A sufficiently strong machine

[Jonathan]

I mentioned something similar to you in an email Jon and feel the machine would benefit from 1 more rail on both Y & Z axis. One Higher up on Y axis and on 3 sides of box on Z axis but more to help with resonance and vibration in all cutting directions rather than increase load ratings.?

As I'm sure you're aware, you can improve the response of a resonant system in just two ways - either raise the stiffness to reduce the magnitude of the deflection, or increase the damping so the error dies away faster. Therefore comparing the stiffness of each option is very much related to comparing the overall response for resonant conditions.

The linear guides do have some damping effect, due to the oil layer between the bearings and rail. Adding rail with two bearings, as you suggest, would therefore increase the damping factor, as it increases the surface area in contact. Instead of adding an additional rail and two bearings, you could increase the stiffness and damping by adding one bearing to each of the existing rails which is likely more cost effective. This also has the added slight bonus of evening out errors in rail straightness and generally aligning two rails is a easier than three. Either way, this damping effect isn't that large as the surface area in contact is small, so you're probably better off improving damping in other areas such as joints in the frame. You can also add non-load bearing sliding contact bearings to further improve the damping effect, which will make a bigger difference as sliding bearings will have a much greater surface area in contact.

If you take the load ratings from the Hiwin datasheet and plot them versus rail size, it seems that the load rating is proportional to the rail size raised to the power 1.7. So if we assume the stiffness of the rail is proportional to the load rating and that the magnitude of the force on each linear bearing is similar, to increase the stiffness by the same factor as adding one additional rail (i.e. 50% as you're spreading the load between 6 bearings instead of 4), you only have to increase the rail size by 27%. So for example going from a 15mm rail to 20mm, or 20mm to 25mm would gain slightly more stiffness than adding the additional rail. Similarly if you want to double the stiffness, then instead of going from two rails to four, you could increase the rail size by 50% (e.g use 30mm rails instead of 20mm).

Looking at it a different way, if you plot the price of the linear rails and linear bearings versus their size, it's a pretty convincing linear relationship. So by increasing the rail size you have an exponential gain (x^1.7) in stiffness for a linear increase in price. If you add more rails you have a linear gain in stiffness for a linear gain in price.

Would also add another ballscrew to Y axis rather than increase bearing plates has the action of 2 screws pushing/pulling together would be smoother than widening the plates.?

Both methods would work. If the bearing spacing is made large enough to prevent racking, then the stiffness of the axis would also be greatly increased since the deflection for a given force due to the bearings is proportional to the bearing spacing squared (based on combining the formulas in the manufacturer's datasheets). Similiarly, adding the extra ballscrew would eliminate racking and thus increase the stiffness, however it cannot increase the stiffness further so the former or a combination of both is required if the axis stiffness is still not sufficient.

Like you say it's very strong already but feel these changes would make it the close to ULTIMATE strong machine.!

Got to draw the line somewhere.

-so you believe 3kw spindle would be better than 2.2kw?

Yes, if one or more of the following are true:
1) You need to get more power than the 2.2kW spindle can deliver at low speed, e.g. for cutting steel, but still require a high speed spindle.
2) The machine is rigid enough for the stiffness of the spindle to be the limiting factor.
3) The machine is rigid enough to make cuts which exceed the power rating of the spindle.

Number 3) is definately the case for the macine in this thread. I'm not yet certain about the rest so wont comment.

-how you did the oiling system? did you use the original nipples and modify them?

No, if you look carefully in the photos you can see that to save space new nipples were machined which were the correct size for the tube to just push on to it and not fall off.

why not grease?

Oil is much less viscous, so it requires less pressure to push through the system which in turn makes the oil distribution system easier to manufacture. Oil also allows higher speed, but that's not really an issue here. Grease is commonly used for the linear bearings, but oil seems more common for ballscrews.

how much oil is used under operational condition? What oil?

Very little. The datasheets for the ballscrews and linear bearings do give recommenced values. Bear in mind the oil is also good at flushing out foreign bodies from the nuts, so certainly no harm in oiling them regularly. You can find the correct oil to use in some of the ballscrew datasheets, but I wouldn't worry too much about what you use as, although not recommended, the ballscrews will last a long time without oil, so anything will be a lot better than nothing. You probably don't want something with too low a viscosity as that would drain out quickly.

-where did you get that spring shims that make the preload? what are the specifications? how did you calculate the distance between the 2 nuts , so that they would be mirroring each other exactly and the shims to fit in?

I got them from Lee Spring, spring manufacturer for a variety of uses. The preload force is set as a percentage of the screw's rating, so you need to find the rating for your screws and then find disc springs which can apply that force without being fully squashed. I didn't calculate the distance, instead a gap is left, measured, then a spacer made to the correct thickness to squash the springs by the calculated amount to obtain the required force.