Meduzas "bolt-on" belt geared feeder upgrade for UM2

[neotko 1,417]

Indeed. I did it before going to sleep so it needs testing. This way noone dares to try yet If I get time to print it I'll update

Edited March 1, 2016 by Guest

[neotko 1,417]

Well I updated the files. The design had a 1-2mm mistake but the gears position does fit and moves ok. I think this gears won't make much noise, but since they are for RC cars I would put some green lube on them.

All seems to assemble nicely, but I only got time to print the beta and update the design, no actual extruding yet =)

https://www.youmagine.com/designs/metal-geared-addon-inspired-based-on-meduza-bolt-on-addon

Edited March 3, 2016 by Guest

[neotko 1,417]

So far my first test was quite bad. Or the design it's 0.5 mm(something like that) off, making the gears stay too close to each other, or there's something weird with this um2 0.9 (bought to ultimaker when that was still possible). I can push more mm3 with my umo no geared feeder than with um2 with gears. I think I'll go back to the umo motor and pack this um2 nema on a bag. Anyhow, I'm not happy at all with it. I think I'll start to save for a Bondtech or a um2+ motor/gear if it comes available as a separate addon.

I'll upload the step files tomorrow if someone wants to continue with my geared version.

[neotko 1,417]

Well the problem wasn't the feeder but the ptfe coupler was dead. So it works I need to run some mm3 test.

[puremoco 15]

Since I have some issues with the "clicking" of the feeder and many aborted print, I have decided to upgrade the feeder.

A reduction with a pulley seems to me more suitable than with gears.

I have disovered meduzas feeder, but I am missing an easy solution to adjust the belt tension. Furthermore, it would be somehow difficult to order the exact belts and pullys here in Germany.

This is my first test, so maybe there will be a second version later.

The stepper will be replaced by a Trinamic NEMA17 (200 steps), which is optimized for microstepping.

I must have a look where to place the whole feeder, since I am using 2.2kg spools (30cm diameter)

Edited March 20, 2016 by Guest

[neotko 1,417]

I finished installing two gear boxes on um2 motor and umo motor. And it's quite weird. UM2 motor keeps giving me bad prints with errors on the retracts/print (40k retracts per print). I removed any other posible think by just placing the um2 + gear on the perfectly working umo gearless. And the errors pop. It's like if the um2 motor fails on a small but noticeable amount of movements.

I started to check about 1.8 vs 0.9 motors and found this info, maybe someone with more knowhow can check this.

As far as I see it by my prints, if I put the um2 motor on a already working machine, my prints get errors (with and without gears). I wonder if lowering the step for the feeder to 1/8 might help to have a more accurate magnetic movement from the motors. I don't get errors (that I can see) when the print it's 10k retracts, but at a full plate 40k for 6 keychains I see mistakes on infills and small letters (compared to two other umo+ with same gcode).

Well so far I would not recommend gears on 0.9 motor unless you lower the stepping to 1/8 + gear. But I been wrong in the past and I need more tests.

https://groups.google.com/forum/m/#!topic/deltabot/uuhSyAI81Hg

"Tim, that's accurate, microstepping only adds precision up to around 1/8 or 1/10 stepping. Beyond that, the angular size of each microstep is likely to be smaller than the natural error tolerance of the motor. Magnetic decent and fab precision issues (eg worn stator stamping dies) cause much larger position errors than a single 1/32 step.

Full steps and half steps are likely to be more accurate in absolute terms than microsteps, because they represent symmetrical magnetic field positions. The magnetic detent effects aren't trying to bias the rotor in a particular direction.  Quarter steps may have poor accuracy, that's when the permanent magnet is most trying to shift the rotor and the coil fluxes are quite unbalanced to boot. It depends on how the manufacturer optimized the geometry. Maximum holding torque designs may mean smaller than desired microsteps near full steps. It's a complicated science."

Edited March 22, 2016 by Guest

[puremoco 15]

@neotko:

For my camera rig, I also use steppers. I also have encountered, that microstepping is not suitable for precise movements. I posted a topic in the pololu-forum:

http://tinyurl.com/gm5eju3

At the bottom, you can see some tests, which I have performed at full stepping and 1/8. As you can see 1/8 is very irregular with a standard stepper.

You can also see, that the Trinamic stepper is performing much better at 1/8 than a cheap "chinese" stepper.

http://www.trinamic.com/products/motors/motors-stepper/qmot-qsh4218

The stepper drivers are also responsible for accurate microstepping. The A4988 in the UM2 are very old technology and there are much more sophisticated drivers, like the TMC2100 (aka SilentStepStick), on the market for the same price:

http://www.trinamic.com/products/integrated-circuits/stepper-power-driver/tmc2100

Unfortunately, the SilentStepStick-drivers can be only replaced with the UMO. The UM2 has the A4988 drivers directly on its PCB... Maybe there will be replacement in the next UM-generation.

Edited March 22, 2016 by Guest

[neotko 1,417]

Thanks @puremoco my digging on this it's a very see-print-see-repeat-check-repeat approach I don't have much knowhow that all I have read and learned this year. But there's something weird on the behavior. Specially when the retracts happen 30k in just 1h

[Torgeir 242]

Hi Meduza,

Really liked your way of doing it.. It's a "Roll Royce" feeder unit, imo. 

I made a very simple belt gear with a 16 and 32 theeth 2GT Pulleys. I used same belt as the two short ones in the UM2. Two small (outer dia. 10mm and inner 5mm) bearing a 5mm shaft from a model helicopter and two adapted 6mm plexiglass sheet. Sure there is 5 ea 18mm spacers in there too.

I have been printing for about 11 hour in this way, an it does not make much noise -only the stepper backtracking is noticed, but this is like the other steppers. I've made this with normal hand tools, not very nice but work very good.

Do you share your stl files for other? I would love to make your pro. ver.. 

Thanks for sharing your idea.

Regards

Torgeir.

[puremoco 15]

Finially had some time to print my version of a pulley reduction feeder:

Still need to print the parts for the feeder, since I don't want to use the original.

[neotko 1,417]

@puremoco video it's private atm

[puremoco 15]

@neotko: Ops, now I have changed it

[puremoco 15]

Now I have also finished the fedder from IRobertI:

As soon as I have finished my "to-do-print-list", I will change the feeder and run some tests.

[puremoco 15]

Finally I had some time to install the new feeder on my UM2:

Here is a quick print test before and after the upgrade:

- 40 mm/s

- 0.4 nozzle

- 220°C

- Colorfabb PLA/PHA

As you can see, the printer is extruding more filament, which is resulting in a more smooth surface. The print settings were exactly the same. With the old feeder, there were a lot of retractions (=ticks) of the feeder, which is an indicator for printing too fast.

With the new feeder, the extrusion is evenly without any ticks.

Some it is now possible to print faster. I also expect a better surface because the resolution of the feeder is now higher.

[SandervG 1,521]

Looks like great results! Perhaps bridging will be easier now as well. Have you used it for any 'real' prints already?

[puremoco 15]

I don't make a lot use of bridging, but I have noted, that retractions are a lot cleaner then before.

 

I did some "real" prints. The biggest difference until now is the the printing speed. With a 0.6mm nozzle I am not afraid of printing >100 mm/s, which was impossible with the standard feeder, because the "ticks"(retractions) were starting a 40-50 mm/s...

And the layers are looking more even/consistent, then before.

Until now, I did not had any extrusion or clogged nozzle problems like I had before regularly.

I wonder why you did not put an geared feeder from the beginning in the UM2...

For attaching the feeder to the UM, I have printed a small holder. I have used two of the steppers holes in the backplate of the UM to attach my feeder with screws and nuts.

This was just a quick-and-dirty solution, the holder could be integrated in the feeder-plate in the next version. Attaching the feeder this way has the big advantage, that is now more comfortable to use 2,2-kg spools, like you see in the picture.

Another improvent might be to mirror the IRobertI-feeder, so that the lever and spring-screw can be accessed more comfortable. I wonder why the feeder wasn't designed this way.

One problems, which I have encountered, is the heat of the stepper motor. I have reduced the current to 800 mA, but the stepper is still getting about 55°C hot. Unfortunately this is also the glass-transition temperature of the Colorfabb PLA/PHA, with which the housing way printed. So the whole housing might deform over time. The small wheel was printed with nGen, which can go up to 85°C. So I will alos print the housing again with nGen and place some spacer between the stepper and the housing.

[ultiarjan 1,223]

Another improvent might be to mirror the IRobertI-feeder, so that the lever and spring-screw can be accessed more comfortable. I wonder why the feeder wasn't designed this way.

 

I guess it was designed like this because mirroring it would require a firmware change to adapt the rotation direction of the motor.....

I like your design.... did you already publish the drawing & BOM somewhere ?

[puremoco 15]

@ultiarjan:

At first I want to implement the changes and print it agian. Once it is done and working, I will publish it.

Mirroring the feeder should be no problem because of the direction. As I installed the Trinamic-Stepper, I damaged the original Stepper as I wanted to remove the knurled wheel. It was really tight, even after removing the screw. So you might want to order a knurled wheel in advance I case it is stuck on the motor shaft.

So I just cut the stepper wires with the connector and attached the same color to each other, since the colors and phases are the same. But somehow it was turning in the opposite direction. By changing one of the phase (black against green) you can invert the direction permanently. So you just need to upload the UM2+ firmware, since it is the same gear reduction (1:3).

I will print again and keep this thread updated.

[neotko 1,417]

Guys anyone saw this Toranado gears? It's a 4.94:1 gears made in usa

http://m.ebay.com/itm/Toranado-Precision-Gear-and-Shaft-Kit-v2-0-/272354110557?nav=SEARCH

They are quite cheap for the gear ratio and could make the @Anders-olsson / @Foehnsturm pancake feeders to keep the weight to a minimum with more than enough gear power (and specially cheaper/lighter than e3d titan)

I'm tempted to buy one to finish my direct drive design. Also ofc it could be used with a redesign Meduza gear addon to keep a high gear ratio on um2 non plus feeders.

Edited September 3, 2016 by Guest

[ultiarjan 1,223]

don't really see how this would make something lighter than a titan? appart from a different stepper ofc

[neotko 1,417]

don't really see how this would make something lighter than a titan? appart from a different stepper ofc

 

The titan design it's 90g without stepper, a pancake motor it's 60-70g. So they use harder materials for all the plastics (yeah more durable). With a printed part I think that, IF that gears weight like mines from misumi that are 20-30g (It was long ago so I should check that again). Then with that other gears it should be possible to print a extruder more light with a higher torque gears (titan 3:1 toranado 4.94:1). Also more torque, more chances to use a pancake with 2.85mm filament).

I did order one so I can report the weight of that gears in a week or two.

Edited September 4, 2016 by Guest

[mariuszem 4]

I found this mod makes UMO extremly quiet printer. Old extruder was noisy, specially while retract. That is awesome (I use medusa 20T to 40T with GT2 110 belt).

[meduza 191]

@mariuszem: The sound level was actually one of the main reasons for me to design this with a belt drive instead of spur gears :-) And yes, the UMO feeder is really noisy on retracts.