It's time for a metal v-wheel

[ejs]

This is a branch of the discussion found here http://www.shapeoko.com/forum/viewtopic ... 7&start=10

So the idea has come up more than once of reworking the v-wheels away from Delrin towards metal. There are several concerns:

~The available steel wheels have dimensions that vary from the Delrin wheel, which would require reworking the plate
~The steel wheels could potentially wear down the soft-anodized aluminum slide
~Brass alloy could be an excellent candidate, but is costly by comparison

Initially I came at the problem with the intent of duplicating the wheels that come with the kit. The pursuit was given up at, first, when I came up with the idea of a better replacement spacer. It's a definite improvement. However, the light on the horizon is finishing up a real, metal alternative to the supplied part in order to further firm up the entire build. With 16 points of contact, there is an undeniable improvement to be had. After some posts regarding the topic of the merits of the reworked spacers, the idea of metal wheels surfaced again. This time, something workable emerged.

A Fix

Trying to make the original wheel requires a multistep process on the ShapeOko, or some serious undercutting. A metal lathe would likely be the best option. Something else was needed for this to be a 2.5D part.

First there is the matter of the metal. With copper costing what it does, an aluminum that is similar to the composition of the slide seems a good choice.

Next is the matter of design. Instead of going extreme on the undercutting, I realized the whole idea of the sandwiching the wheel between the bearings could, itself, be reworked to make the wheel the bread. If we reuse the bearings they can be put right next to each other. The cream on top of my redesign, is both how and how well it keeps the bearings together.

The Details

By using two differing stock thicknesses, I've put together a solution. The mock ups are below. First, note that there are spurs on the inside of new wheels. These are slightly undersized and slightly inward sloping to create a secure fit from under to oversized bearings. The cut away material allows for slight deformation of the spurs which facilitates the snug fit.

The opening in thicker stock is actually deeper than the bearing itself. This allows one bearing to be completely sunk into that half of the wheel, and the remaining stock that extends above the bearing creates a space for the second bearing to register into. This creates a system that works to keep both bearings extremely aligned and centered with one another.

The shorter half is drilled though the edge and the taller half is drilled and tapped to allow #00 bolts to keep the halves together, clamping the bearings in between. The lip that retains the bearing closest to the plate should require no other changes to the setup. The design offsets the centerline of the "V" by only about a quarter millimeter, well within the allowance of the ShapeOko's plates.

Action

So the stock and the appropriate mill are on there way. Reusing the current bearings will ease cost. My design is very good, but will likely need some light tweaking as projects tend to. I expect to have things nailed down by the end of November. I'm really very excited about how this will play out and expect the results to be excellent. So who else would be interested?

As an FYI, here's the wheel that doesn't fit.
http://store.makerslide.com/index.php?m ... ucts_id=75

For your engineering pleasure:

Components


Components: Isometric


Assembly: Solid bearings, transparent wheels show bearings' hairline meeting one-half millimeter below the gap in the "V"


Assembly: Isometric with solid bearings and transparent wheels showing detail of the spurs


Glamor shot: showing the full assembly and the M5 socket screw head.


Concerns

Two things I have a care about. First, the four bolts holding things together will be #00-90 as mentioned. These are 1.19 mm in diameter for the major dimension. The minor is just under 1 mm. Four of these will be able to produce about 120 pounds of force between the two halves without worry about them deforming. I think that should be plenty, but additional bolts is a consideration. Fasteners with a larger diameter are out since they would interfere with the positive "V" that comprises the rail.

As for the inner lip nearest the plate. It should clear the plate, but a second washer may be needed if the current washer is undersized. This would move the centerline of the "V" out by about 3/4 of a millimeter--a change that should not be a problem for the supplied plates.

Please post your interest and thoughts below. I'm very interested to hear what people think about this solution.

Best,
EJ

[daveczrn]

The only slight flaw I see is captivating the bearings without having some type of hard stop between them. Currently when you tighten the screws the inner races will touch first as usually they are wider than the outer race. That will then push the outer race closer together as the inner race cannot as it would have a hard stop. It might work out of luck having enough play I. The bearings where it won't matter where there would be enough combined play in both bearings to take up the distance of the overhanging inner flanges create. Thus giving a hard atop for both inner race and outer race. If one side of the wheel was machined to not have the gap between the wheels then none of this would be guess work.

Dave

[CptanPanic]

Look great, can't wait to see how it turns out.

[cvoinescu]

I admit I don't understand what @daveczrn said. In the bearings I have, the inner and outer races are the same size, and the rubber shields are slightly recessed, so there's no reason why two bearings can't be squished together and work just fine.

My concerns and worries are:

• If this is milled on a ShapeOko, the parts won't be perfectly round (because the ShapeOko isn't perfectly orthogonal, the X and Y belts aren't tensioned exactly the same, and because the Z axis flexes more in the Y direction than the X).
• The parts won't be smooth enough as milled, and sanding or polishing them by hand introduces further imperfections.
• Even with round parts (e.g. turned, not milled), they will deform slightly, coming a little further apart between the screws; that will make them wobble as they travel on the rail.

Of course, I have no idea how bad each of these effects is, whether they make any difference or not, or even whether they occur at all or not. These were just my first thoughts when I saw the design. I hope it works way beyond my expectations!

[T4b]

the ShapeOko isn't perfectly orthogonal, the X and Y belts aren't tensioned exactly the same

Can't you fix that by measuring how far the machine travelled in one axis when it should have travelled, say, 170mm, and then adjust the number of steps per unit of length accordingly?

[daveczrn]

Sorry if it doesn't make zense . I posted that 5 minutes after I woke up. I'll measure my bearings when I get home and see if I can explain it better

[cvoinescu]

the ShapeOko isn't perfectly orthogonal, the X and Y belts aren't tensioned exactly the same

Can't you fix that by measuring how far the machine travelled in one axis when it should have travelled, say, 170mm, and then adjust the number of steps per unit of length accordingly?

Sure, I'm just pointing out potential sources of error. Many can be eliminated or reduced.

[ejs]

This is interesting, Dave. I had no idea that the inner races tend to be taller than the outer. But I wonder if this would be an advantage. Per cvoinescu's notes (different thread) on the twisting distortion between the inner and outer races, my observation is that with the inner races taller, pushing them together may pinch the inner balls together and allow for less of this twisting.

As for specifically having a hard stop, I'm trying to avoid it for the reasons that we see with the current wheels--tolerances. If I put a hard stop in place then the concern is that something is always going to be not quite tensioned enough. It's the same problem right now that we have with two hard stops in the form of the inner Delrin lip and the precision spacer. Both are spec'd at 1.00 millimeter, but there is 7%+ percent variance in some cases. The washers tend to be oversized, in my experience, and the lip is undersized. In light of your note about the inner races being taller, this condition should be reversed.

Though really, as it currently stands, there is only a clamping force on the center races, and a large part of the slop is caused by the outer races not engaging the lip with any level of ferocity.

What my solution brings about is clamping both the inner and outer races together, which is an advantage over the current solution and creates a much more solid assembly. Effectively, both sets of races would be acting as their own hard stops.


For cvoinescu's comments: these are all good. It's a chicken and egg thing. Or perhaps it may be a Honda and Ferrari thing. If we can continuously create components that are better than those that created them, we keep on winning. Once my first set is cut, they will replace the Delrin, and will be deformable as you say, but much less so than the plastic they are replacing. If I then cut a set with the metal wheels in place that set will be even more precise. And one more iteration might not be bad just for kicks. I would also volunteer that that way I have the fastener throughs spaced, a twenty degree offset is mandated when assembling the wheels. I would expect, first, that any offset would be incredibly minor and that any trouble would soon wear flat. Secondly, any post processing imperfections would be a nick or bump on one side of one of the beveled flats would likely be a non-issue. In terms of imperfection size, the beveled flat is pretty wide. And also consider there are two separate flats pressing down at the same time. This two point contact should level out any nooks and crannies.

As for the matter of the stretch of the bolts, that's where the car analogy comes in. Honda's last forever because the material science is quality, but the tolerances are just slack enough that you generally don't notice something is failing until it fails. With the Ferrari, everything is tuned, so anything even slightly off brings the system to a halt. We'll call the ShapeOko, oh, Mitsubishi. An under recognized solid base with room for tuner kit. So metal bits will tighten things up. But the tuner parts still need to have a little play. The longitudinal stretch in the compression bolts keeping the wheel together, in tandem with the spring quality in the M5 bots once the eccentric nuts are tightened, should help dampen any wonkiness. I also have to believe the squareness of a tightly bolted ShapeOko gantry (ideally with dual Y in place) of standard size should not deviate to any relevant degree.

Lastly regarding the belt tension. That's an important note and will be looked to. I would definitely invest in a tension meter if I put these into production.
Or use it to justify getting a lead screw setup


And Cptan, thanks!

[roberlin]

I will be really amazed if you can muster enough accuracy cutting metal with the shapeoko to beat the delrin v-wheels.

See for example the wavy lines on my heatsinks in the member-projects section. I've also cut a bunch of aluminum plates
for making an ord-bot and the circular holes are quite visibly out of round. Now, I'm sure you can get significantly better results,
but I think there is a big gap to cross to do what you are proposing.

Not that you shouldn't try. I think re-inventing the wheel is highly underrated in general

[minorthreat]

Be careful preloading your new aluminum V-wheels, and make sure you lubricate, lubricate, lubricate.

Aluminum on aluminum has a nasty tendency to gall up (cold weld). Brass or bronze would be a better material choice, and I believe the original design intent was to have the (less painful) V wheels wear and not the (more painful) Makerslide. Most auto parts stores sell an anti-seize compound (grease) that contains molybdenumdisulfide and copper, bike stores sometimes have a similar product in a spray can that's effectively drops a moly film on stuff. Either should work. Good luck.