How much deflection to expect...

[Wal]

Hi all,

Yep, I know the right answer is zero, but with home-built machines I guess we need to be a little realistic.

I mounted my z-axis yesterday, here it is:



No doubt it'll be on and off a few times more for shimming etc, but while I was at it, I torqued it all up and out of interest measured the deflection at the nose of the spindle with the z near it's downward limit. Pulling hard on the plate that the spindle is mounted to, my dial gauge read 0.05mm. Is that any good?

Vid here:



It's on TBR20 supported rails, how big an improvement would I see with profile rails?

Cheers!

Wal.

[JAZZCNC]

Pulling hard on the plate that the spindle is mounted to, my dial gauge read 0.05mm. Is that any good?

Nobody can answer that question Wal to many variables mate.!! . . . . . And Anygood for what.?

Edit: Thou I will say it's an order of magnitude better than others will be getting from machines they paid several G's for.!!

[Jonathan]

It's a reasonably good starting point, but bear in mind this is only half the story as you need to add to this value the deflection between the bed and frame, for the same force.
If you want a less subjective answer, then measure the force you're applying and divide by the deflection. As a rough guide, if you get at least 1000N/mm stiffness you'll probably be fine for most things, but clearly it depends on what the cutting force and accuracy requirements are.

[Wal]

>And Anygood for what.?

Ah, sorry Jazz, was being a bit vague there - just ballpark any good for cutting alu plate. I'm not too bothered about crazy feed-rates and am happy with smaller step-overs. Reassuring to hear that it's a good starting point!

One more question, the pillow blocks which house the bearings that ride on the rails - they have grub-screws that you can tighten to get a snug fit between the bearings and the rails - how tight do you go with these? Just a nip front and side?

Wal.

[JAZZCNC]

One more question, the pillow blocks which house the bearings that ride on the rails - they have grub-screws that you can tighten to get a snug fit between the bearings and the rails - how tight do you go with these? Just a nip front and side?

It's a feel thing but basically until they have a slight resistance with no play.!!. . . Thou if your planning on cutting Ali then you'd be better cutting your losses now at this stage before going further and buy some profiled linear rails. These rails are not really upto it and will need constant adjustment if you want any thing like decent accuracy.

[Wal]

Thou if your planning on cutting Ali then you'd be better cutting your losses now at this stage before going further and buy some profiled linear rails.

I hear ya Jazz, it's a well known fact that the profile rails are the way to go for superior rigidity, but I think to swap out at this stage would mean too much heartache, ball-ache and expense on what was always going to be a learner build. I may as well build another machine, which I have every intention of doing..!

While I'd like to 'precisely' cut ali on this, think more ornamental than super high precision...

Wal.

[Jonathan]

You'll be fine with those rails to start with. I'm not saying the rails aren't important, but it's the whole picture not just the rails that affects the overall stiffness and the rest of your design looks pretty strong (as it's a fixed gantry) so you'll get away with it. For example, I cut plenty of aluminium on my machine before changing the Z-axis rails to profile rails and didn't notice much difference from the profile rails as the weakest point lies elsewhere.

[routercnc]

You can look at it another way which is if the machine stiffness is 1000N/mm, and you are cutting aluminium (guide figure 50N load), then you would see 0.05mm deflection at the tool (50/1000). This assumes a rigid tool (as your data was from the collet), and ignores resonances which would add slightly to the deflection in a difficult to calculate way.

If you can't be bothered with the maths - I just checked our kitchen scales (the flat digital type) and they go up to 5kg (50N). So if you have something similar handy use them to push against the collet until they read 5kg. If the DTI reads less than 0.05mm you are in business (>1000N/mm).

[Wal]

Thanks routercnc, helpful info. I guess I could also hang 5kg's worth of weights off it..?

Anyway, today's progress:



All coming together reasonably well. I must say that I'm blown away by the minute measuring capabilities of some of these gauges. My dad's a retired engineer whose specialism was in measurement, he hasn't seen the machine and is pretty much bemused by what I'm building, heh - I guess you would be if you're used to measuring hydraulic gear to MOD tolerances... He's retired now and has given me a load of his old gauges, one of which is a Mahr with increments of .002mm - it's mad to see those otherwise imperceptible forces at work!

Anyway, one thing I will say at this stage is that the gantries move fairly smoothly without much effort when turning the screws by hand, although pushing the gantries around (back-feeding the screw) takes a fair bit of effort with a bit of binding towards the ends of the screw. I'm not entirely convinced on the quality/consistency of the screws from China. Pretty sure that the z-screw is slightly bent and I've noticed that the x ball-nut is a bit 'clicky'... Not going to get too anxious about it just yet, I'm keen to see how it all runs under power...

I think I overdid it on the thickness of the rubber aprons I'm hoping to fit to shield the running gear. 4mm is WAY too thick. I think 2mm should do it, even then it might run under the gantry, I'll have to experiment. If anyone has a spare bit of 2mm rubber sheet lying about...



More updates as the week progresses...

Wal.

[Jonathan]

If you can't be bothered with the maths - I just checked our kitchen scales (the flat digital type) and they go up to 5kg (50N). So if you have something similar handy use them to push against the collet until they read 5kg. If the DTI reads less than 0.05mm you are in business (>1000N/mm).

Good idea - I used 20kg hanging scales as they're exceptionally cheap on eBay and have a hook on the end which makes is easy to grab a bar/tool in the spindle to pull on. Do remember that, especially for a fixed gantry machine, we need to consider the stiffness of the bed.

You can think of the stiffness readings as spring constants, which can be combined as you would for springs. So in this situation you would have the same force applied to the spindle (i.e. Z-axis) and bed (i.e. X and Y axes), which is like having two springs in series - so to find the overall spring constant (i.e. stiffness), measure them both and use the following formula:

.

Then if . is greater than 1000N/mm, you're in business. Unfortunately you can't really simplify that, but if one of the stiffness readings is significantly lower than the other (say 10*), then you can just use that as the overall stiffness wont be affected much by the other one.

Anyway, one thing I will say at this stage is that the gantries move fairly smoothly without much effort when turning the screws by hand

If it doesn't, you're doing something wrong.

although pushing the gantries around (back-feeding the screw) takes a fair bit of effort

That's normal.

with a bit of binding towards the ends of the screw.

Not good! That implies the bearings on either end of your ballscrew(s) are slightly misaligned, so they're applying significant radial force to the nut when the ballnut is near the ends, but not in the middle as the ballscrew itself will just bend. Try moving the nut to one end of the screw, loosen the bearing mount (so it self aligns), retighten carefully, then do the same at the other end. You can also try leaving the bearing fixed and loosening the ballnut mount.
Do it now.. it's the sort of thing that you'll get away with for a while, but will cause problems soon enough as it prematurely wears the bearings/nut.

I've noticed that the x ball-nut is a bit 'clicky'

I'd look into that sooner rather than later, before it's too late. If you dismantle the ballnut you'll probably find there's a ball in the wrong track, or bits of swarf in it. I've had that problem with new nuts more than once.