RepRap 3DR Delta printer Released

Been busy building a 3DR and now have everything printed out.

I found a great way to make up the rods that do not require the adaptor.

I used carbon fibre tube with threaded rod bonded into the end of it, once the glue is set this then screws into the traxxas joint and will not come loose.



Finished rod end



There's some more info over on my build thread at FizzFab - [fizzfab.com]

Had a lot of trouble sourcing the Bosch extrusion here in Australia, Managed to find some in the end but would recommend reducing the size of the web to work with a 5mm slot as all of the other brands of 20mm by 20mm extrusion use a 5mm slot not a 6mm like the Bosch. Even better would be to remove it altogether as Nicolas Arias has done. this will work with any extrusion and also RHS too.

DM/

Nice going DM!

I found the slot webs make for a tight and square fit of the extrusions (crucial to the precision of the printer), so I wouldn't want to miss those.

As mentioned in a previous post, I built the connecting rods similar to yours (but slightly larger Traxxas parts and 5od/4id carbon tube. What I didn't like is the weight of the set screws, I should use a carbon rod instead next time. Generally I found that, with those kind of printers, every added gram of moving weight will reduce the quality of your printed parts and/or the speed you can print.

Currently doing some modifications - added a heated bed and replacing the extruder by a planetary motor type (as seen on Johann's Mini Kossel). Although it worked quite well in my 3DR, I didn't care much for the bulk and also for the added slack (thus hysteresis for filament rectracts) of the M8 hobbed bolt in the bearing. You suspected problems from the outer bearing not being well supported: didn't find an issue with that because of the force of the idler pressing down on it.

Next thing to contemplate would be an automatic Z height probe (also as seen on the Mini Kossel), though this will add dead weight to the printing platform. Alternatively, I think about adding a quick & easy calibration function to the Marlin firmware: cycle through four positions (close to the X, Y and Z towers and center) and each time use the panel knob to adjust the Z height so a paper strip will just catch between nozzle.and bed. The resulting offset values would then be stored in EEPROM.

Hade,

are you using a recent marlin, with the M666 command ? (I think it was implemented 2-3 weeks ago)

M666 allows you to set an offset after homing, saves messing with mechanical adjustments on endstops (or messing about adjusting hall sensors). My hall sensors aren't adjustable, so it came in really usefull. You can only add -ve adjustment i.e. lower each tower carriage after homing.

I was using it and pronterface to do what your suggesting regarding nozzle and paper (my lcd isn't connected,still haven't designed a case to hold my ramps and :LCD)


Heated bed? sounds good got any pics? ,was thinking of a "luke warm" bed for mine so I can use PVA on glass, don't really need too high a temp for it to stick well. did you use a silcon heated mat ? what did you use to insulate the PLA base?

Edited 3 time(s). Last edit at 09/27/2013 06:32PM by Masnachu.

I added pictures of the new extruder and heated bed to my Flickr 3DR album:

[flic.kr]

No, my FW is based on this Marlin branch: [github.com]
I've seen the M666 command, simple enough... but I'd like to have (read: write myself ;-)) a user interface for display/knob for doing the actual adjustment procedure - the storage and usage of the adjustment values would be quite similar to the M666 implementation though.

As for the heated bed: just an experiment... actually two self adhesive heating foils, pretty mild with 15W each, a larger one with 160 mm and a smaller one with 80mm diameter layered on top of each other. I can plug in either one, the larger will heat up to 60°C and the larger one beyond 90°C. The insulation is cork. We'll see how it works, have just finished the modification but no prints yet.

Edited 1 time(s). Last edit at 09/27/2013 07:08PM by HaDe.

Hi,

I'm building a 3DR but would like to have a bigger XY area (bigger bed), anyone knows what are the rules I should follow in order to build it correctly? I guess the only thing, apart from top and bottom mounts, is the carbon fiber rods length that have not idea what's the relationship between them and the XY size... Any clue?

HaDe Wrote:
-------------------------------------------------------

> As mentioned in a previous post, I built the
> connecting rods similar to yours (but slightly
> larger Traxxas parts and 5od/4id carbon tube. What
> I didn't like is the weight of the set screws, I
> should use a carbon rod instead next time.
> Generally I found that, with those kind of
> printers, every added gram of moving weight will
> reduce the quality of your printed parts and/or
> the speed you can print.

This is a good point and one that I have pondered..

For me, the biggest advantage to using a separate bolt and a hollow ball type of rod end is that you can crimp the rod end to remove any play (see my build thread for more details on how to do this). Removing the play from the rod ends then allows you to be able to remove the backlash springs which is a saving of at least 2 grams per axis (depending on spring size), add this to the 4 grams saved for not needing the rod adaptors and thats a 6 gram saving per axis. I'm sure that this is much greater than the additional weight of the M3 bolts and slightly larger rod ends. (The weight difference between the M3 and M2 bolts is not measurable on my scales.)

A completed rod of my design including the M3 nuts and bolts weighs 7 grams. This uses a 120mm long carbon tube (I think that it is possible a little lighter than this - maybe 6.7 / 6.8 grams but my scales do not have the resolution to measure less than 1 gram increments).

I would be interested to know how much a standard rod weighs for comparison.

Incidentally, the Traxxas rod ends I measured had 0.5mm of play so the backlash springs are definitely a good idea if you do not want to crimp the rod ends.

/DM

Edited 1 time(s). Last edit at 09/28/2013 01:00AM by DeeEmm.

From my experience with model helicopters, crimping would help temporarily but the slack will come back :-) ... ok, forces are much lower here so we might get away with that.

Before the Traxxas, I tried Igus KBRM-02 industrial rod ends - NO play but VERY tight so the M2 screws kept rotating and I had to tighten those so much the balls deformed - no good. Maybe I try KBRM-03 MH next which are larger but the 3mm ball bore is metal sleeved so the mounting screws can be tightened properly.

[www.igus.de]

Then there are also (internally) spring loaded aluminum rod ends which look promising - below you see M2 but there are M2.5, M3 (two variants with M2 or M3 thread) and even larger ones too:

[shop.lindinger.at]

I didn't have those in my hands yet and couldn't find dimensional drawings, but I suppose you could simply slip a carbon tube over the shaft and fix it with CA.

I also looked at industrial type rod ends but could only find steel versions locally. Not sure of the weight but no doubt much too heavy. The Igus stuff is really good quality I have used a lot of their energy chain with industrial robots.

The aluminium type also looks really good, I wonder what the weight is? Perhaps use a hollow 3mm rod with 2mm bore and use M2 studs as connectors? This could be very lightweight.

I also just finished some calcs on the torque requirements for the motors as I want to experiment with some smaller (2kg/cm) motors I calculated the torque requirements at 0.576 kg/cm not accounting for transmission losses in the bearings. I think that an extra few grams for better rod ends should not be an issue at modest print speeds, especially for a 4kg/cm motor like most people are using. Maybe it might make a difference at very high print speeds? Probably not.

One thing that puzzles me is that the marlin firmware says that it has linear acceleration. This is very bad for speed. Much better to have S-shape acceleration curves.

DM/

---

[deeemm.com]
[fabribles.com]

DeeEmm Wrote:
-------------------------------------------------------
> I used carbon fibre tube with threaded rod bonded
> into the end of it, once the glue is set this then
> screws into the traxxas joint and will not come
> loose.

The same approach was used here (with the traditional rostock). Moreover one end used right handed threaded rod and the other one left handed. The ball links did not have any thread. The end of threaded rod cut its own thred in the ball link. Since the threads on the opposite sides of the carbon rod have different direction, it is easy to precisely set distance between ball links. The ball links used did not have any play at all. And they still do not have any play after about 20 hours of printing.
These links were used: MP JET Ball Link (Ø7 mm, M3, M3 short); code: MPJ2451B


The rods with the ball links work very well so far.

Edited 1 time(s). Last edit at 09/28/2013 05:39PM by hercek.

Thanks for the pointer... I always found MPJET to make very fine modeling accessoires, alas those can be hard to find here in Germany - I certainly would like to give MPJ 2451B a try.

Just added new bed support spacers and a 140 mm fan.

Edited 1 time(s). Last edit at 09/29/2013 06:54PM by HaDe.

HaDe Wrote:
-------------------------------------------------------
> Next thing to contemplate would be an automatic Z
> height probe (also as seen on the Mini Kossel),
> though this will add dead weight to the printing
> platform. Alternatively, I think about adding a
> quick & easy calibration function to the Marlin
> firmware: cycle through four positions (close to
> the X, Y and Z towers and center) and each time
> use the panel knob to adjust the Z height so a
> paper strip will just catch between nozzle.and
> bed. The resulting offset values would then be
> stored in EEPROM.

OK, I do not read this thread. My previous post mostly happened because the pictures made me read DeeEmms message. But I read a bit more of it now. So here is a comment.

If your geometry is correct then you do not need to measure the center of the bed. It is anyway useless without some math to convert it properly to either diagonal rod or delta radius adjustement. Math for this is simple though (if you assume endstops are set correctly which is a big if). Anyway, my point is that once a printer is properly calibrated, than there is no reason to measure bed center provided you do not change platform, diagonal rods, cariages or towers themselves. Hardly anybod does this with any non-neglibe frequency.

Or maybe you want to measure whehter your bed is bent? It should not happen with well mounted glass.

It looks like people sometimes change bed or hot ends. If so then they may need some support (z-probe) for quick endstop adjustements. Only measurements at each tower base are needed for this. That means only 3 points. If you would be changing only the hot end, then only one point is needed.

Also, if you do not measure the point near a tower base at the very edge of the theoretical built volume then the measured data cannot not be trivialy moved to endstop adjustements. So some math would be needed again. It would not be trivial if you cannot assume top endstops are precisely set.

I actually did M666 as a support for a bit of delta printer calibration mixed with a small math exercise. If you want to read about it more then start from here: [forums.reprap.org]
Some people asked me to try to push M666 upstream and that is the reason it finished in ErikZalm branch.

>> If your geometry is correct then you do not need to measure the center of the bed <<

Agreed... alas it actually *does* seem to be hard to find a really flat glass bed. I had one locally made from 4 mm glass and also ordered the Borosilicate one from Trinitylabs: both are slightly (and rather irregularily) bent, despite a 3 point moint which should not introduce bending loads. The total error is approx. 0.1 to 0.2 mm which will invariably show in larger prints directly on uncoated glass. It's much less of a problem when using blue tape though.

I'll receive and check out a sample of a surface ground carbon fibre bed (made from high temp prepreg material) soon, should be interesting.

@ Hercek. Thanks for the heads up on the MP JET Ball Links, will check to see the availability here in Australia. Would much prefer to use a link that does not need to modified to remove the play.

@HaDe. Has anyone considered / tried grinding their own glass beds to get them flat? Maybe a ground finish would also be good for print adhesion? I might give this a go.

HaDe Wrote:
-------------------------------------------------------
> The total error is approx. 0.1 to 0.2 mm
> which will invariably show in larger prints
> directly on uncoated glass. It's much less of a
> problem when using blue tape though.

Interesting, do you have any information how flat mirrors are?
I would assume producers take more care selecting flat glass for mirrors. But may be not.

The printer I did the callibration excercise with uses a mirror and I did not notice any problem with its flatnesss. Unfortunately, I do not have any way to measure it. The 3dPrinter does not tell me how flat the mirror is since it was calibrated against it. And I do not have something else about which I could say it should be really flat/straight.

Haven't tried mirror glass yet... as mentioned I should receive a carbon bed soon, will see how that one works. Ground glass sounds good, too.

As for measuring flatness: to get a rough figure, I rocked a straightedge along various axes of the bed surface (that it could be rocked at all made me suspicious in the first place) while observing the upper edge of the straightedge with a dial gauge.

You could try covering it in dry erase marker and then move a straight edge such as a metal rule across the surface.

HaDe Wrote:
-------------------------------------------------------
> As for measuring flatness: to get a rough figure,
> I rocked a straightedge along various axes of the
> bed surface (that it could be rocked at all made
> me suspicious in the first place) while observing
> the upper edge of the straightedge with a dial
> gauge.

I have actually done this with a cheap plastcik ruler. I did not notice any problem. But I do not consider my measurement reliable since:
* it could not be roled but this detects only concave surfaces; if the surface is convex I would not notice; I doubt my eyes are good enough to notice e.g. 0.1 mm distance from a ruler in the center
* I do not believe much that the ruler was stright - though probably not an issue sice probability the surface curvatuve would always match ruler curvatuve is low (I do not exactly remember what positions I tried)

I think I can safely assume the mirror surface is not concave for now.

DeeEmm Wrote:
-------------------------------------------------------
> You could try covering it in dry erase marker and
> then move a straight edge such as a metal rule
> across the surface.

Nice!
I'll try this (or some variation) when I'll be changing kapton tape the next time.

Hi guys,

Is it possible to make 3DR to use microswitch endstops instead of expensive Hall effect sensors? What changes should I do to put microswitches in and use?

Am I right, that basically, when using microswitches I'll just need to precisely drill two holes to mount them in the upper plastic triangle, just above where the carriage reaches top?

Yes microswitches are possible and a simple modification. All you need is a nice place to mount them where the carriages will trigger them safely. I think this is the reason for Hall effect. These don't require contact which should extend the life of the machine. The switches are wired the same as any other machine. Some choose NC and some NO. There are also breakout boards but unnecessary.
Default configuration prescribes Hall effect though. this means you'll have to whip out sketchup to make the modifications to suit the new switches.

Masnachu Wrote:
-------------------------------------------------------
> what aluminium profile are you using ? , the
> printed parts are designed for Bosch-Rexroth (BR)
> profile, using different profile would explain the
> size discrepancy

The extrusion is the same. The printed part does not have the dimensions of the part I have measured in sketchup. It is also not the slicer because I have checked the G-Code and measured the path, which is also correct.

The problem is that the plastic expands and so I get a thickness of 6.50 mm instead of 5.xx mm...Or the nozzle is not ok. I don t know yet. I will try a new hotend next week.

I have solved this by down resizing the plug. And now it fits. The pulleys ID holes also do not fit 4 mm instead of 5.xx mm. I will bohr them to enlarge. But the holes for the line are so small, this will not be easy thought...

Thanks for the help!

Edited 3 time(s). Last edit at 10/03/2013 09:18AM by zacbot.

---

video of my 3DR

CidVilas Wrote:
-------------------------------------------------------
> Yes microswitches are possible and a simple
> modification
....
> Default configuration prescribes Hall effect
> though. this means you'll have to whip out
> sketchup to make the modifications to suit the new
> switches.

A3144 Hall switch clones (OH3144) are incredibly cheap on eBay. I'm using three of these on some scrap perfboard cut and filed to fit the existing board mounts. Works great with the M666 software endstop trim in Marlin.

Yep,was going to ask why you wanted to use expensive mech endstops

Have done the same as pokey, seems to be working fine. Just connect VCC,GND and Signal from the hall sensor directly to the same pins on your controller board, and off you go.


good news on getting the prob sorted Zacbot

Edited 1 time(s). Last edit at 10/03/2013 06:04PM by Masnachu.

Almost done with mine. Here are some pics of the process

The hobbed bolt is a bit larger and I had to add extra washers to prevent the extruder from bumping into the carriage.

I passed extra cables thru the columns to place a female USB in the base (it has some erratic behavior, probably from interference from the motor's cables) and a power connector.

The MAX Z I get is 217mm with 3mm cristal, which baffles me (it's supposed to be 211mm). But I measured it many times and is correct

Working on the hotend now. So far so good. Not bad for my first 3D printer

Looks fine! (and nice choice of color ;-)

My intent was to have a USB socket at the base, too... but the space inside the columns is too cramped already.

217mm Z... mine had 200mm with hotends.com J-Head and a 3mm bed, now 194mm because of the heater & insulation. Column length is 480mm.

Keep us updated...

If you're having interference problems you may try twisting the stepper motor wire pairs (one twisted pair 1A+1B, and a separate twisted pair 2A+2 -- it's recommended anyways for stepper motors so won't hurt anything.

HaDe Wrote:
-------------------------------------------------------

> 217mm Z... mine had 200mm with hotends.com J-Head
> and a 3mm bed, now 194mm because of the heater &
> insulation. Column length is 480mm.


Ah, there you go, mine are 500mm. there is the 20mm difference, and maybe I set my endstops bit too close while calibrating and ended up with that 217mm mark.

Nyarlathotep Wrote:
-------------------------------------------------------
> If you're having interference problems you may try
> twisting the stepper motor wire pairs (one twisted
> pair 1A+1B, and a separate twisted pair 2A+2 --
> it's recommended anyways for stepper motors so
> won't hurt anything.

Yeah, I'll try some twisting, but I won't pass the cables again (pain in the ass). The USB works fine until the motor in that column moves
Maybe some ferrite cores would help to kill the interference?
The plan is to use it mostly with the SD card, so is not vital to make it work.

Hey guys. Has anyone gotten a 3DR going? I need some pictures on how to attach the LCD, and setup the hotend with the extruder. Lastly, pictures or something on the electronics with the board. If you could please let me know/help that would be awesome! Thanks!

You can see how I setup mine in my gallery

For the LCD I used M3 screws, but some of the holes in the case and display are 2 or 2.5mm, I had to make them a bit wider. The fan I had made it difficult to pass the LCD cables, so I used wire ties instead of screws, that left some room to pass the cables.

My J-Head mount had some overhang inside (where the hotend fits), I had to heat and drill some there (not happy with the messs, but it works). Also you have to drill the M4 holes deeper so they hold the hotend in place, and more drilling for the filament hole (2mm).

I aligned the electronics towards the front and where the fan will be (opposite to the usb plug). I had to cut down one of those screw columns so it didn't push the board up.

My 3DR is printing now, but I think it still needs some work on the extruder feedrate or printing speed.