eShapeOko, twist around Z

Post Reply
superfrog
Posts: 19
Joined: Tue Sep 08, 2015 6:50 pm

eShapeOko, twist around Z

Post by superfrog » Thu Mar 31, 2016 4:25 pm

After nearly a year of not having enough time to finish the machine (and a few month of pre-order wait before this), I have had been enjoying a functional if not completely polished eShapeOko for a few weeks.

I am happy enough with X and Y, which come with their limitations but work for my purposes. The only problematic thing I can see is the fouling of the wheels while machining increases drag as it goes; it has not caused a problem yet and a dust/debris shoe is planned sometimes soon.

I am somewhat unhappy with the Z axis though. While it is secure along the X axis and looks acceptably steady along the Y, it twists along an axis perpendicular to the XY plane quite significantly (XZ and YZ appear ok). As the tool axis is somewhat away from the axis of rotation (which seems to be at a first approximation between the V-wheels contact points) this translates into movement +/-0.4mm at the tool when applying sideways force of say 10N +/-50% (i.e. pushing it with a couple fingers). This would be acceptable for slow cutting but as soon as the spindle is accelerated sharply, it is bound to cause problems (and I suspect is the cause for some experienced artefacts).

I am suspecting (excessive) bearing clearance and/or sideways vwheel deformation. Has anyone observed similar symptoms?

My first inclinations was to find low clearance 625 bearings to replace the current ones. They do not seem common and would probably be prohibitively expensive.
My second inclination would be to preload the hell out of the bearings (hell being only a few N on such small bearings, I suspect). With the design if the vwheel it is only possible to do so by squeezing the inner races and with the design of the machine, doing so using a fairly rigid spacer. Haven't tried, but feels like it may work.
My third solution would be to dump the Z assembly and replace it with something with less slack, using linear bearings of some kind (Why are linear rails and their carriages so expensive...). I would be tempted by a pair of HIWIN 15mm to allow the spindle to sit closer to X carriage plates. This solution comes with its own problems of price, weight and custom implementation.

Suggestions ?
eShapeko #377: "Uncle Bob" 500x1000 deluxe frame, with nicely coloured plates.
Arduino Due + GAUPS 1.1 + 4xDRV8825 at 30V running G2
800W ER11 "Ol' Sparky" Chinese spindle + VFD

cvoinescu
Posts: 4442
Joined: Thu Jul 19, 2012 6:50 pm
Location: Camberley, UK
Contact:

Re: eShapeOko, twist around Z

Post by cvoinescu » Thu Mar 31, 2016 9:35 pm

Can you tell where the movement occurs?

First of all, please check that the eccentric spacers are adjusted for a tight fit. If you can barely turn the V-wheel against the stationary rail with your fingers, it's about right. The Z axis is trickier to adjust, because it has a long screw with a single eccentric spacer. You may have to loosen those two screws and wiggle the back end so that the screw is closer to perpendicular, then re-tighten the nut and the eccentric spacer.

If it's indeed the bearings having axial play -- which they do, a little, and better ones are indeed very hard to find -- you can preload them by reducing the thickness of the washer between the bearings. The precision washers that come with the kit are 1.00 to 1.01 mm thick (which should match the width of the inner lip of the V-wheel), but the regular M5 washers are around 0.95 mm, and the Form B washers (part EM-W5B in Pack 5) are 0.75 mm thick. You should have a few extras of both kinds. You can try those in the V-wheels and see whether they reduce play without seizing the bearings when tightened firmly.

You may have bearings from an older batch. I've changed suppliers a few times in search for better quality; I'm reasonably happy with the current one. I can send you a set of the current type of bearings, in the hope that they'll be better than what you have.
Proud owner of ShapeOko #709, eShapeOko #0, and of store.amberspyglass.co.uk

superfrog
Posts: 19
Joined: Tue Sep 08, 2015 6:50 pm

Re: eShapeOko, twist around Z

Post by superfrog » Fri Apr 01, 2016 12:50 am

While I have the chance, allow me to thank you for the work you did on the eShapeOko. This is a really nice variation on the original machine and I recognise and appreciate the work you have done keeping everything running ( and the MegaRail stuff).

Thanks for your offer of bearings, I'll come back to you if new bearings are required but I'll try the preload option first. I suspect I got bearings from your previous provider, one of them was shot on arrival, but as you included a spare so it was not an issue. That said, I doubt the quality of bearings would make much of a difference as bearings inherently have clearance designed in to allow for some misalignment when fitting. The difference is probably in the quality of the race finish and the roundness of the balls, neither being important in our case, as well as how many duds you get in a set.

The spindle/mounts/Z makerslide assembly appears rigid and visibly (as in can be seen to) twists between the 4 Z v-wheels. Tightening the v-wheels through the eccentric spacers, more than I would want as I fear making flats on the soft delrin wheels, does decrease the twist a bit to around +/- .35mm at the tool (vs +- .40 )
I went through another round of measurements, this is with the v-wheels tightened a tad more than normal so that I can not make them slip with my fingers anymore (checked on all four, just in case)
For information, the distance between the tool axis and the v-wheel plane is about 70mm. I am measuring along X.
The shift is +.38/-.35 when applying similar force. Amusingly enough, after flexing in one direction the assembly settles around .05mm in the last flex direction. That would be consistent with a .05*2/7 = 14 um axial play, not exactly a lot or surprising. The rest could be down to elasticity of the metal and the delrin of the wheel.

I'll try using the .95 washers next and see how it goes. If my clearance theory is correct, .05mm could bring reasonable preload with the delrin acting as a spring for the 20 odd microns that are left. The 0.75 washers will probably be too thin (I do have shim washers somewhere though if I can find them) and may cause 'rapid unplanned disassembly' of the bearings.

For general information, in term of other flex, there is about +/- .15 mm along the X axis (under my scientific unit of a reasonable nudgeworth of force) probably because of end-plates a v-wheel assembly play. I have not tried the Y axis. For both X and Y, belt stretch probably adds much more transient positioning error than plate flex.
eShapeko #377: "Uncle Bob" 500x1000 deluxe frame, with nicely coloured plates.
Arduino Due + GAUPS 1.1 + 4xDRV8825 at 30V running G2
800W ER11 "Ol' Sparky" Chinese spindle + VFD

superfrog
Posts: 19
Joined: Tue Sep 08, 2015 6:50 pm

Re: eShapeOko, twist around Z

Post by superfrog » Fri Apr 01, 2016 12:53 am

Ohh, and I am also curious to know whether some of the other DTI wielding members of the community observe similar things on their machines !
There is no reason for any of the other designs to be better ( except for SO3 which appears to have wider Z wheels pacing, which would help some)
eShapeko #377: "Uncle Bob" 500x1000 deluxe frame, with nicely coloured plates.
Arduino Due + GAUPS 1.1 + 4xDRV8825 at 30V running G2
800W ER11 "Ol' Sparky" Chinese spindle + VFD

superfrog
Posts: 19
Joined: Tue Sep 08, 2015 6:50 pm

Re: eShapeOko, twist around Z

Post by superfrog » Fri Apr 01, 2016 10:20 am

In five thousand words worth.

Image

Image

Image

Image

Image
eShapeko #377: "Uncle Bob" 500x1000 deluxe frame, with nicely coloured plates.
Arduino Due + GAUPS 1.1 + 4xDRV8825 at 30V running G2
800W ER11 "Ol' Sparky" Chinese spindle + VFD

superfrog
Posts: 19
Joined: Tue Sep 08, 2015 6:50 pm

Re: eShapeOko, twist around Z

Post by superfrog » Fri Apr 01, 2016 6:48 pm

Upon further examination using a similar attachment method to the previous photos, deflection along the Y axis is only +.04/-0.03 with as small .0.01 dead zone in the middle either way.

I'll report when I manage to redo the wheels with some preload.
eShapeko #377: "Uncle Bob" 500x1000 deluxe frame, with nicely coloured plates.
Arduino Due + GAUPS 1.1 + 4xDRV8825 at 30V running G2
800W ER11 "Ol' Sparky" Chinese spindle + VFD

superfrog
Posts: 19
Joined: Tue Sep 08, 2015 6:50 pm

Re: eShapeOko, twist around Z

Post by superfrog » Fri Apr 01, 2016 9:10 pm

I just redid the wheels with 0.95mm washers (selected from a bag using a micrometer). Light drag can be felt on three wheels. On the fourth wheel, one of the two bearings has a slightly less clean race and gives a bit of a gritty feel, but the assembly is good enough to use for now.
The washer were previously 1.036, 1.030 , 1.031, 1.029

Play is somewhat reduced to +/- 0.2 with a deadzone of 0.05 (+0.02/-0.03 but the centre is less clear than before).

I think I can live with this for the moment, I don't think much more preload is possible without degrading bearing function. I will have to check whether machining results are going to be better.
eShapeko #377: "Uncle Bob" 500x1000 deluxe frame, with nicely coloured plates.
Arduino Due + GAUPS 1.1 + 4xDRV8825 at 30V running G2
800W ER11 "Ol' Sparky" Chinese spindle + VFD

Post Reply