Welcome to the SeeMeCNC Community

Hi,

I've been working on making a big delta printer recently, but that's another thread on its own. This one is just about the rollers that will go on it.
Pictures first!
OK now the write up.

My big delta printer uses 15 series T slots, which are 1.50x3.00 inches. They have different slot sizes so I can't buy rollers from open source hardware stores. These are the rollers I made to ride on the linear carriage of the printer.

First, the basics. Each roller is made of an appropriate engineering polymer. Metal rollers would require tighter tolerances and smoother surfaces to work right, but they would still weigh more and would wear the anodized aluminum T slot without modifications. The rollers are 1.500 inches long like the T slot. There are two 13mm bore holes for 624 ball bearings on each side of the roller.

An M4 screw is put through the bearings, then nuts or spacers clamp the inner races together. This part is critical to the rigidity of the linear carriages. Without preload, the bearings could slide +-0.05mm axially. The preload makes everything stiff and precise to a few microns.

A hole of somewhat arbitrary size is bored through the center of the rollers. In this case it is about 0.26 inches in diameter, but it is not precise (nor does it need to be). The 13mm bearing holes must be as concentric as possible to the external roller surface, otherwise the linear carriage will be loose in some places on the T slot and tight in other places.

I tried to avoid this from the beginning but I don't have the measuring tools or machines to solve it. When I bored the second bearing hole on each roller, I had to re-chuck the whole part, which lead to about 0.002 inches of concentricity error. The carriage will still roll down the T slot smoothly, but it will accelerate and decelerate in a sine pattern in phase with the high and low points of the concentricity error. It still stays preloaded and accurate, but I know that it may be wobbling up to 0.002 side to side during the motion. Less than perfect.


As far as materials go, I picked four types of plastic that seemed right for the application. Two were Delrin, which is a brand name for Acetal Homopolymer (as opposed to Copolymer, which has reduced mechanical properties). One was a high grade PET material, and the last was Molybdenum Disulfide filled Nylon 6,6 (MDS Nylon, TECAMID).

All of them machined decently on the lathe, but they were all very stringy. I picked these four plastics primarily for wear resistance, operating temperature, stiffness, and creep resistance. Creep means that the plastic will bend over time if a pressure is constantly applied to it.

So far I have only tested the MDS nylon rollers on the linear carriages. Over the course of a day or so, they developed dimples from creep where the tapered surface rolls on the T slot rail. So MDS nylon is ruled out as a possible roller material. Black and white Delrin should perform the same; I'm unsure about the PET. Black Delrin is what SeeMeCNC uses on their Cheapskates.

I also considered using PEI rod for this, but it is significantly more expensive and I'm unsure about the wear properties. I'll give more updates on this when I make more progress this coming weekend. I bought more black Delrin rods because I'm going to have to make more of these rollers. I might also change the roller length and the tapered part that rolls on the T slot rail. I'll probably shorten the rollers and put fillets on the rolling surface that match the profile of the T slot extrusions. This will hopefully get more line contact area to prevent creep, and switching from MDS Nylon to Delrin should help too.

Why not try teflon?

PTFE creeps really quickly. The surfaces that contact the T slot would flatten out after a few minutes or hours.

My first recommendation would be some THK, NSK, or even Hiwin rails. They will be tight, and remain tight while holding tighter tolerances than you can imagine and never loosen up. Its about $75 per rail on ebay. Not that big of an expense in my opinion.

To address material, Ultem(PEI) is an awesome material for all things precision plastic, its stronger, creeps less, wears better, and handles higher temps. I use it regularly in the medical field. It is expensive.

To get the wheels concentric, consider a 4 jaw chuck, indicate on the bore that you previously cut - bore the thru hole accurately as well as your bearing counterbore so you have something to indicate on.

I thought about getting linear rails for the stiffness and straightness, but they would be quite expensive for the entire printer. Each tower would need 1200mm of rail, plus at least one carriage. The carriages are heavy from what I've heard, and I'm trying to keep the weight to a minimum to allow for faster accelerations. I may end up using them, but the preliminary tests from the first carriage I made with rollers showed much better stiffness than I expected, even when the load is cantilevered 15 cm.

I'll check out PEI as a roller material again based on what you said. The creep and temperature properties look good, you say the wear is good, and parts I've made before out of PEI turn out really nice. It would be around 14 dollars per foot if I changed my design to use 0.75 inch round bar, otherwise $21 per foot for 1 inch bar.

Originally I planned on using it because I needed a plastic that could operate at 150C . I'll consider it for other parts too, but if Delrin works for the rollers I'll probably stick with it.

I do need to get a 4 jaw chuck; I don't know if indicating each part would be necessary as long as two of the jaws were never moved between parts. I would need a more accurate indicator too.

My lathe is weak and lacks precision and capabilities, I'll probably stick with my inaccurate method for the next rollers I make, although I will definitely make a self centering drill bit to keep the hole on center more if possible. I'll probably end up making the rollers on a full sized CNC lathe with a subspindle if I get to production.
This picture is the first iteration of the carriage. My newer ones are 0.25 thick aluminum on the inside and a printed plate on the outside. For low loads and high accelerations, I think lightweight roller carriages may be better than the heavier recirculating ball carriages. If I were making a subtractive machine, I would definitely go with the precision rails though.

Mineature linear rail carriages are very light. Few ounces each for 12mm wide rails(wide=wider than 12mm)

I'll have to do a comparison test sooner or later between the guide rails and the roller thing I'm using now. I changed this design to be lighter and use less material per roller. Total weight of my carriage is a little bit under 300g. I also changed the roller contact points to be fillets rather than chamfers.

If the roller fillets match the T slot fillets exactly, they will have large line contacts to distribute the pressure. The fillets are slightly larger so they contact at a known angle and still distribute some of the pressure. I'll have to test it out more before I can be certain. I guess what I want out of this is a long lasting, precise roller system that rolls super smooth with no degradation over time or with use. So far the cheapskates from my Rostock Max V1 have done that but they aren't quite so precise since they come in two pieces with an OD that isn't even turned. Plus a single 608 bearing is heavy and has no preload.

This is an excellent thread that I enjoy reading.
I like to read how you created your large capacity delta and I am looking forward to see
the completed delta. Thanks for starting this thread.

I made the new short rollers, about 12 total, since there are 3 rails and each needs 4 rollers. I actually designed the linear carriages for 3 or 4 rollers but I think 4 would be more rigid.
Anyway, 4 of the new short rollers are made out of PEI. They all look rougher in the pictures than they do in real life.

I tested a carriage on the T slot with the rollers and it seemed very rigid to me. I put cantilever forces on the carriage and it didn't bind or deflect any visual amount. So far so good.

I had to make a new drill to bore the bearing holes in these rollers. The exact depth of the bearing hole is critical, so I turned a drill from 1144 steel with a big step at exactly 5.08mm. When pressed against the face of the roller, the drill goes in and stops at exactly 5.08mm. The bearing holes come out to 12.89mm for a 13mm bearing, so they press fit tightly and don't come out without some serious force!

I think I solved the concentricity problem by through-hole drilling with a side cutting endmill, then drilling with the self centering 12.89mm drill. This ensures an on center hole even though the endmill isn't on center. The 0.2500" endmill makes a hole about 0.26" in diameter.

The 12.89mm drill I made wobbles with the part and stays on center because it can't cut sideways like an endmill. Since it can't cut on the sides, the sides rub against the plastic. The PEI was difficult to drill because the rubbing would generate a lot of heat, which caused galling on the inside of the bearing hole. The black delrin machined nicely because it has a much lower coefficient of friction and heat deflection temperature.