My ongoing 'biggee' Kossel development

Here is a pic of my CAD so far. It's 2020 struts and I am scaling for a 300mm bed. I probably should have used a 350mm bed assembly to get a full 300mm build space, oh well...

Some things you might notice:
-Frame is somewhat larger than the bed; I wanted to keep things inside as much as possible for potential enclosure later. It forces an increase in the rod length but also decreases the maximum angles of the rods at extreme positions which should help with various issues near the edges of the bed. I am also using a heavier-gauge rod (from Tridprinting).

-My current Kossel Mini rod spacing is 40mm. With almost double-length rods for the bigger design, that leaves a large aspect ratio of rod spacing to length, which magnifies flexing and tolerances, impacting on position and especially tilting. What you see is my current attempt to scale the carriages & effector to 80mm rod spacing.

-I'm trying to reuse parts from the Mini, which includes the ball-bearing linear rails that have a 20x20 screw pattern on the stage. For such large carriages with the overhang, it might be wise at some point to move to extrusion-roller carriages to give me a wider footprint. They seem pretty stiff though so I'm sticking with them for now.

-I'm enclosing the underside of the bed to keep the motors from being exposed to the extra heat if the the chamber is enclosed later, to block small parts from dropping inside, and because it looks nice. Right now I have white Melamine-covered sheet of MDF for the cover; since MDF is not very stable with humidity there are metal struts that actually support the bed assembly and the only functionality of the MDF will be to hold the sliding bed clamps. The flammability of MDF is another reason I might give this up (although I don't have an alternative at the moment).

-Stuff under the bed may require cooling fans now that the steppers are boxed up (there is also a bed pan on the bottom of the frame to hold stuff, also MDF).

My priority here is stiffness over speed because I worry the accuracy will suffer and that is more important to me than speed.

Oh, it has the RobotDigg metal vertex corners.

So..am I doomed with my current path? I am still bucking the trend and sticking with 3mm filament for now so stiffness is the important thing (I have some investment in 3mm I am not ready to let go of yet).

Two comments:

1. I think the MDF is a bad idea. It will flex and warp with the heat from the bed, so the bed will not stay level. It will also prevent the stepper motors cooling by convection, causing them and the space under the MDF to get hot. Why not support the bed from the metal frame and put insulation under it, like everyone else does? Or at least support the bed using metal standoff pillars attached to the metal frame and passing through holes in the MDF.

2. I really think you should get this machine working well with 1.75mm filament before you try it with 3mm.

---

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].

I recently completed a large delta; here are a few things I learned:

(1) 2020 is likely to be too wobbly for the vertical towers. I'm using 2040 and it is borderline.
(2) After doing a LOT of analysis of bed size, printer size, rod length, etc. (including using the quite handy kinematics visualization scripts here) I came up with a few dimensional rules of thumb which I am sure will ruffle everyone's feathers but which nevertheless are working well for me:

given a print radius DELTA_MAX_RADIUS, in my case 150mm, I got what I felt produced the best tradeoff between size and print quality at the outside of the radius with:

DELTA_DIAGONAL_ROD = 2.5 * DELTA_MAX_RADIUS = 375
PRINTER_RADIUS = (4/3) * DELTA_MAX_RADIUS = 200

(3) I bought motors that had WAY more torque than necessary and drive them at 80% of the rated current. They don't get hot.

(4) Nevertheless, I made the 6mm craft-ply plate you see in the picture to mount the bed to specifically so that the electronics and the motors would cool more effectively. The bed itself is supported on springs so that I can level it and so that a head crash is a little less painful. The Onyx PCB is backed with a layer of cork (5/32" I think) which is spray-glued to a piece of 1/8" craft ply, and that insulated the bed quite well, makes it heat more efficiently, and keeps the heat away from the motors and electronics. If I had to do it again, I might make the big plate even stiffer, though. 6mm ply works, but I don't think MDF is such a great idea,

(5) Long bowden tubes can be a pain in the tail. A 3mm Bowden is something most people will tell you to shy away from, and if it makes anything at all worse to deal with you are going to be quite frustrated

Quote
wfcook
I recently completed a large delta; here are a few things I learned:

(1) 2020 is likely to be too wobbly for the vertical towers. I'm using 2040 and it is borderline.

Interesting, my 2020 1m high Kossel is not wobbly at all. I think the metal corners are a big help. Also I used 2060 for two of the lower horizontals, which has a similar effect to adding corner braces.

Quote
wfcook
(2) After doing a LOT of analysis of bed size, printer size, rod length, etc. (including using the quite handy kinematics visualization scripts here) I came up with a few dimensional rules of thumb which I am sure will ruffle everyone's feathers but which nevertheless are working well for me:

given a print radius DELTA_MAX_RADIUS, in my case 150mm, I got what I felt produced the best tradeoff between size and print quality at the outside of the radius with:

DELTA_DIAGONAL_ROD = 2.5 * DELTA_MAX_RADIUS = 375
PRINTER_RADIUS = (4/3) * DELTA_MAX_RADIUS = 200

Looking at your photo, I think you could use a much larger print bed. XY resolution improves as you get farther from the centre. Compare my photo here:



The bed radius is 165mm, of which the printable radius near the towers is 150mm. I provided the extra 15mm for bed supports, bed clips, and to give a heated margin around the print to discourage warping.

Quote
wfcook
(3) I bought motors that had WAY more torque than necessary and drive them at 80% of the rated current. They don't get hot.

I agree, 80% of rated current is a sensible maximum. I find 50% of rated current is enough on my printer, except when I am using insanely high speeds or accelerations to test firmware changes.

---

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].

Hello,

I don't have much delta design experience but the following things strikes me:

The carriages are really wide. I understand what you say about aspect ratio (rod spacing to rod length) but will the tolerance in your rails not create problems with slack in the setup? As tried to illustrate below.



For that reason it seems attractive to have the rods closer together unless your rails are really good. The introduced error illustrated above will be dependent of direction of movement whereas problems with rod length will be more static and can be solved by fine adjustment.

Second - Where will you mount the z - probe?


Finally, small SW tip: In the upper tabs (where you have file, edit, view and more) you can chose "view" and then "screen capture" and then "image capture". That picture you can then paste into paint or what ever like a normal screenshot. The picture becomes cleaner that way.

Quote
LarsK
Hello,

I don't have much delta design experience but the following things strikes me:

The carriages are really wide. I understand what you say about aspect ratio (rod spacing to rod length) but will the tolerance in your rails not create problems with slack in the setup? As tried to illustrate below.

For that reason it seems attractive to have the rods closer together unless your rails are really good.

I don't think the wide carriages will make things worse. If a carriage rotates due to slack as you have illustrated, then the effector will rotate by the same angle as the carriage, no matter what the rod spacing is. OTOH, a wider rod spacing will improve rigidity of the spider assembly, and will also make the system less sensitive to tiny differences in the lengths of the diagonal rods in a pair.

It occurs to me that if 2040 extrusions were used with the long edge in the tangential direction, then wide carriages with rollers running on the outside of the extrusions should provide very stable system.

---

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

Quote
LarsK
Hello,

I don't have much delta design experience but the following things strikes me:

The carriages are really wide. I understand what you say about aspect ratio (rod spacing to rod length) but will the tolerance in your rails not create problems with slack in the setup? As tried to illustrate below.



For that reason it seems attractive to have the rods closer together unless your rails are really good.

I don't think the wide carriages will make things worse. If a carriage rotates due to slack as you have illustrated, then the effector will rotate by the same angle as the carriage, no matter what the rod spacing is. OTOH, a wider rod spacing will improve rigidity of the spider assembly, and will also make the system less sensitive to tiny differences in the lengths of the diagonal rods in a pair.

It occurs to me that if 2040 extrusions were used with the long edge in the tangential direction, then wide carriages with rollers running on the outside of the extrusions should provide very stable system.

You are correct. My first point is invalid as you say. I still notice that I have not seen many other designs so wide but that should not be a reason not to do it of cause.

I agree that using the 2040 like you say seems like a good idea. It will give a lot of stability.

Quote
dc42
Looking at your photo, I think you could use a much larger print bed. XY resolution improves as you get farther from the centre. Compare my photo here:

Resolution on a delta, particularly on the edges off the bed, is difficult to describe as getting "better" or "worse", since as you know one rod may bee gaining resolution while the others are losing it. I tend to think that the system works better when things are more evenly balanced, such as they are in the center of the bed, where this printer with 20T gears, 0.9 steppers, and 1/32 microstepping already has some ridiculous resolution.

That said, I agree that it could handle a larger bed if I get the gumption to make it happen. The curves of tower resolution begin to get more uneven rapidly beyond the points I described, which is why I chose them.

It's looks like it is resting on the MDF but the bed has a cork lining under it and all is resting on metal struts underneath. (the sliding clips are on the MDF but with a solid bed it shouldn't matter that much).
@DC42:
-I don't see any kind of cooling on the duet stepper drivers, which seems counter to the usual heats sinks on RAMPS (although that is probably why there is no cover?)
-It looked like you mounted your bed using metal standoffs (in your blog) attached to the heat spreader but that seems like that would cause cold spots where the standoffs attach. (althought ceramic standoffs might work well)
-I agree about up-sizing the bed radius for the reasons mentioned.
-XY resolution does increase at the edges but isn't there a big tradeoff with speed as one of the rods goes nearly horizontal? That is one reason I scaled the frame (and rods) to completely surround the bed.
-I really wanted a more polished solution than the popular clips to hold the bed down, although I admit their simplicity is hard to beat. It's just so temporary looking...
-Do you ever plan to enclose your rig for heating?
-The SeeMeCNC deltas have an enclosed bottom, what have they done to mitigate trapped heat down there? I was thinking of using forced air venting.

Quote
shadowphile
@DC42:
-I don't see any kind of cooling on the duet stepper drivers, which seems counter to the usual heats sinks on RAMPS (although that is probably why there is no cover?)

The stepper driver chips, like all SMD parts, are designed to be cooled via the printed circuit board. The Duet design does this well (as do most electronics boards with on-board drivers). The stepstick drivers used on RAMPS do it badly, because the PCBs that the chips are mounted on are far too small.

Quote
shadowphile
-It looked like you mounted your bed using metal standoffs (in your blog) attached to the heat spreader but that seems like that would cause cold spots where the standoffs attach. (althought ceramic standoffs might work well)

The M3 standoffs are 10mm long and quite small in diameter, and the aluminium plate is 5mm thick. So I think the spots around them will be only slightly cooler than the rest of the plate, and in any case they are in the bed margin outside the printable area. Ceramic spacers are a nice idea though. PTFE spacers are another possibility.

Quote
shadowphile
-I agree about up-sizing the bed radius for the reasons mentioned.
-XY resolution does increase at the edges but isn't there a big tradeoff with speed as one of the rods goes nearly horizontal? That is one reason I scaled the frame (and rods) to completely surround the bed.

Only if your stepper motors and/or electronics are not up to the job. The rods never go more than about 20 degrees off the horizontal anyway.

Quote
shadowphile
-I really wanted a more polished solution than the popular clips to hold the bed down, although I admit their simplicity is hard to beat. It's just so temporary looking...

Picture frame clips look much neater - I use them on my Ormerod because they take up less space (the Ormerod bed was designed without a margin).

Quote
shadowphile
-Do you ever plan to enclose your rig for heating?

Yes! One option I am considering is to fix a plywood or acrylic panel about 80mm wide to the back of each tower. Then fill in the spaces between the edges of those panels with much wider acrylic panels, hinged on one side and with magnetic catches on the other. Then a lid on top.

Quote
shadowphile
-The SeeMeCNC deltas have an enclosed bottom, what have they done to mitigate trapped heat down there? I was thinking of using forced air venting.

If whatever is under the bed (PSU, electronics etc.) and the stepper motors can't cool by convection, then a fan is obviously needed.

---

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].