Bigger 3d delta printer size

I am designing a 3d delta printer with 1500mm column height, but even after going through all the available data the actual calculations are not available for 1) distance between the two column (triangle length) 2) effective arm length. 80% rule i guess is not applicable for bigger size. I got calculations for kinematics but need basic geometry calculations for effective arm length vs triangle length. The concept i got from google is attached. Need help!

Delta Calculator

Try this. I used it to design my delta of similar height.

I think there are two choices to determine all of the lengths after deciding the print area. If you set the minimum angle of the arms to 20 degrees, do you want the effector to miss the belts or do you want the maximum print area with the shortest horizontals?

I suggest this approach:

1. Estimate how close the nozzle will be able to get to the drive belts in front of the towers. From that and the print radius you want, you can work out how far apart the belts need to be, and from that you can work out how far apart the towers need to be. Example: the nozzle on my effector can get to within 38mm of the belts. Add the required print radius of 150mm and 5mm extra margin, and the belts need to be on a 193mm radius. The belts are 35mm in front of the tower centres, so the tower centres are on a 228mm radius. Using standard corners this required 355mm horizontals.

2. Choose the rod length to put the rods at about 60 degrees to the horizontal when the effector is centred, and not less then 20 degrees to the horizontal when the nozzle is at the edge of the print area.

3. Choose the rod spacing. I suggest about 1/5 of the rod length. To small and you have less rigidity; too large and you sacrifice too much print area because the effector needs to be larger.

4. Check whether the chosen rod spacing requires a larger effector than you planned, and if so, go back to (1).

Edited 2 time(s). Last edit at 10/21/2015 05:22AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

... and strengthen the frame because (as it is shown on the picture) it would be a wobbly piece of ****

But you probably will be printing only very slowly anyway because it will be hard to find belts which are not too springy with one belt length of 2 * 1.5 m.

Maybe you should start with deciding what accelerations and what precision you want to achieve. Then you can derive how stiff your frame and belts need to be.

Quote
hercek
But you probably will be printing only very slowly anyway because it will be hard to find belts which are not too springy with one belt length of 2 * 1.5 m.

I don't see that. CoreXY printers have very long belts and are known to be fast.
Also the long diagonal rods will counter the disadvantage of long belts, because the steppers will move much slower to reach the same effector speed. Or is my imagination fooling me here?
-Olaf

CoreXY is faster compared to simpler cartesians (which e.g. haul around the whole build plate) but hardly compared to delta. Besides the print head assembly, it still needs to drag two rather heavy smooth rods around in one direction if you use CoreXY. With Delta, you drag around only the diagonal rods which can be much lighter because they are loaded almost exclusively in the direction of their long axe. In CoreXY the smooth rods holding the carriage are loaded in the direction perpendicular to their long axe. Therefore they must be strong compared to diagonal rods and also they must be hard to withstand the linear bearings movement.

Notice that your printer will not print fast if you keep your acceleration low. Models with small infill areas which require changing print head direction too often (while each segment length is only few mm long) will lead to print times which are very dependent on acceleration. With small accelerations you often do not achieve the maximum speed limit. E.g. the key-switch part of this keyboard case depends on acceleration a lot. Any model with a lot of holes will be similar.
The higher the acceleration, the higher the force. The higher the force, the bigger the bet elongation. The bigger the elongation, the bigger the initial amplitude of the dampened print head oscillations. And the oscillations will lead to waves in the printed parts.

You can probably see these waves in your own models. E.g. look at models which have horizontal holes in vertical planar surfaces - search for waves around the holes. Or make an experiment if you have a delta printer which is otherwise solid (i.e. it does not have any unwanted play in the platform, diagonal rod joints, or carriages) and you want to see a consequence of springy belts and smooth-rods/towers:

• print a test cube (you need the sharp corner a cube has)
• it is probably good to slice it using spiralize outer contour (you need to print only the outer shell for this experiment)
• set print speed to at least to 120 mm/s
• set tower stepper current as high as the steppers and the drivers allow so that you can set high accelerations
• set high acceleration to at least about 7000 mm/s²
• notice how there are waves after the corners of the cube; the waves will be biggest just after a corner and then they will be disappearing as the head moves nearer to the next corner
• play with the acceleration, the bigger it is the bigger the waves, the smaller it is the smaller the waves
• if you would have two printers which are the same and the only difference is the printer height, therefore also belt length then you would see that the taller printer has bigger waves at the same speed/acceleration settings

I considered these issues in the design of my 1500mm vertical riser enclosed delta. I chose not to use belts and instead went with stainless steel syncromesh cables and matching pitch 15 tooth pulleys. The cables are smaller in diameter, lighter, and said to hold tension presets well. My design recessed the cables into the riser which allows placement of the carriages much closer to the vertical face and eliminates the belt/effector interference problem of most delta designs. I'm waiting on quotes for the enclosure wall panels and will begin documenting construction as soon as the cut panels arrive.

I would say the main advantage of CoreXY compared to older cartesians is that the forces on carriage are balanced. That leads to smaller wear on smooth rods / linear bearings. And statistically smaller backslash problems especially if the rods/bearings are already worn out. As for as the speed, well CoreXY and H-bot do not heed to haul one stepper so that will lead to higher acceleration in one axe. Though I do not know how useful it is if your acceleration in X can be e.g. two times as big as in Y.

Core XY provides for both XY steppers to be fixed where neither axis carries a stepper in the motion...right? Therefore the acceleration in both would be the same, would they not?

Quote
simspeed
I considered these issues in the design of my 1500mm vertical riser enclosed delta. I chose not to use belts and instead went with stainless steel syncromesh cables and matching pitch 15 tooth pulleys.

I'm playing with a design which would use steel cables too. In my case, I would like to achieve at least 7g accelerations with about 0.05 mm dynamic precision while being able to go without bowden. Which is rather hard since the extruder stepper on platform makes the moving part much heavier.

Good luck with yours.

Quote
simspeed
Core XY provides for both XY steppers to be fixed where neither axis carries a stepper in the motion...right?

Yes. As well as H-bot.

Quote
simspeed
Therefore the acceleration in both would be the same, would they not?

Well, you need to realize that in one direction you need to move also the two horizontal smooth rods. So that means the acceleration in one direction may be bigger (the direction in which you do not move the smooth rods). Not sure whether any firmware can take this into account though. Because it actually translates to one diagonal acceleration being bigger. I doubt CoreXY/H-bot firmwares take this into account.

Quote
hercek
I'm playing with a design which would use steel cables too. In my case, I would like to achieve at least 7g accelerations with about 0.05 mm dynamic precision while being able to go without bowden. Which is rather hard since the extruder stepper on platform makes the moving part much heavier.

Good luck with yours.

Thanks Hereck...I'm going to try both extruder approaches. For the Bowden tubes I'm using Bondtech extruders mounted mid height within the enclosure wall to shorten the tube lengths. For the direct drive I'm using the Berrystruder design with the geared Nema 11 motors to save weight. I'm also using a constant force spring reel to balance the weight of the drive components hoping to minimize the load on the axis motors. We'll see which works best for this design.

Bondtech looks nice. I did not try it yet.

I also considered some kind of flying extruder but I'm afraid how it would perform when there are long quick non-printing moves. It certainly can eliminate high frequency / low amplitude vibrations from small infills. Well, I guess that is just the matter of making the bowden still long enough so that the long move is not jerking the extruder stepper too
It looks to me that it is just easier to go Nema23 and stronger cables and smooth rods and the extruder stiff on the platform. I mean compared to the flying extruder.

If you balance the total weight with a counterweight or CF reel, I don't see the need to "fly" the extruder except for clearance about the effector itself. Werner Berry offers a dual extruder effector using his Nema 11 gear motor extruders. He uses a lead counterweight for his delta and hides the weight inside a sheetmetal riser cover. My enclosure design doesn't provide for concealment so that's why I'm planning to use a CF reel. I bought the two Bondtech extruders for a Core XY machine I designed but have put on hold until after the Hex Delta build. I'll give them a try here first. They're heavy but super strong for long Bowden tube designs.

I should try some kind of "flying" extruder too. It would allow for higher accelerations.
And I'm leaning to the idea that whatever we can do to shorten the bowden ought to be tried. Especially if it does not slow down the printer a lot.

Quote
dc42
2. Choose the rod length to put the rods at about 60 degrees to the horizontal when the effector is centred, and not less then 20 degrees to the horizontal when the nozzle is at the edge of the print area.

Thanks for response. i think i'll go with this 60 deg concept. Also, can i use lead screw of 1.5 mm pitch with nema 17 for up-down movement instead of belt/cable, coz in case of power failure the end effector will be held in a position. bt does this affect speed much?

Quote
simspeed
Delta Calculator

Try this. I used it to design my delta of similar height.

This site is not working??

I just tried it....the link works fine for me. Sorry...

Here is the specs image for my DeltaHex configuration

Edited 1 time(s). Last edit at 10/22/2015 12:02AM by simspeed.

Ya i tried on other laptop, it worked. Thanks.

Can anyone tell me about using lead screw of 1.5mm pitch for up-down movement instead of belt, does it affect speed much??

Quote
harsh
Can anyone tell me about using lead screw of 1.5mm pitch for up-down movement instead of belt, does it affect speed much??

I haven't used lead screws, however I have some words of caution:

1. Lead screws are commonly used on the Z axes of Cartesian, CoreXY and HBot printers. The Z axis moves very little compared to the X and Y axes, so they get relatively little wear. In contrast, the carriages of a delta move all the time during a print. So you will need to keep them well lubricated. My worry would be that they will wear out quickly anyway, but I don't have enough experience to know whether this fear is justified.

2. A typical delta using x16 microstepping has between 80 and 200 microsteps per mm. Yours would have 133.33 full steps/mm. So you definitely don't want to use x16 microstepping, because the electronics would not be able to send step pulses fast enough. Unless you print very slowly, you would need to use either full stepping or half stepping (266.67 half steps/mm). The motors are likely to be very noisy.

3. The limiting speed factor is likely to be the inertia of the motor rotors. So you will need to run the motors at close to their rated current to get a good torque/inertia ratio. Choose your motors carefully so that you can achieve 90% of rated current using whatever drivers you plan to use.

If your goal is to prevent the effector dropping in the event of power failure, I think there are better ways of achieving this. If the effector and carriages are light enough, the motor detent torque is enough to hold position with no power anyway. Also, just holding the effector may not save the print, because the nozzle will probably weld itself to the print. My suggestion for handling power failure is to use a small UPS or some supercapacitors to supply power to the electronics and stepper motors (but perhaps not the hot end, and certainly not the bed heater) for a few seconds. Use a power failure signal to trigger the firmware pause macro, and have that macro retract some filament and move the head safely to a parking position, outside the normal print area. The firmware can also save the current file position and shut down, allowing you to resume the print from where it left off.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Yes, a lot, and also the accelerations will be lower since the lead screw has some moment of inertia and it takes more time/energy to spin it up/down. You need to check the torque charts for your steppers.

E.g. I have a chart here for a 1.8° Nema17 stepper: At 24V it can do about 1300 steps/second and still have some usable torque (the speed could be raised to 3000 steps/second if you would use 48V). That means the maximum speed at 24V for a carriage would be only 1300/200 *1.5 = 9.75 mm/second. That is absolutely terrible speed even for a cartesian printer, not even mentioning a delta. You want to get at least to 100 mm/s, preferably 300 mm/s.

The lead screw would not have the springy problem of a belt but they are extremely slow. You can find screws with much bigger pitch. But it would be still slow.

Well there are multi start lead screws so speed is not the issue (if you pay up). But I think it will be hard to find one that do 15-30 mm/turn and is light enough to get descent accelerations. Moment of inertia is the name of the culprit.

Just had a wild idea about the lead screw adaption.
How about fixing the leadscrew and instead move the steppers up and down together with the carrier?

No more belt stretching
No massive amounts of rotating mass to accelerate
The stepper would turn the brass lead nut with a closed belt and a decent pulley ratio to re-enable 1/16 stepping.

Like I said, it's a wild idea
-Olaf

It's actually fairly common on larger CNC routers. It reduces the leadscrew whipping at higher speeds.

Another interesting way to drive the carriages is with a fixed belt, similar to the way tank treads work. Here's a short thread that explains it: [www.cnczone.com]

If moment of inertia of the screw results in slower accelerations than inertia (weight) of the stepper then it makes sense to keep screw fixed and move the stepper with the carriage.

But the stepper itself is rather heavy.

I did not investigate this much but I would say that for a delta the order would be something like (from best to worse):

• GT2 belt with steel core (these are rare)
• T2.5 belt with steel core (get high quality pulleys which do not have much play when matching the belt)
• GT2 belt with glass core (most GT2 belts sold have glass core)

It is good to prefer GT2 belt to T2.5 belt if they have the same core. The reason is that GT2 belts match more precisely to pulleys. The problem is that most GT2 belts use glass core and therefore are too springy.

It is good to prefer belts since it is very easy to work with them. If one really cares about dynamic precision (i.e. does not want to have "waves" in the high speed prints) then the most easy is to use a simple steel cable. They are very stiff and very easy to get in any store selling bicycle parts. This of course means drums. And drums introduce static error based on the position of the carriage (because the location where the cable is on the drums changes with the carriage position). So it is good for small deltas. Maybe for bigger ones it would make sense to use a steel cable too - with a drum which has a big diameter and a stepper with transmission. One would get backslash from the transmission but maybe it is not that bad. I do not have experience with them.

Young's modulus of common cores (you want as high value here as possible):

• steel - 200 GPa
• spectra - 121 ±11 GPa
• kevlar - 112 GPa
• glass - 70 ±20 GPa

I use GT2 belts that are nearly 2m long on my 1m high delta, and I haven't observed any problems that I can put down to belt springiness. So I am not convinced that you need to avoid using belts. I see that you can get GT2 belt in 9mm width. Why not use that, assuming you can find compatible pulleys? The extra 50% width would reduce the springiness by 1/3.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
dc42
I use GT2 belts that are nearly 2m long on my 1m high delta, and I haven't observed any problems that I can put down to belt springiness. So I am not convinced that you need to avoid using belts. I see that you can get GT2 belt in 9mm width. Why not use that, assuming you can find compatible pulleys? The extra 50% width would reduce the springiness by 1/3.

Not that hard to find GT2 belts with steelcore, not that expensive either.
Plenty of sellers with different widths/timings on Aliexpress.

Actually just ordered 10m PU GT2 with steelcore ($22, free shipping) just to see if there's any noticeable difference at the speeds I am printing, up to 80mm/s.

Quote
dc42
I use GT2 belts that are nearly 2m long on my 1m high delta, and I haven't observed any problems that I can put down to belt springiness. So I am not convinced that you need to avoid using belts. I see that you can get GT2 belt in 9mm width. Why not use that, assuming you can find compatible pulleys? The extra 50% width would reduce the springiness by 1/3.

I actually recommend belts even for delta. They are easy to use, comfortable and fool-proof. I'm just saying one should be picky what kind of belts especially when quick printing on a tall delta is anticipated. And of course, using wider belts is good. My speculations about using steel cable are not for the general user base.

I typically print at 120 mm/s, 7400 mm/s², 20 mm/s jerk. I use 6 mm wide T2.5 steel core belts which are 155 cm long. I can see waves in the surface sometimes. The most easy place to see them is around horizontal holes in vertical surfaces. I believe the waves are caused either by belt stretch or smooth rod bending. I doubt it is caused by the oscillations of the stepper rotor. I believe these would be dampened by stretchy belts. There was also a post on google groups about a user who had a very tall printer and used T2.5 steel core belts. He bought GT2 glass core belts based on some web recommendations and then returned back to T2.5 because GT2 was too stretchy. He was complaining quite a bit about the belts.

I did a theoretical computation what is the maximum belt stretch of a GT2 glass core belt. The elongation for 1m long GT2 glass core belt and 57N force was 3.7 mm. 57N is the maximum force my 3dPrinter Nema17 stepper will do with pulley which has about 15 mm diameter. Computed based on the specification of the belt producer. Of course when you are getting to the limit of the stepper holding torque and use e.g. 1/16 micro-stepping then even the stepper position can be wrong by 8 micro-steps.
Anyway, the point is that 3.7 mm is too much. Of course nobody runs steppers at their limit. But there are people who would run them at about 1/10 of their limits. And at that accelerations you get 0.37 mm belt elongation ... definitely easily visible in your the printed parts.

I guess most people do not notice this since they ran their deltas too gently. It is a delta rock it up a bit. It is supposed to be quick.