A different multi-extrusion approach - UM tool / printhead changer

[thefrog 7]

Here are the STLs for main body parts

Mechanical Parts will follow tomorrow

Edited February 29, 2016 by Guest

[thefrog 7]

There's a 60T gear from misumi (or I think it was 58T) that has a inner dia of 12mm. With that you could insert the mk7 inside and align the screws so both are together in love

I didn't went that path because I just found out after buying my 50T

 

Would still be to big ...

You have to remove the hub part of the gear completely

Both gears are glued onto the axle so I'm not 100% sure if they hold up well ... but I'm working on a mechanical connection between the two gears much like a claw clutch.

But my glue is capable of much higher torque, I'm just not sure about surface area :/

Edited February 29, 2016 by Guest

[foehnsturm 969]

That's what I made. If just glued to the axle it became loose over time.

[jonnybischof 60]

There's a 60T gear from misumi (or I think it was 58T) that has a inner dia of 12mm. With that you could insert the mk7 inside and align the screws so both are together in love

I didn't went that path because I just found out after buying my 50T

 

Do you mean GEFBB0.5-60-2-12-W7-H14-TPC3 ?

This could be nice, and very solid because you can put the parts together using one long setscrew. But you can only connect the gear's hub to the MK7, not the gear itself. That's not quite the same as foehnsturm did.

Anyways, you could try Misumi's Nylon gears (GEABM0.5), drill out their hub to fit it onto the MK7/8 and glue the parts together. The Nylon is easy enough to drill through even with "toy grade" tools.

/edit:

GEABM0.5-90-3-B-9 has a 9mm bore. But 90 teeth (min for that bore) is quite a lot. Would of course be great for torque, but that gear has a 45mm diameter.

Edited February 29, 2016 by Guest

[mastory 44]

That's what I made. If just glued to the axle it  became loose over time.

 

@foehnsturm & @thefrog,

Loctite (Henkel) make a variety of shaft locking compounds that should be far more than adequate for the torque these hubs need to handle. 638 comes to mind, but there are many options. If removal becomes necessary, add 250C heat. Otherwise, silver solder is also an option for metal hubs, sufficiently strong and easy to join.

[thefrog 7]

Thats what I used ...

It's rated at ≥25 N/mm² after 24h. So I suppose I'm good

Edited February 29, 2016 by Guest

[thefrog 7]

So now onto parts. You'll need :

1x 13t gear modul 1

1x 60t gear (modul 1, 4mm wide) -> I machined one side down so its only 3.6 wide but has a small hub part thats 4mm wide, Ø4 bore

1x Extruder drive gear (MK8 wheel) Ø9 OD, ca. Ø7-7,5 on the theeth diameter, 8mm wide, Ø4 bore (I've just hobbed the wheel with a M6 tap like

) -> if you switch to a different bore size (e.g. Ø5-6) you can buy MK8 wheels here

1x Ø4x18 axle (or whatever bore you did choose) -> cylindrical pin

2x flanged bearing 4x8x3 MF84 2RS -> here

1x 3x10x4 bearing 2RS version -> here

1x M3x7,5 threaded pin

1x ca.51,2-51,5 length Ø4 PTFE

1x Ø8x16 spring medium/high strength

[neotko 1,417]

Now that today my board heater 1 choose to die and I have finish packing one of the hotends since also soon I'll get a china um2 hotend to try on my umo+... I wanted to share some of my critical points about my mod.

Vibration

After many prints with my umo+ unmodded (except 1.75 ofc) I can clearly see that a four points fixed hotend delivers more quality than a magnet grip. On this last 2 months I haven't use the dual and I even added the fourth screw to hold the hotend evenly, it didn't make a difference. Also my design has a big flaw and it's that it's way too long, so the lateral moves did translate vibrations to the print. To see the difference you had to print the same stuff over and over like me, but still there was a difference.

Cooling fans

I been obsessed to get great overhangs at decent speeds and with my design I can get a perfectly centered airflow. This clearly it's the best thing of redesigning the hotend grip. And my x2 40x40x20 24v fans deliver enough air. Anyway without a propper air simulator it's hard to get anyclose to a crossflow fan solution.

Magnets

having the magnets just on a side it's ok, but at high speeds vs a fixed hotend it's not enough. I really think that a hardware controlled magnets would work much better allowing x/y magnets to remove any print shake vibrations. Ofc the difference in quality, as I said, you can only distinguish it by printing the same over and over again, but it's there.

Lack of room

The umo+ (um2) bed doesn't allow any real parking area. Only reliable place it's parking the heads near the bed bottom, and that means a) stronger grip to hold the hotends = slower head change or b) Redesign the feeders and move the cables so the bowden tension helps keeping the tools there.

And I mean lack of room to purge the hotends between head changes. Also, the lack of software doesn't help. I tryed some wipers that did somehow work, but they need programming beyond what I know and even with s3d it wasn't good enough (the wiper). The sollution could be drilling the glass on the parking areas or using a different size bed glass so the drip from the first layers didn't affect the hotends, because since they are only really hold by x or y with the magnet, a hit on the oposite angle of this magnets can very easily make the hotend fall or a speed bump with a nozzle full of drip on the tip.

Well that's my experience with my version of the mod. Now it will rest in a box as spare parts

Edited March 11, 2016 by Guest

[foehnsturm 969]

I have the magnetic coupler running on three different setups now. The newest one  uses 2 x 4 cylidrical magnets of size 6 mm in diameter and 3 mm in length. This one is really super-strong. You can literally grab the printhead and throw the x/y gantry whereever you want without any sign of disengaging. You're of course right that a longish printhead will be prone to vibrations (even without a magnetic coupler).

And yeah, the lack of room. So true with the Ultimaker (elserbot)  type gantry. Adressing this needs a smarter firmware, a modified (shorter) buildplate or an other type of dock.

For the wiper, I dare to claim that no one investigated thoroughly enough so far. That's my impression when I compare my latest design to all the other solutions out there. Sounds bold, I know But all the prints I showed recently were made with the tool changer and without using any prime/wiping towers or ooze shields.

The wiping device needs some more testing and isn't ready for publishing yet. But you'll get to know within a few months

Edited March 11, 2016 by Guest

[neotko 1,417]

Indeed if you where able to do the wiper, that's amazing. You can be bold, I googled a lot and the only decent sollution I found was that one I posted with the servo. I was going to make a wiper using the movement of the sideblocks but not anymore

[foehnsturm 969]

Priming towers and ooze shields are greatly overestimated

[neotko 1,417]

By the reflection on the bed that looks like a um2?

[SandervG 1,521]

Priming towers and ooze shields are greatly overestimated

 

Nice, very impressive

In the 'door frame', if that is the best word, it looks like some zig-zag between the layers and 2 colors. Do you think that could turn into a sharper transition by increasing resolution, or is it just difficult because it is such a small strand you are printing each layer?

Just wondering, you are kinda becoming the expert here!

[foehnsturm 969]

You are correct with your assumption. Thew door frame is about 1mm in square. If you want this to become better you have to switch to super-slowmotion. But this part has 8 of them on every layer so it would take ages. Additionally, I just threw S3D at it without any special optimization. I'm sure there's a lot of potential for better slicing.

[neotko 1,417]

If you want to separate the prime towers on s3d remember to ungroup the processes (not well documented s3d) this way tou can have a different x/y for each tower and different ooze shields (quite handy)

Edited March 11, 2016 by Guest

[macua85 52]

Foehnsturm, any more updates on your machine?

[foehnsturm 969]

@macua85

Unfortunately not, still working on the other project.

[foehnsturm 969]

With a new clip and after some polishing

[macua85 52]

Have you guys seen this: http://e3d-online.com/Titan-Extruder

Could be a nice package for easy use with this changer system. Maybe swap for a pancake motor and would be sufficient? What do you think?

[ultiarjan 1,223]

no idea if a pancake is sufficient, but I thought the same when I saw this... think a nema17 is enough to bend your axis

[neotko 1,417]

Have you guys seen this: http://e3d-online.com/Titan-Extruder

Could be a nice package for easy use with this changer system. Maybe swap for a pancake motor and would be sufficient? What do you think?

 

Indeed that should work but since it's only a 3:1 it might only work with 1.75 with a pancake. But the design it's quite compact and it should be easy to make an adapter like the one anders use with his pancake gear-less. For 3mm the pancake might work with a fan cooling down the motor and 1amp setting. But anders test show that the motor at 1amp gets quite hot without a fan.

 

And info about his mod here

[foehnsturm 969]

To me it looks like they are having an eye on this forum

Anders came up with a pancake stepper, I added the combined drive wheel / gear (which is almost exactly the same they are using), thefrog made a smart housing not so far away from that one E3D is offering.

Anyway that's ok. With 1.75 mm this design should work with a pancake, with 3 mm there are serious doubts. With the bulky and heavy stepper they use, you kill some of its benefits. But 3D printing currently is its SUV period (compared to cars) ... It's all about (feeder) power, no matter if you can use it

[neotko 1,417]

Indeed they read this forum. But it's nice to have a comercial version of the gears. Misumi it's very expensive.

[foehnsturm 969]

BTW just finished a nice print with soluble support.

Had no success with the Eames chair https://grabcad.com/library/eames-chair so far. Not directly because of dual extrusion issues, but consistently extruding just 1 mm of filament every 2 min (the legs) even with reduced idling temp and extensive priming seems to be demanding.

Edited April 23, 2016 by Guest

[ipcalit 5]

Did anybody try the tiny geared DC micro-motors from Pololu?

https://www.pololu.com/category/141/micro-metal-gearmotors-with-extended-motor-shafts

Some of them come with built-in encoders and between 30 - 120rpm should provide enough torque to drive 1.75mm filament. Other commercial product seem to use similar micro-motors for compact extruders:

http://nscrypt.com/docs/nScrypt-nFD-Brochure.pdf

At ~10g motor + 5g for bearings + 5g whatever spring/hardware you add, the package would be adequate for deltas directly on the effector.