A Big, Rigid Delta (Picture heavy)

[Generic Default]

Words are boring, so I'll keep them to a minimum in this thread.
I'm building a large Delta printer. Print space is ~17" (430mm) diameter, ~22 inches (550mm) tall. Right now my design is using fast travel precision leadscrews for linear motion, but the frame is compatible with belt/pulley drive as well. Almost all of the parts are aluminum or steel, made adjustable with tight tolerances. The printer will be fully enclosed with temperature control up to at least 150C, probably higher. I can't yet disclose some of the details of the enclosure. For now I'm using my Tri hotend for triple extrusion on a non-magnetic ball joint effector.

I'll post more pictures as I make progress, and I'm actually further ahead on design and prototyping than I'm showing on these forums.

[Generic Default]

These are the balls I'm considering. They're 0.375 diameter with 8-32 tapped holes. They come in plain brass and nickel plated brass, shown next to a 3/8 chrome steel ball bearing for comparison.

Unfortunately the threaded balls are not perfect since they were made on an old automatic screw machine. They are round around their central axis, but distorted around the other axis. Sort of like how planets are slightly elliptical because of spinning. The plain brass ones are about 0.0015 to 0.002 shorter than they are wide, and the nickel plated brass ones have backlash flat marks that look to be caused by about 0.002 of backlash in the lathe that made them.

0.002 will show up as imperfections in printed parts because the balls mate to a cylindrical hole. The ball bearings don't have this problem because they are spherical to about a micron. The surface finish is perfect with the ball bearings, but I think it is still workable with the turned balls. The 0.500 CF arms have aluminum cups on them that rub on the ball surfaces, and I don't know how well aluminum on brass or aluminum on chrome steel will work. It feels rigid and it has low friction, which I want, but I want to know how long the arm cups will last with sphere-on-circle wear. I think the surface finish of the ball will matter a lot.

This is what the effector plate looks like mostly assembled. All of the parts are milled flat because 1/8" extruded aluminum bars are bowed. I wanted to make sure this part turned out perfect because the hotend and effector set up is critical to print performance. This thing is as rigid as it gets, and the hotend is perfectly level to the effector.

I'll get pictures of the frame and linear motion parts this coming weekend when I have time. I'm working on the enclosure now.

[Tincho85]

That's looking awesome!
Great job.

[Generic Default]

Here is my plan for the heated bed. It's a sheet of PEI on top of a 5/16"x16.75 (8mm thick) Mic-6 plate. I have some wet-and-stick E glass fiber sheets to put between the coil and the conductive aluminum plate, then another glass fiber sheet under that to insulate the rest of the printer from the hot bed. The whole stack will be held up by silicon standoffs so that I can adjust the bed on 4 points with screws for manual leveling. The frame of the printer needs to stay room temperature even though the bed and enclosure will be able to get up to around 200C. The mountting bar (notice that there are two T's because it's not a mistake or anything) will also probably be covered with thicker E glass sheets to avoid heat transfer to the frame.

I'm going to start on the heated bed as soon as I get some Nichrome wire. I'll have to do some basic electrical calculations to figure out what size of wire to buy. I'm not sure if I should go with a 120v solid state relay or if I should just use the 24v 360 watt PSU that also powers the smoothieboard. I'm leaning toward the SSR because it can have a much higher power output without possibly shorting the smoothie.

Does this heated bed idea look viable? Has anyone here made their own heated bed?

[Eaglezsoar]

hot plate stack render annotated.jpg

Here is my plan for the heated bed. It's a sheet of PEI on top of a 5/16"x16.75 (8mm thick) Mic-6 plate. I have some wet-and-stick E glass fiber sheets to put between the coil and the conductive aluminum plate, then another glass fiber sheet under that to insulate the rest of the printer from the hot bed. The whole stack will be held up by silicon standoffs so that I can adjust the bed on 4 points with screws for manual leveling. The frame of the printer needs to stay room temperature even though the bed and enclosure will be able to get up to around 200C. The mountting bar (notice that there are two T's because it's not a mistake or anything) will also probably be covered with thicker E glass sheets to avoid heat transfer to the frame.

I'm going to start on the heated bed as soon as I get some Nichrome wire. I'll have to do some basic electrical calculations to figure out what size of wire to buy. I'm not sure if I should go with a 120v solid state relay or if I should just use the 24v 360 watt PSU that also powers the smoothieboard. I'm leaning toward the SSR because it can have a much higher power output without possibly shorting the smoothie.

Does this heated bed idea look viable? Has anyone here made their own heated bed?

A nice nichrome wire calculator - http://www.jacobs-online.biz/nichrome/NichromeCalc.html

[Generic Default]

That's a nice calculator! I found a table in the video that explains it, shown below;

[img]http://hotwirefoamcutterinfo.com/_NiChr ... perage.jpg[/img]
[img]http://i.stack.imgur.com/jkJvu.png[/img]

If I use a solid state relay, I could use the smoothieboard output to the heated bed to control it. The Smoothie output would trigger the SSR to close the 120 VAC circuit, which would supply the heating coil with power.

So at 120 volts, I would need 3-7 amps to get in the power range I want. I don't want the wire to get so hot that it burns the glass fiber or starts a fire. The power supply calculations would be as follows;

120 volts from wall outlet plug
5 Amps of current
600 watts from 120 volts * 5 amps

120 volts / 5 Amps = 24 Ohms of resistance
22 AWG wire at 5 Amps = ~900 degrees F (Maximum for glass fiber sheet is 1100F)
22 AWG wire has 2.8 Ohms/meter, 10 meters needed

10 meters of doubled Nichrome wire would be 5 spiral passes around the circular alum plate. I think I can fit 5 spiral revolutions in the area of the plate. I'll have to leave enough space between the spiral arms so that there isn't any arcing between them. That would be a disaster at 120 VAC.

I really have no idea how to accurately bend a wire to make a heating coil. Can I just bend 22 AWG Nichrome by hand into a long, straight length of 10m, then fold it at the half way point and spiral the parallel wires? I know how springy wires can get, and I can't have any wires within a few millimeters of each other or they may arc. Once the glass fiber is wetted and stuck in place, it will give keep the wire held in place. I don't know if the 120 VAC will arc between the wire and the aluminum plate because the glass fiber is a porous cloth.


Backup plan is to buy a heat gun or hair dryer and control it with the SSR. After all, this entire thing will be in an enclosure.

[Eaglezsoar]

Another nice article. http://airtripper.com/698/resistance-wi ... -tutorial/

[ZakRabbit]

That's a nice calculator! I found a table in the video that explains it, shown below;

[img]http://hotwirefoamcutterinfo.com/_NiChr ... perage.jpg[/img]
[img]http://i.stack.imgur.com/jkJvu.png[/img]

If I use a solid state relay, I could use the smoothieboard output to the heated bed to control it. The Smoothie output would trigger the SSR to close the 120 VAC circuit, which would supply the heating coil with power.

So at 120 volts, I would need 3-7 amps to get in the power range I want. I don't want the wire to get so hot that it burns the glass fiber or starts a fire. The power supply calculations would be as follows;

120 volts from wall outlet plug
5 Amps of current
600 watts from 120 volts * 5 amps

120 volts / 5 Amps = 24 Ohms of resistance
22 AWG wire at 5 Amps = ~900 degrees F (Maximum for glass fiber sheet is 1100F)
22 AWG wire has 2.8 Ohms/meter, 10 meters needed

10 meters of doubled Nichrome wire would be 5 spiral passes around the circular alum plate. I think I can fit 5 spiral revolutions in the area of the plate. I'll have to leave enough space between the spiral arms so that there isn't any arcing between them. That would be a disaster at 120 VAC.

I really have no idea how to accurately bend a wire to make a heating coil. Can I just bend 22 AWG Nichrome by hand into a long, straight length of 10m, then fold it at the half way point and spiral the parallel wires? I know how springy wires can get, and I can't have any wires within a few millimeters of each other or they may arc. Once the glass fiber is wetted and stuck in place, it will give keep the wire held in place. I don't know if the 120 VAC will arc between the wire and the aluminum plate because the glass fiber is a porous cloth.


Backup plan is to buy a heat gun or hair dryer and control it with the SSR. After all, this entire thing will be in an enclosure.

I "play" with nichrome all the time as a vaper, you can anneal it before shaping it so that it won't be so springy.

[Generic Default]

I'll just ignore the heated bed for now, I'll resume working on it once I have this thing moving and printing.

This is what the printer looks like half-assembled next to my Rostock Max V1.

This is a close up of a tower.

Notice that it uses a NEMA 23 stepper with a damper. It is all bolted together with 0.25" (6.35mm) thick alum plates for structural strength and rigidity.
The motor shaft is coupled to a 3/8th inch fast travel lead screw with 1/2 inch of travel per revolution. This means each full step moves the linear carriage 0.0025".
The lead screw is axially rigid but can pivot in pitch and yaw several degrees to prevent off-axis wobbling. It is held together with shaft collars and two preloaded ball bearings in a spindle, with silicone square O rings as spacers that absorb vibration and allow the radial angling. It helps that the linear carriage is ultra rigid.


Since I went with a lead screw instead of a belt drive, I had to take backlash into account. My goal was absolutely zero backlash since this printer needs to be precise. I reached the goal with the nuts shown below;

I already printed a bunch of these with different variations. Material is trimmer line nylon. I call them "Creep-Preload Nuts" because they are initially tight with high friction, but after several days of being on the lead screw they lose nearly all of the friction but keep their zero backlash properties. I measured with a dial indicator and I saw no backlash whatsoever, although I can only be certain to within 0.0002" because the dial indicator is only that accurate.
The nuts turned out much better than I expected, and I'm going to use them on any CNC machines I build in the future.



I'll try to get a video up next weekend once I get stuff moving.

[Generic Default]

I had to make a jig to ensure that all of the delta arms are exactly the same length.
I don't have any saws so I can't cut my 0.5 inch CF tubes to equal length. The aluminum ball sockets on each end of each CF tube can slide a bit, and have close concentric tolerances with the CF tubes. They are locked in place with set screws.

The jig consists of two printed parts on a 24 inch long T slot extrusion. The black block on the right has a 3/8 socket with a 3/8 chrome steel ball glued into it. The left black block is flat except for a small detent pit to accurately position a 3/8 ball.

The CF tube with the aluminum ends is placed so the right socket mates with the 3/8 ball on the black jig block, then the other end is positioned with a 3/8 ball in the socket into the left side detent. The left side aluminum end slides to the left until it is pressing the 3/8 ball into the detent, then the set screws are locked. All six delta arms are calibrated this way.

So long as the room temperature is kept within a +- 1 degree Celsius range, the arms will all be equivalent in length to within 1 thousandth of an inch over a 500mm span. I can't think of any way to get more accurate than this without metrology measurement machines.

A delta printer will work pretty well so long as the arms are consistent in length to within about 4 thousandths of an inch; any more will tilt the effector near the edges of the build plate. The black blocks on my jig are printed with PETG.

[Heffa]

This looks awesome!
I really like the Tri Hotend effector - any chance you would be willing to share the source for that? It looks to be exactly what I was planning to build for my own large Delta

[stonewater]

thetri hot end is a polystroooder http://www.polystroooder.com ..... they have some pretty awesome engineering inside of them!


Tom C

[Heffa]

Thanks, I know about the Polystrooder and am thinking about using one myself.
I was interesting in the aluminium effector platform for the hotend - it didn't look like the one at Youmagine from Generic Default

[Eaglezsoar]

A great post that I hope continues, great job on the build!!

[Generic Default]

Hey sorry for not responding recently, my computer died and I haven't had access to one for about a week. I won't be able to do much here for another week or so when I get another computer in.

The hotend mount is made from machined aluminum extrusions. I do plan on making them to sell later this year. I'll get a picture of the whole setup pretty soon, it's pretty cool looking.

The Youmagine one is meant to be printable and easy to work with. Most people don't have CNC mills or waterjets at their disposal. I'll try to get the gcode and CAD files for the mount shown on this thread if I can recover the data from my old computer, otherwise I'll have to remake that stuff.

[McSlappy]

If this thing is even near the quality of the Polystroooder I think you'll have the best printer around

[Jrjones]

Care to explain a bit more on the anti-backlash nuts? Is the helix different along the length of the nut or is there no backlash due to the tight fit?

[Generic Default]

They're just plain nuts, the helix is constant and matches the fast travel acme profile. Screws are never perfect, if you take a long enough nut and try to put a threaded rod through it, it will bind. The nuts on this printer are close in tolerances to the screw and they're just long enough so that they take up all of the backlash without adding too much preload from binding.

They run dry, and since it's flexible nylon on burnished steel, there is very low wear and very low friction. It's much stiffer than belts but it's a bit more complicated to make.

[SEBRET]

So to Make sure I'm understanding this. You used the slight variations in the thread as a feature rather than a con by matching it with the more consistent nylon thread? Meaning that at least some part of the Nylon is firmly gripped and therefore backlash free? I know it's a probably a simple concept, but i like to really spell everything out when it comes to mechanical stuff.

[Generic Default]

Yep, the flexing of the material is what allows the backlash to be taken up. This trimmer line nylon has a modulus of about 1 gpa, normal steel nuts have a modulus of about 200. I was unsure if the design would work until I tested it with a dial indicator. With small nema 23 steppers, it can still drive the axis pretty fast with high acceleration. My next test will be on a more rigid lathe setup; I know the backlash is near zero but I want to check the static and dynamic stiffness under bigger loads.

[Jrjones]

Ah, that makes sense. Bonus that you don't need to lubricate nylon either.

[Generic Default]

I finally mounted one of my Tri hotends on the effector. Looks awesome. I wish I had more than 2 minutes a week to work on this thing, though!

At this point I have the machine moving at standard printer speeds of up to 150mm/sec. Next, I have to wire the hotend and test print, then install all of the parts for the enclosed chamber and wire that so I can print in a 100C oven and eliminate warping.

[Eaglezsoar]

That is the best arm - effector combination that I have ever seen, absolutely fantastic work!

[Generic Default]

Bringing this back from the dead because I've been working on it recently. I had a computer death in August 2015 and I lost all of my files and CAD for the printer. I remade them of course, but the data loss combined with moving and being busy with other stuff kept me from working on the printer for a little bit more than a year.

Anyway,


I'm to the point of enclosing the build volume of the printer with heat resistant materials so I can have a heated build chamber up to 200 degrees C. The build plate itself is an 8mm (5/16") x 425mm (16.75") mic-6 aluminum slab. The build surface will either be PEI or high temperature Garolite sheet (for nylon). Right now the 12x12" square sheets don't cover the entire surface so my build area is 12x12x22".


I just got the new PEI sheet in today with the adhesive. It's clear rather than amber, which has me wondering if it's a completely different polymer. It seems more flexible as well. Has anyone here tried both on printers?

Green - high temperature Garolite
Yellow - PEI from 2014
Light Blue - New PEI

[Qdeathstar]

how are you planning to cool the hotend? i'm assuming a fan wouldn't do a lot at an enclosure temp of 150c?