DIY Laser levelling using webcam and laser level

[John McNamara]

Hi Joe

I do have an optical microscope, A cheap student version I used to align my lathe in 2011 using wire. The is a longish post here that discusses lathe alignment and wire, https://www.model-engineer.co.uk/for....asp?th=136771 In this case I lashed up a measuring stage that allowed me to position the wire centrally in a slot by eye using my bio computer then measure off the error using a micrometer. It worked! Page down the thread to see photos.

In particular note the simple wire positioning stages clamped to the lathe bed made from scrap steel, rock solid.

Using a USB webcam or microscope with your software is clearly a better solution. It eliminates a number of sources of error.

I don't have a working USB microscope I will order one! I cant wait to test the software.

In many ways two cameras will be even more useful, for XY measurement, Ideally using one computer, although old PC's are easy to source for nothing or next to nothing, if two instances of the software could interface to two USB ports on one computer it would be more convenient.

Regards
John

PS stretched wire plumb lines are used to this day by heavy machinery installers, for instance passenger lifts and giant boring mills to align the column. Normally the bob weight is suspended in a container of thick oil to damp out any tendency to sway like a pendulum.

[devmonkey]

Quick update,

I've managed to set the master rail reasonably straight with the setup. It is within +/-5um along the entire length. It was not easy for a few reasons I will outline:
1. The microscope axis is offset in both X+Y from the centre of the carriage, I did this because obviously it has to be offset in Y as you can't 'see through' the carriage but also in X since I couldn't fit the microscope next to the carriage and remain over the frame. This arrangement is highly non-optimal as you are measuring an offset AND an angle. Eventually you get used to how much adjustment to make, however there must be a better arrangement that avoids this coupling.
2. Twisting the carriage hard on the rail can induce a few um of error.
3. Pressing hard on the machine frame and deflect the entire frame a few um.

The conclusion is that with a machine constructed like mine from 80x80x3 box section over a long dimension of 1300mm aligning to an accuracy of +/-5um is probably the absolute limit and it isn't sensible to try to achieve any greater. Also the frame is unlikely to hold this in actual use due to the deflection under force measured. Still it is good to be able to measure these things to get an idea where the limits are.

So in conclusion I've managed to align to ~1/2 a thou with a setup that cost a few quid. It is all entirely dependent on the taut wire actually being straight, however if we assume it is then this setup is equivalent to a DIN 874/00 spec straight edge, one of those 1500mm long will cost £2k then you still have to actually align the rail with it.

Here is a snippet from the alignment:


Quite pleased, time to get the gantry on now!

[Kitwn]

2. Twisting the carriage hard on the rail can induce a few um of error.

To me this says a lot about the quality of your rails and carriages. It also makes the case for a well aligned frame if you want smooth movement without binding.

[devmonkey]

To me this says a lot about the quality of your rails and carriages. It also makes the case for a well aligned frame if you want smooth movement without binding.

I would say it says that standard hiwin rails and carriages are pretty good, would you agree? This is an artificial setup as carriages wouldn't be used singly in a machine. The related interesting fact is that with these 20mm hiwin rails, with a section bolted down hard, moving the microscope 60mm to the next unbolted hole, finger pressure is sufficient to deflect the rail +/-3um although a small contribution will be from frame deflection. As said before everything looks like jelly when you look closely enough.

Also torquing up the mounting bolts with light pressure on an allen key moves the rail 1-2um, I set the rail at -2um with the bolt snugged down, then tightened it with the allen key that brought it to zero. I've just redone the straightening procedure this way and I'm now +/-2um along the entire length, job done!

The rail I chose as the master had two 'bends' in it, it was 'S' shaped, with the two ends zero'd there were two intermediate points along the rail which were roughly +/-70um respectively. I verified this wasn't some artifact of the taut wire or the software algo by changing the wire twice, moving the wire guides and changing the end that had the hammer hanging on it.

This shows that you definitely need to straighten the rail, it is not sufficient to use it out of the box.

[Kitwn]

I would say it says that standard hiwin rails and carriages are pretty good, would you agree?

That was my meaning, ' a few um' detected by the microscope is not much movement at the carriage. These rails are known to be intollerant of poor alignment, non-straight frames etc. and that's what you'd expect where there is so little play. When I decided to go for fully suported round rails on my machine part of the decision was that I was unsure I coud make the frame accurately enough for the more expensive Hi-Wins not to bind.

[John McNamara]

Rails are never straight out of the box they always need to be bolted down. The bolting should be done with a torque wrench for uniformity. If you can get hold of this book or find it at the local library you will see a large section on mounting rails and other components.

https://www.amazon.com/Precision-Mac.../dp/0872634922

I got my copy at Foyles many years ago for a lot more, A great book if you build machinery.

In it he explains the way the rail distorts when tightening it down. Chasing microns is great fun to me! they hide in every corner of your design.

I would not be concerned about a small amount of movement when you rotate the carriage axially on the rail. If it is a brand name rail you should find the type number engraved on the rail and the bearing blocks. By looking up the manufacturers website you should be able to determine the bearing clearance or preload and accuracy for the particular combination you have. typically around 3 clearance levels may be offered in addition to different accuracy levels. If you are ordering a new rail and bearing set for a project it is tempting to order the maximum preload, this is not necessarily a good idea at all. maximum preloaded bearings are often rated for a shorter life span and more importantly they require more accurate mounting surfaces, using the wrong clearance type can seriously reduce the bearing life and greatly increase the force required to move the bearing placing a high load on the drive system. Fortunately all the bearing manufacturers offer excellent manuals available for free download.

I ran this Google search today and it yielded some interesting results.

precision machine design slocomb mounting linear rails

Or use the link below.

https://www.google.com/search?rlz=1C...4dUDCAw&uact=5

Regards
John

[devmonkey]

Rails are never straight out of the box they always need to be bolted down. The bolting should be done with a torque wrench for uniformity. If you can get hold of this book or find it at the local library you will see a large section on mounting rails and other components.

https://www.amazon.com/Precision-Mac.../dp/0872634922

I got my copy at Foyles many years ago for a lot more, A great book if you build machinery.

In it he explains the way the rail distorts when tightening it down. Chasing microns is great fun to me! they hide in every corner of your design.

I would not be concerned about a small amount of movement when you rotate the carriage axially on the rail. If it is a brand name rail you should find the type number engraved on the rail and the bearing blocks. By looking up the manufacturers website you should be able to determine the bearing clearance or preload and accuracy for the particular combination you have. typically around 3 clearance levels may be offered in addition to different accuracy levels. If you are ordering a new rail and bearing set for a project it is tempting to order the maximum preload, this is not necessarily a good idea at all. maximum preloaded bearings are often rated for a shorter life span and more importantly they require more accurate mounting surfaces, using the wrong clearance type can seriously reduce the bearing life and greatly increase the force required to move the bearing placing a high load on the drive system. Fortunately all the bearing manufacturers offer excellent manuals available for free download.

I ran this Google search today and it yielded some interesting results.

precision machine design slocomb mounting linear rails

Or use the link below.

https://www.google.com/search?rlz=1C...4dUDCAw&uact=5

Regards
John

Mine are medium preload (ZA) mate not high preload, rail is now straight enough for me, although huge fun (not!) there is no point chasing down +/-2um when I can deflect 300mm of 80mm box section that the rail is mounted on between uprights by 5um just by leaning on it. I listed all those example measurements to show what the system is capable of, e.g. obviously the carriage will twist on the rail when twisted, if it didn't then it wouldn't move, but in reality this is only a problem if you have a microscope bolted to the carriage as in actual use two carriages would be used mounted to a rigid plate, I was highlighting this as a source of error with how I mounted the microscope not a criticism of the hiwin system. Possibly a better (although more work) microscope fixture would be to register to the reference edge machined on the hiwin rail rather than use a carriage at all, couldn't be bothered though...

[John McNamara]

Hi Joe
I agree medium preload is a good choice, and when the carriage is installed there will be 4 bearing blocks distributing the load, as they are spaced apart any axial twist will be much reduced. I suspect that if you ran the test with the microscope mounted in the carriage running on your already precision aligned rails You may get even better numbers due to some of the residual errors in the rail alignments averaging out.

Hi Kitwin
You mentioned that you were working with round rails, Maybe if you first fit both rails with the mounting screws just tight enough to allow them to move sideways then after sliding on the bearing pairs onto each rail you fit the carriage plate on the bearings in the same manner just tight enough to move you will be able to align one as a master rail using wire alignment and the microscope before setting up the other.
It may take a couple of cycles of this procedure but it will iron out any errors.

This worked for me when building a small All laser cut CNC router with a mate for a Men's shed in Bright Victoria in 2012 It is still in use daily apart from the Covid shut down.

It uses round rails, and yes timing belt drives apart from the Z axis which is a ball screw. Remarkably the accuracy is a lot better than you would think. After wearing out 2 Makita routers due to the high usage It now has a Chinese 2.2kw high speed spindle.

https://www.google.com/search?q=Brig...%3Apinterest.*

Regards
John

[Kitwn]

Hi Kitwin
You mentioned that you were working with round rails, Maybe if you first fit both rails with the mounting screws just tight enough to allow them to move sideways then after sliding on the bearing pairs onto each rail you fit the carriage plate on the bearings in the same manner just tight enough to move you will be able to align one as a master rail using wire alignment and the microscope before setting up the other.
It may take a couple of cycles of this procedure but it will iron out any errors.

This worked for me when building a small All laser cut CNC router with a mate for a Men's shed in Bright Victoria in 2012 It is still in use daily apart from the Covid shut down.

It uses round rails, and yes timing belt drives apart from the Z axis which is a ball screw. Remarkably the accuracy is a lot better than you would think. After wearing out 2 Makita routers due to the high usage It now has a Chinese 2.2kw high speed spindle.

John,
My main concern with the round rails was that a single carriage will rotate around the rail so some form of 'gantry' going accross both will be required to mount the microscope. Probably not a real problem in fact, especially if the slave rail is quite losely mounted. Nearly all the error seen will be from the nearest rail and a few iterations should sort things out as you say.

I'm not sure you can fully adjust round rails in practice. The steel rail must be far more rigid than the aluminium support and whilst shims may well sort out vertical errors, I suspect any attempt to nudge the thing sideways will more likely distort the support than the rail. I don't speak from experience however.

One day I might decide to do a serious upgrade and build a mostly-new machine in which case I will be using all the wisdom gathered from you and Joe to chase down every micron I can. Alternatively somebody with better coding skills than I will come up with a method that can detect all the errors on my machine and writes a dedicated post-processor to take them out in software. Oh No! Not another rabbit hole to head down!!!

Very amused by your last comments, my version reads "After wearing out 2 Makita routers due to the owner being being stupid it now has a Chinese 2.2kw high speed spindle"

Kit

[John McNamara]

When we built the router in 2012 it was for a community men's shed. we were on a parts budget, My mate and I put our time in for nothing so there was no cost there. A Makita router was available, from memory it was donated.
The Chinese spindle. Is much quieter than the Routers with their brush type motors. I guess I should of put this in a separate thread, it is a bit off topic. This thread is about Alignment and software.