Showing off - Video of my Mega Delta printing CNC parts in PLA

Alternate subjects:
Delta's can be fast..... and accurate
or
How not to print PLA by the book.

One weak point I've identified in my design is the belt joiner and tensioner set up. To fit the space the printed part is more fragile than i would like with two failing already. I've decided to have them made from Aluminium now and nearly hit buy the other day on a mini mill to do it myself before deciding I may as well just build a CNC Mill after seeing the Mostly Printed CNC on thingiverse.

The video (linked below) shows that I'm getting high quality prints from PLA at 90mm/s, what you can't see but i've measured, all parts are within .1mm of expectations with only the small M3.5 holes being fractional small.

Slicer: Simpliy3D
Nozzle: 0.4mm
Filament: cheap PLA, about 6 months old and unsealed (what not to do)
Retraction: 1.2mm @60mm/s
Layer: 0.2mm - 3 perimeters
Infill: 55% full honeycomb
non print XYZ travel at 350mm/s

Duet:
M201 X5400 Y5400 Z5400 E1000 ; Accelerations (mm/s^2) Set all E-motors the same
M203 X24000 Y24000 Z24000 E3600 ; Maximum speeds (mm/min) Set all E-motors the same
M566 X2600 Y2600 Z2600 E1200 ; Maximum instant speed changes mm/minute. Set all E-motors the same

Video: [vimeo.com]

Nice one ! Don't suppose you are anywhere near Canberra are you?

Quote
deckingman
Nice one ! Don't suppose you are anywhere near Canberra are you?

Depends if you consider Flynn to be in Canberra.....

Quote
aussiephil

Quote
deckingman
Nice one ! Don't suppose you are anywhere near Canberra are you?

Depends if you consider Flynn to be in Canberra.....

For a Brit, it's in the ACT so that's near enough. My daughter moved out there a couple of years back to be with her Kiwi boyfriend. They were in in Gungahlin but have recently move to Casey. They are getting married in September so I'll be out there for a couple of weeks for the wedding. It'd be good to see your printer in the flesh but thinking about it, I doubt if I'd get a pass. It could end up being a dual ceremony - one wedding and one divorce.

Quote
deckingman

Quote
aussiephil

Quote
deckingman
Nice one ! Don't suppose you are anywhere near Canberra are you?

Depends if you consider Flynn to be in Canberra.....

For a Brit, it's in the ACT so that's near enough. My daughter moved out there a couple of years back to be with her Kiwi boyfriend. They were in in Gungahlin but have recently move to Casey. They are getting married in September so I'll be out there for a couple of weeks for the wedding. It'd be good to see your printer in the flesh but thinking about it, I doubt if I'd get a pass. It could end up being a dual ceremony - one wedding and one divorce.

Sorry for the flippant reply, for some reason I had it in my head you were a local.......

Casey is just 10 minutes away so please come over, September should be getting a little warm. I'll bet you'll not only get a pass but be told to enjoy yourself. By that time I hope to have a mostly printed CNC up and running as well.

So cheers and hope to meet you in September

Is it OK if I PM you?

Quote
deckingman
Is it OK if I PM you?

Yep, anytime

Surprised that the video hasn't raised questions or comments in general.

Last night I finally seen the effects of belts getting a little loose, looks like the adjusting screws used with the belt clamps backed off a little plus the belt clamp/adjuster needs a small redesign. This resulted in some corners with backlash type misalignment.

I've ordered some nyloc nuts to use now, hope to have them in a couple days.

Quote
aussiephil

Surprised that the video hasn't raised questions or comments in general.

I'm not. If it'd been a pile of crap you'd have had plenty of comments. As there is nothing to criticise, there are no comments. Take it as a compliment. (Sorry, that's just an old man getting more cynical every day).

looks awesome.

what was the total cost of the system? (if you've been keeping track, that is... I know sometimes it gets out of hand and you just forget about keeping track lol)

Quote
deckingman

Quote
aussiephil

Surprised that the video hasn't raised questions or comments in general.

I'm not. If it'd been a pile of crap you'd have had plenty of comments. As there is nothing to criticise, there are no comments. Take it as a compliment. (Sorry, that's just an old man getting more cynical every day).

I'll second that thought as well?

Doug

Aussiphile,

I read all of your build thread and I think your Delta has turned out amazing! I hope to build a Delta soon and yoursee is an inspiration!

Quote
deckingman

Quote
aussiephil

Surprised that the video hasn't raised questions or comments in general.

I'm not. If it'd been a pile of crap you'd have had plenty of comments. As there is nothing to criticise, there are no comments. Take it as a compliment. (Sorry, that's just an old man getting more cynical every day).

Hey I'm old and cynical myself, it sort of feels weird though as my main hobby of christmas lights tends to have a lot of success stories posted in addition to "oh crap it doesn't work"

Quote
Careless
looks awesome.

what was the total cost of the system? (if you've been keeping track, that is... I know sometimes it gets out of hand and you just forget about keeping track lol)

As above, xmas lighting can get ridiculously expensive each year and I try to not keep track with the only goal to make sure all the bills are paid each week....
A quick estimate (remember everything cost 3x in Aus well maybe 2x ).... some of this is poor memory
20mm rods - $240
20x40 extrusion $230 (8x2M lengths) with most of one left over
cast plate bed 8mm x 600mm - $250
Extruder and hotend - $200
Silicon Heat pad 500mm - US $53
Carbon fibre arms - US $73
Bearings/Screws/nuts etc maybe close to $200
Nema23's - around $100 delivered
Belts/pulleys - $40
a couple kg of ABS - $60
MDF - $70
Duet/paneldue/7"screen/height sensor - $300 (roughly)

Looks like I have well north of $1700 invested in this build and more likely above 2k if i added all the receipts up... oh crap that was scary and I haven't finished spending money with the full heated enclosure to finalise.

This was never meant to be a cheap build but a high quality result build that could be used as a pure workhorse like I have done with the Flashforge.

Quote
BuildAThing
Aussiphile,

I read all of your build thread and I think your Delta has turned out amazing! I hope to build a Delta soon and yoursee is an inspiration!

Thank you very much....

I've gone back to the design board with all the lessons learnt from this build to design version 2 parts that are easier to work with building a machine. Redoing the corners and carriages to enable easier build and maintenance.

Really must add some more info to the build thread... was never good at keeping them up to date

If it makes you feel better, that ain't expensive. e.g, here in the UK 20x40 extrusion (V slot that is) is about £5/linear metre so 16 linear metres would set you back about £160. When I was over last Xmas, the exchange rate was about AUD2 to the £ so it works out about $320. Hmmm, I've got 40Kg luggage allowance when I come over in September.......................

Great build thread , I ´m impressed , I´m planning to build a simular sized Delta based on the german hexagon delta basics, however your design makes me rethink that route. So are you still planning to continue with that design version 2 ?

What made you decide to not go for a linear Hiwin carriage design ? aren´t such long rods (even 20 mm ones ) kind of risky regarding their straightness ? Why no Vulcano 0,8 to 1.0 mm nozzle with this kind of big-build space ?

anyway my compliments !!

Eflin

(lübeck germany)

Quote
eflin1
Great build thread , I ´m impressed , I´m planning to build a simular sized Delta based on the german hexagon delta basics, however your design makes me rethink that route. So are you still planning to continue with that design version 2 ?

What made you decide to not go for a linear Hiwin carriage design ? aren´t such long rods (even 20 mm ones ) kind of risky regarding their straightness ? Why no Vulcano 0,8 to 1.0 mm nozzle with this kind of big-build space ?

anyway my compliments !!

Eflin

(lübeck germany)

Hi elfin1,

Firstly thanks, it's been an interesting experience so far. Lets me answer the questions in order

So are you still planning to continue with that design version 2 ?
Yes the v2 design is really just a redesign of the top and bottom 2040 corners and new carriage design that allows for relatively easy belt changes. The new carriages are done and are going into my machine now but it is a significant pull apart but that has given me a chance the true it up better. I'm also replacing the 10mm steel core belts with standard 10mm GT2 belts

What made you decide to not go for a linear Hiwin carriage design ?
There was a couple of reasons, cost being one and the other being that rods could be sourced in Australia and a belief that they are the best option. (easy to debate of course)

aren´t such long rods (even 20 mm ones ) kind of risky regarding their straightness ?
If we were talking about 8/10/12 i would say yes but 16/20 there is a risk that they could get bent by the courier but that applies to rails/tslot etc at that length.... those 20mm rods came wrapped and packaged that the odds of being bent were very low and trying to bend one to see deflection, ah i'll never worry again
As I bought these from a mainly CNC supplier i wasn't concerned about straightness from the supplier.

Why no Vulcano 0,8 to 1.0 mm nozzle with this kind of big-build space ?
I have both 0.6mm and 0.8mm nozzles for the Dyzend but wanted to fully bed in experience with the 0.4mm in both PLA and ABS before moving up in nozzle size, I know what to expect and generally look for after 5000+ hours of ABS at 0.4mm on the Flashforge printer.
Besides most of the stuff at the moment suits 0.4mm but I do have a couple prints coming up that will be days faster with a 0.8mm

Now some thoughts.
Hiwin rails could make a simpler design but watching how the 2040 flexes with no load on it I'd want to use 4040 or better at 1.6M and longer.
I'm debating the top mounted motors but it does make practical enclosing with motors outside harder
I was surprised at the amount movement the entire frame has under short moves at high speed and high accelerations and how little it actually appears to impact on the print quality.
rock solid effector mounting is essential and i have zero joint movement with the ball/spring setup
GT2 steel core will fatigue with the large amount of small movements around the pulleys, made worse I believe with a couple times of slack tension that had the belt serious "flapping", I'm swapping to normal black belt to test.

Edited 1 time(s). Last edit at 07/10/2016 12:34AM by aussiephil.

That is some really nice printing and engineering, congratulations! You have inspired in me that my biggy-Kossel build can give higher performance. I have a few questions that might help me with my own construction:
My Design
Duet running on a 1M tall 350mm bed with 300 build zone, 300mm 500W silicone heater under 3mm borosilicate glass, black thermally-conductive double-sticky tape from heater to glass, no heat spreader (I could not get high enough temps using a spreader), nema 23 motors, 20x20 towers, extruded aluminum vertex pieces from Robotdigg, 80mm-spaced rods, 3mm filament with a geared-down stepper, some simple Bowden-tube strain-relief such as yours.

1) How accurate is your build? I have been trying to square, etc my system but getting (repeatable!) measurements of both ends of the rod spacing, tower squareness, etc below 0.2mm or so has been hard with just a 6" digital caliper. (I borrowed a long caliper from work but only for the original construction).
2) How is your bed stackup? I keep sticking with a glass bed because it is impervious and perfectly flat and really popular and inexpensive, but it has been difficult to heat properly and also mount solidly. You have a metal bed: is the silicone pad just sticking underneath with the usual adhesive that comes on them? How is the bed supported on the frame? This has been a nightmare for me because my silicone pad has a bump where the cable exits, making it hard to get a flat build out to the edge of the bed. But the tape gets soft when hot and the suspended heater pad starts to pull away, so I have cork and cardboard insulation then another metal plate. I wanted to mount the glass bed directly to the frame so now everything underneath is suspended and I have to put a light bed support under the middle, blegh!
3) Do the printed carriages slide ok on the steel rods? How do you keep the friction down? I also keep trying to use wheel-bearing-mounted carriages but that has also been a pain with warpage because printed carriages with wheels just warp too much; I have been using some polycarbonate plates to stiffen them. With my frame design I have to use the 20x20 towers as the rails.
4) You have printed vertex parts? What are your horizontal beams?
5) With those print speeds and heated bed and no cooling, how does not print not slump? I just built in a fan-shroud to direct air just under the nozzle and now I feel that may have been a waste of time, sigh.

After looking at your design choices I feel like I am back to square one, but I see a lot of custom parts on your setup and that flying extruder, as clunky as it looks, must help a lot with print-speed and quality. I'm still not convinced the performance of a normal Bowden setup can be fully compensated by just increasing the retraction settings, especially with jam-prone PLA, but you do it fine even at 90 cm/sec! I tend to stay around 30.

thanks

Quote
shadowphile
That is some really nice printing and engineering, congratulations! You have inspired in me that my biggy-Kossel build can give higher performance.
Duet running on a 1M tall 350mm bed with 300 build zone, 300mm 500W silicone heater under 3mm borosilicate glass, black thermally-conductive double-sticky tape from heater to glass, no heat spreader (I could not get high enough temps using a spreader), nema 23 motors, 20x20 towers, extruded aluminum vertex pieces from Robotdigg, 80mm-spaced rods, 3mm filament with a geared-down stepper, some simple Bowden-tube strain-relief such as yours.
thanks

Thanks appreciate that, the engineering itself could have been better as quite a bit was developed on the fly from sometimes random ideas and inspiration from other builds. let me try and answer your questions.

Quote
shadowphile
1) How accurate is your build? I have been trying to square, etc my system but getting (repeatable!) measurements of both ends of the rod spacing, tower squareness, etc below 0.2mm or so has been hard with just a 6" digital caliper. (I borrowed a long caliper from work but only for the original construction).

All my 2040 frame spacing elements were cut using a jig to be essentially identical and square, this makes the towers accurately spaced in both distance and angle. I have never measured between rod pairs instead relied on the cut accuracy of the frame, using a jig removes the need for long calipers for this. I did buy a 400mm per side digital square to check that the bed is 90 degrees to the towers. The towers were plumb bobbed to get them vertical and i've improved this method during the rebuild i'm doing.
My current vertical measured errors are sub 0.2% and aiming to get under 0.15% which is around 2mm over the 1600mm. I'd call this acceptable but can be improved but will have to wait till i have cnc capability.

Quote
shadowphile
2) How is your bed stackup? I keep sticking with a glass bed because it is impervious and perfectly flat and really popular and inexpensive, but it has been difficult to heat properly and also mount solidly. You have a metal bed: is the silicone pad just sticking underneath with the usual adhesive that comes on them? How is the bed supported on the frame? This has been a nightmare for me because my silicone pad has a bump where the cable exits, making it hard to get a flat build out to the edge of the bed. But the tape gets soft when hot and the suspended heater pad starts to pull away, so I have cork and cardboard insulation then another metal plate. I wanted to mount the glass bed directly to the frame so now everything underneath is suspended and I have to put a light bed support under the middle, blegh!

Glass beds seem to be the flavour but with the problems heating glass due to low heat conductance I decided Aluminium was better and I was used to it anyway on the small printer. Let me describe the base of the printer from the ground up, remember I have plenty of vertical height to play with..
Base - 32mm High density fibre board (MDF in OZ)
Printer base frame - 2040 Tslot, printed corners, tslot is secured to mdf via 2 M5 bolts on each of the six sides, corners are screwed down to this as well to reduce movement.... welded aluminium corners will be the next upgrade. Steppers also are bolted down to the 32mm base.
Now the heatbed sectio your inerested in....
Next - above this sits a 32mm MDF bed base, this is three point adjusted to be level and square to the towers using 90 degree brackets underneath bolted to the 2040 verticals. This base is essentially free floating as it's not attached to the brackets... it's weight alone holds it in place
Next - 12mm layer of cork that has a groove carved out for the cable connection to the silicon pad.
Next - two 3mm layers of ceramic fibre paper are used, layer closest to the cork covers the entire heat bed edge to edge, the layer closest to the heat bed has the center cut out so it fits around the heat pad.... the heatpad is around 3mm in thickness.
Next - the silicon heat pad (1200w) is stuck directly to the aluminium plate using the supplied adhesive tape that was attached to it.
Then we have the 8mm plate, this makes a total bed stack of around 50mm thick but it is stacked so that everything just lays flat...
BTW, as the bed doesn't move and the plae itself is heavy, none of the above stack is bolted or clipped to the frame.
I firmly believe that for a delta setting up a base to stack the cork/heater/print plate onto makes sense and actually makes it easy to deal with the "bump" of the 240v connection.

Quote
shadowphile
3) Do the printed carriages slide ok on the steel rods? How do you keep the friction down? I also keep trying to use wheel-bearing-mounted carriages but that has also been a pain with warpage because printed carriages with wheels just warp too much; I have been using some polycarbonate plates to stiffen them. With my frame design I have to use the 20x20 towers as the rails.

I'm using LM20UU linear bearings with the carraiges/rods and my test of both smoothness and alignment is to place the carraige at the top of the rod pair and let it fall to the bottom under it's own weight (not attached to belt), I've seen no sign of jerkiness indicating that rod spacing is good and bearings are correctly lubricated and smooth. Straight from the factory they were crap and needed to re-lubed.

Quote
shadowphile
4) You have printed vertex parts? What are your horizontal beams?

Answered somewhere above but they are 2040 t-slot.

Quote
shadowphile
5) With those print speeds and heated bed and no cooling, how does not print not slump? I just built in a fan-shroud to direct air just under the nozzle and now I feel that may have been a waste of time, sigh.

At this time of the year the ambient temperatures are under 20c, I doubt i could get away with even printing PLA in summer when it goes above 40c in the same area...
I always planned to be printing ABS so any PLA printing has been secondary and one of the reasons I never designed a fan setup... having printed about 8 rolls of PLA now the only need a see for a fan is to get better bridging performance and to print with higher ambient temperatures.
I learnt with ABS that the plastic layers need time to cool enough for the next layer... in the video i'm printing quite a few parts so the layer that goes down has more than enough time to cool before the next layer goes down..... small parts I always print in muliples or carefully adjust all the speed settings in the slicer. Fans do have a place, especially bridging performance.... I tend to design and slice to eliminate or reduce any bridging required.
I've got a 30mm single wall cube that shows no sign of "slumping" despite being printed by itself at around 80mm/s

Quote
shadowphile
After looking at your design choices I feel like I am back to square one, but I see a lot of custom parts on your setup and that flying extruder, as clunky as it looks, must help a lot with print-speed and quality. I'm still not convinced the performance of a normal Bowden setup can be fully compensated by just increasing the retraction settings, especially with jam-prone PLA, but you do it fine even at 90 cm/sec! I tend to stay around 30.

Hopefully I've given you some ideas. Yes mine is essentially all custom parts.
The flying extruder is "clunky" but i'm working on it as part of the redesign, despite the clunkyness it performs really well and certainly allows very short retractions to be effective....
well they seem effective based on the lack of blobbing but some of that is affected by the high travel speeds so even if there is a a little ooze it has so little time during the high speed travel.
I had dismissed using a flying extruder originally and was always going to mount the extruder on the effector and the only reason that didn't happen was the Titan extruder and Dyzend hotend do not neatly slot together and I was forced to use a short bowden tube hence the on the fly design

Cheers
Phil

Edited 2 time(s). Last edit at 07/12/2016 10:09PM by aussiephil.

Looks like a great printer. I'm a big advocate of flying extruder and have designed (using the term loosely) a bracket for a titan extruder. Yours does seem to wobble about a lot but perhaps this doesn't show in the prints as the effector and arms have much more mass than a little kossel mini. The normal scheme isn't going to work though as the rods are long and the extruder would be quite a way above the hotend meaning a much longer bowden tube. Why not just mount the titan as a direct extruder since its not really suspended the way you have it configured anyway?

Edited 1 time(s). Last edit at 07/19/2016 08:20AM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Quote
DjDemonD
Looks like a great printer. I'm a big advocate of flying extruder and have designed (using the term loosely) a bracket for a titan extruder. Yours does seem to wobble about a lot but perhaps this doesn't show in the prints as the effector and arms have much more mass than a little kossel mini. The normal scheme isn't going to work though as the rods are long and the extruder would be quite a way above the hotend meaning a much longer bowden tube. Why not just mount the titan as a direct extruder since its not really suspended the way you have it configured anyway?

Really frustrated with the printer the last few weeks, looks like i may have failing stepper motor.....

The consequence has been though a near full rebuild including a rethink on the suspension of the extruder, haven't changed the basics but changed the spring arrangement in the middle to longer but stronger springs. The whole extruder flying arrangement is also counterweighted. The movement when seen in person is not that bad and I think less than some videos i've seen.

I was and always intended to mount this as a direct extruder however the Titan does not directly attach to the Dyzend hot end so that plan was shelved for the time being.

AussiePhil,
I'm quite impressed with your results - and from my experience I know that engineering a printer that large - that works well is difficult, time consuming, and yes, can be frustrating.

And also rewarding! Well done, keep at it!

I see you are getting very good results with the Duet. 32-bit is the only way to go (I use a Smoothie and like it).

The flying extruder mod is an interesting development to watch. I've gone in a different direction, using post-processing for pressure compensation with a bowden. It is working very well, and I'm working on v3 of the software using the things I've learned (probably no end of that).

Did you have any difficulty adapting S3D for your printer? I'm interested in how S3d is working for you as your printer is also custom, and S3d may solve some problems for me as I will relate below:

You have the capacity to print tall, have you tried printing very tall items with support yet? The reason I ask is: even though my printer has no mechanical or thermal problems printing tall (it's heated and enclosed) - good very tall support structures are always a huge problem. Ideally support could be designed into the part, however I do get already designed models I want to print that are without it.

Slic3r will generate tall, thin support that breaks while it's printing, so it's not suitable, and support that is too close to the edge of a model. I've used Slic3r up to now because of the wide array of very nice features and it's commented gcode that I use in my post-processor. I was also able to import my custom bed shape - very handy for part placement. But I need better support!

I know S3D has good user-definable support (I've watched the videos). If you haven't done it yet - do you think it would it work well to generate support over 300mm tall? If not, perhaps someone else will know?

I've tried working with Meshmixer to generate support:
- Meshmixer's support generation uses twiggy structures that are totally incompatible with printing tall - they are too bendy.
- Using manual support placement to make the support stronger is a total pain!
- Larger diameter twiggy structures are inefficient - as they are solid circular forms - not tubes.
- I tried combining objects to make support, but this is clumsy and problematic, and I do not want the support structure sliced the same as the rest of the object.
- And they did not bother making good documentation. Too bad!

I would have also used Hwin type rails as well if I could find/afford them at the time I built my printer. I used some precision V groove bearings and Makerslide. I don't recommend that method due to wear in the aluminum and resulting variations in friction. I plan to upgrade to Hwin type rails when I can. I end up having to fix all my price/performance compromises in the end...

EDIT:
I emailed S3D last night and they replied already this morning!. I asked them about Smoothie compatibility, a custom bed shape, other configurations for a custom printer, any limitations on support height generation, and gcode comments (I use Slic3r's comments to identify loop types in my post-processor)
I've included their answers here is I hope they will be helpful for others. Yes, I'll be purchasing S3D in a hot minute - the time saving to set up one big print will easily pay for it.

Hi Paul,
That looks like a very impressive printer that you have built. We have many homebuilt machines in our office as well, so the software can easily be configured to work with any of these DIY/homemade printers. I will try to answer all of your questions below.
1) Smoothieboard and smoothieware firmware are fully supported. We have several printers that use this firmware as well without any issues.
2) You can easily import a custom STL that represents the bed layout, similar to what you have done with other programs. In fact, many of our pre-configured profiles already include custom bed STLs that represent the build platform of the printer.
3) Yes, all of these things are easily configurable as well.
4) There are no limitations based on build volume. You could easily create supports up to 2 feet tall (as we work with many printers and manufacturers that have machines over a meter tall)
5) Our software already includes many useful comments in the gcode, and you could also add your own comments if needed. So I imagine you would have everything you need to perform similar post-processing.

Edited 2 time(s). Last edit at 08/23/2016 01:20PM by Paul Wanamaker.

---

My printer: Raptosaur - Large Format Delta - [www.paulwanamaker.wordpress.com]
Can you answer questions about Calibration, Printing issues, Mechanics? Write it up and improve the Wiki!

I haven't read through the entire thread yet, but quickly looked at the video. This is exactly the type/size of printer I want to make for myself after my current rebuild is (assuming successfully) done. Thank you for posting this and nice job!