Why not use unsupported rails?

[Jonathan]

The purpose of this simulation is purely to compare rails, not to say one type of rail is better than another in all circumstances, clearly each of the 3 main types of rail has its place.
What the supported rails are mounted to has to be included, as nobody would mount them without any support, however including other factors such a bed strength is unnecessary complication and so should be considered separately. Clearly when designing a real machine you must consider how all the parts interact, so ideally one would simulate the machine as a whole. However since the rails are a major source of error it is reasonable to look at them independently to help with selection.

The main advantage with profile rails is they can tolerate high preload, so the stiffness is much higher and there is no play (so long as the preload is not exceeded), which is why I will soon replace my round rails. Round rails have much less, if any, preload so the deflection of the bearing block relative to the rail is non linear and greater. In the case of just supported rails the load rating/stiffness varies depending on the direction of the force which is a significant disadvantage if not mounted 'opposing'. The accuracy of a profile rail is not much different to a round rail - it depends how you mount them as how many of us correct the bend in a profile rail by comparing to an accurate straight edge reference? Not many as for the vast majority of machines it's excessive.

Your right john, it would be interesting to see what the equivalent would be of unsupported against...for example would 50mm unsupported be the same as 25mm supported?

Depends what you mount them to...I've run simulations to find the equivalent in supported rail of 30mm unsupported. Since you suggested comparing 40mm extrusion I have used 40mm *light* aluminium extrusion. Same forces as before.

Table, Ratio colum shows deflection relative to the equivalent loading on 30mm unsupported.



Pretty pictures attached:

Simulations.zip

So the 30mm unsupported rail is similar in bending to the SBR16 rail on weak 40mm extrusion, yet currently using the cheapest prices for each I can find the 30mm system costs 70% more. That's for the rails, bearings and end supports, or supported rails, bearings and extrusion. That implies that now the prices of SBR (and TBR) rails have decreased significantly since 2eopoz bought his it is no longer makes economic sense to use unsupported. The exception of course is a small machine (Jazz's 300mm estimate sounds reasonable) with lower forces, or laser and plasma cutters.

The thing to do now would be to run the simulation with two rails as that is more realistic, however I don't think it will tell us anything new.

[JAZZCNC]

The accuracy of a profile rail is not much different to a round rail -

Sorry Jonathan that's Rubbish total utter rubbish they don't compare in any way.!

Just the fact the bearing blocks are ground flat and true on profile bearings makes a massive difference. Bearing in mind 90% of round rail is cheap chineses tuff I've yet to see a round rail bearing block that's flat on top with sides machined 90deg. Only the very best brand quality are any where near and these can cost the same as profiled.

Then we get to the actual bearing it's self, the round rail bearings are sloppy from the begining just use a dial gauge and apply side pressure to see that, you won't find this on profiled bearings due to preloading and the precision machining of the rail.

They are night and day apart.!!

[Jonathan]

Sorry Jonathan that's Rubbish total utter rubbish they don't compare in any way.!

Sorry I should have been clearer, I meant the accuracy of the straightness of the rail, not the tolerances they are machined to. Obviously I agree that overall profile rails are machined to much better tolerances and thus by far the most accurate rails. Round rail bearing blocks and supports are merely extruded.Sloppy - i.e. clearance not preload which is exactly why I'm replacing mine. Simulation of a realistic gantry, same as in first post except both rails and 10mm aluminium plate stuck on the back, forces in X and Z and constrained at the ends:

[JAZZCNC]

Sorry I should have been clearer, I meant the accuracy of the straightness of the rail, not the tolerances they are machined to. Obviously I agree that overall profile rails are machined to much better tolerances and thus by far the most accurate rails. Round rail bearing blocks and supports are merely extruded.

Ok but even this is not quite correct.? They are very different in how they made and meant to be used.

Obviously unsupported are just straight bar with no support other than the ends so will flex in what ever direction the forces are applied.
Supported rails have the base support which dicate the strightness, again in my experience this base is far from straight or flat so if it bends or twist's so does the rail follow.

Profiled rail is meant to align to a registration egde on one rail. The rail is ground on at least 2 surfaces and in most case's all four. The rail actually as a special edge that's ground straight and true, often indicated with a arrow or machined grove/pattern. This edge is used to butt agianst the ref edge and is classed as the master rail while the other rails is left to float.
Now I know you know this but my point is that yes they are more far accurate just by the very nature of how they are meant to be used and all the ground square and true surfaces's. This is also what makes them very intolerent of uneven surface's and sloppy workmanship.

[Jonathan]

As soon as you properly fix down a profile rail it wont move, which is clearly an advantage, however if the rail is bent to start with you'll still have a bent rail.

I just checked the straightness of two profile rails and a SBR20 rail on my surface plate to check. Took readings 3 times for each over 400mm length in different places so should be reliable...

SBR20-400mm:0.03mm height variation.
Hiwin 15mm (760mm long): 0.06mm variation.
NSK 20mm (520mm long): 0.05mm variation.

The hiwin catalogue says +-0.1mm for normal precision height, but I don't think it states weather that is +-0.1mm for random rails, or +-0.1 over a single rail.

This agrees with my saying the the straightness is similar, but absolutely no reason to think SBR rails are better as they're clearly not!

[JAZZCNC]

SBR20-400mm:0.03mm height variation.
Hiwin 15mm (760mm long): 0.06mm variation.
NSK 20mm (520mm long): 0.05mm variation.

The hiwin catalogue says +-0.1mm for normal precision height, but I don't think it states weather that is +-0.1mm for random rails, or +-0.1 over a single rail.

So what you checking here Height or straightness.? Very different things.!! Also did you fasten them down along there full length.? Doubt you did and untill you do then the height check means diddly squat.!

Also your not 100% wrong about bent being bent just 50% wrong.???

Profiled rail won't tolerate any bend if it's to work correctly but that doesn't mean they won't come with a very slight bend to them over there length.? Hense the ground edge's so it can be registerd against a reference edge.
They are designed to flex ever so slightly so they can be manipulated perfectly straight and also why you have fixed master rail and a floating rail that gets adjusted by moving the gantry of what ever the bearings are attached too slowly across the rails length while tightening the rails fixing bolts, so tweaking out any bend or tight spots.! This is also why most profiled rails have 60mm centres between the holes.

Now if round rail is bent then it's bent and your stuck with it without resorting to off machine straightening. Often thou it's not the rail it's self but the rubbish base that's bent or twisted and often not flat on the mounting surface.

Profiled and round are very different beast's so can't and shouldn't be compared they are not in the same league.

[Shinobiwan]

SBR20-400mm:0.03mm height variation.
Hiwin 15mm (760mm long): 0.06mm variation.
NSK 20mm (520mm long): 0.05mm variation.

When comparing the parts yep but a more real world example would be that whilst installing on a machine you can further level the profile rails providing you've got the measurement tools. So those figures you've hinted at aren't set. They could even be made worse if you make a mess of the fixings which is less likely with the round rail as its only fixed both ends and not at regular intervals. Its as Jazz said, they're designed for precision and will give more of that over round rail but you have to exercise another level of precision when installing them to reap that.

[Jonathan]

I didn't fasten them down as the purpose of the check was to show that even before you mount the rails they are bent (I measured straightness for the profile rails) which is clearly true. The test was not intended as a realistic indicator of height accuracy in operation, as clearly either will conform to the mounting surface.

Sliding the gantry back and fourth and tightening the bolts to tweak out any tight spots also wont get the rails straight. It will merely copy any bend in the master rail to the subsidiary rail. Very few DIY builds rest the master rail against a reference edge, or indicate it to get it straight, hence for those any bend in the rail as shipped will remain (in one plane). For example if I just bolt down one of my 15mm rails to a piece of nice flat aluminum tooling plate, then use the sliding method to align and fix the other rail to the same plate, the error I measured will still be present so the measurement is relevant. That's what most people do and it's perfectly valid for supported rails.

Its as Jazz said, they're designed for precision and will give more of that over round rail but you have to exercise another level of precision when installing them to reap that.

Which is what I said originally; "the accuracy of a profile rail is not much different to a round rail - it depends how you mount them as how many of us correct the bend in a profile rail by comparing to an accurate straight edge reference?"