printed grooves so that one can regulate mating strength by adjusting the distance of the steel plate.

The Electropermanent magnet sounds rather intriguing; although the assembly looks a little bit on the big side, it seems like a solution that fits right to the problem.

Edited 1 time(s). Last edit at 01/09/2014 07:14AM by uGen.

Linear extruder changer


Rotating extruder changer

Quote
uGen
@MattMoses: Can you elaborate why magnets as attractive force and kinematic couplings may not be such a good idea?

It is not necessarily a bad thing (and again it works very well in many Delta designs) but if you "de-couple" the task so that one part of the design holds the pieces together, and another part performs the alignment, it gives you more design choices.

For example, if you make a mechanism like this:



You can use common metal (or even plastic) balls, and the shape of the magnets is no longer especially important. This could be handy if you were trying to put this together out of stuff you had sitting around. But if you don't mind obtaining spherical magnets, then that may be the better option to use.

Spherical magnets in one piece, with a metal backing plate in the other piece, sounds like it would work fine to me.

Edited 1 time(s). Last edit at 01/09/2014 05:12PM by MattMoses.

MattMoses, I like this idea of a single stack of magnets. The force of the magnetic coupling could be largely negated by parking the tool over or under a stack of opposite polarity magnets mounted in the magazine. The only problem is, the magnets are where I want the tool body (extruder, digitizing probe, engraving spindle, etc.) To be.

Edited 2 time(s). Last edit at 01/09/2014 11:13PM by Dale Dunn.

Quote
Dale Dunn
MattMoses, I like this idea of a single stack of magnets. The force of the magnetic coupling could be largely negated by parking the tool over or under a stack of opposite polarity magnets mounted in the magazine. The only problem is, the magnets are where I want the tool body (extruder, digitizing probe, engraving spindle, etc.) To be.

So one magnet and a thin steel plate.

TCase

@ddseeker: Ah, I have seen the first animation a rather long time ago. Just didn't remember it. While I like the idea of a single motor to save cost, the penalty of that would be that you have to align the drive coupling precisely with the mating part. So either one would have to always calculate the current position of the motor coupling and save the positions of the extruder couplings or rotate the motor coupling arbitrarily until it fits...all of this doesn't sound very appealing to me. This and the rotating extruder changer have another drawback: you need an extra motor to move the stored tool heads. Of course, as soon as you have more than 2 extruders, the complexity/cost turns in favor of adding a motor for stored tool head movement. However, if the carriage has to be light and the printer is based around bowden extruders, it is more beneficial to let the carriage move to fixed positions where the tool heads are stored individually.

@MattMoses: I thought using the magnets as mating parts had mechanical implications that I wasn't aware of. But if it is purely a matter of design flexibility, I think that using magnets as mating parts will in my case reduce complexity and enable me to shave as much weight off the carriage as possible. Thanks for the insight!

@Dale Dunn: This sounds interesting. I will have a look at it - the simpler, the better is my approach for this problem at the moment...

I've been working on magnetic kinematic couplings for 3D printers. I think it's the preferred way to go for multiple extruders, especially for delta printers, where you can more easily sacrifice vertical spacing for your extra tooling.

My progress has been slow, and I'm a bit stuck on how to make the geometry of the nozzle holder work so that I can repeatably/reliably engage/disengage the tooling. Basically, I want to maximize the allowable positioning error when I do my tool change move. I've been having some issues with the bowden cable and wiring getting in the way, especially if I want to disengage the tooling closer to the top of my Rostock workspace.

Using steel pins is a really easy/cheap way of making the base surfaces for the kinematic coupling. I'm using some steel balls left over from a Rostock mod for the tooling portion. They're attached to M3 bolts with JB weld, and I use heat-set, threaded inserts on all my parts as fasteners.

Really glad to see others also interested in fast tool change. I think the more common, static multi-extruder setup has too many limitations.

Maybe printing some kind of tapered groove to guide the tooling part into the head part could help with repeatability? After all, the final mating between ball and pins should be accurate, so the groove might only be needed for the approach.
Seeing your pin solution is very similar to using the side of cap screws that I can get my hands on easier, this might be the way to go.

One thing popped up in my head though, and it concerns me: if one was to use a bowden extruder on a non-delta platform, the bowden cable has to be rather long in order not to create enough tension to pop the tooling off the carriage. Ideally, the PTFE tube has to run a long stretch vertically before entering the nozzle holder. In an enclosed printer, this might be a rather difficult setup...

The mechanical aspect of a tool change package like this hasn't ever been the stopping point for me. It's the electrical contact situation. I need some sort of "laptop docking port" connector that is small enough to fit an extruder, but large enough a human can attach wires to it somehow. DB9 or DB15 panel mount connectors work, but they are really bulky. I want more than 4 pins because I want to run the hot end heater, fan, thermocouple, and a couple accessories (servo / lights) through that same connector, and possibly have enough pins to allow for swapping in a dual extruder configuration or removing it and adding a single extruder for faster printing. Once you have a connection along those lines that you can build into the carriage and your extruder body, then it just becomes a matter of positioning and mechanics.

If anyone knows of a connector that would work along those lines, and more importantly where to get one, please let me know.

They make spring loaded contacts for this purpose, the industrial robot tool changers use them.

Try:

[www.mill-max.com]

They show them as type SLC

Are you looking at disconnecting the extruder every time it's parked? You'll have to wait on temperature to come back up every time you take it out of the magazine (Once per layer, I hope). I guess that's the price to be paid for not having a separate controller for each extruder?

I am also more for individual connections to each tool head. As Dale Dunn correctly states, one would quickly run into heating problems if one was to disconnect the hot end every time. Also, this might be rather fatal for all-metal hotends that require cooling even after switching off.

I can see this being useful for material type switches. Go from a standard hot end to a paste extruder to a flexable filliament extruder, etc. You could easily have the docking area provide power to the extruder to keep it warm if you didn't particularly care about regulating the temperature at that point, just keep it "Warm enough"?

I'll check out those spring loaded pins. that might do the trick!

Could you design a 3D printed interconnect. I'm thinking adhesive copper foil stuck to some plastic tabs that act like sprung contacts. Attached is a quick sketch of my idea. Orange =copper foil, Pink = wires.

With a bit of thought you could probably create a better design.

Simple coil springs landing on copper pads would probably do it. (With dielectric grease to limit contamination from arcing, or cut power during a change.) But you'll want a PID controller to maintain correct temperature, I think. Otherwise you risk burning something up or waiting several seconds every tool change. Or I guess you could fiddle with a current-controlled supply until you get something you're happy with at a given ambient temperature.

To control the temperature of docked extruders you could replace the heater/thermistor wires with power/data wires. Protocols such as i2c or SPI would allow you to address each extruder individually to control their temperature. This would have a few advantages:

• Fewer analogue inputs and PWM outputs required on master controller.
• Temperature control calculations (PID) offloaded onto each extruder.
• Minimise length of thermistor wires, thus reducing noise levels.
• Ability to control other devices such as cooling fans, servos, milling heads etc. without any additional external wires.

Each extruder would need it's own micro-controller, although something like a cheap ATtiny might be sufficient.

I still don't really get why you guys are so keen on detaching the extruder from a dedicated umbilical cord. You would need n+1 controllers (vs. n controllers) for n extruders since each one has its own parking position that needs to be addressed by one controller plus the one that handles the currently active one. The only advantage I see is that you only have one longer bundle of cables swinging around. The added complexity looks like many points of failure to me.

samp20 brings up something entirely different and exciting though: networked microcontrollers. What if every mechanical component (motors, extruders etc) has got an own built-in controller and the main µC just sends motion and timing commands to every component? This might be useful for "smarter" extruders that know their own PID / steps per mm / ... settings / retraction lengths etc and thus become plug-and-play and interchangeable between printers even or motors that can be fitted with some closed feedback loop, do their own motion planning etc. At least, this might free up some processing power of the main µC.
I don't know the details, but I think a plug-and-play capable protocol would be CAN-bus. If detachable extruders were implemented like this, I might get behind it more.