Setup and Alignment Discussion

[Cartierusm]

I would like to start this thread in the hope that a lot of you journeypeople would help with some of us less experienced people. I think a lot of us have problems with setup and accuracy.

The way I have my CNC setup is I have a Laser, from http://www.lasercenteredgefinder.com/, permanently bolted in my Z gantry and it's bottom is level with the bottom of my collet so it will clear anything my bits will. I have the polarizer on it, which in my opinion is very neccessary. I then move the laser, about 1 1/2" away from part, to a centerline I've draw on my part and traverse down the length, Y Axis, and keep nudging the part until the laser is exactly along the Y axis. I then bolt or vacuum the part down and recheck. Once I can move the laser down the Y axis and know the part is perfectly in line I then zero out the DRO. I then find the line I drew on the part at the X axis and do the same. I don't realign the part as I know my gantry is square on all axises. The laser is very accurate, but only as accurate as my eyes, which are very good but starring at it can be daunting trying to be super accurate.

Now on my Bridgeport, non-cnc, I use a touch probe that goes into my spindle and once it comes in contact with my part it lights up, then I offset the radius of the diameter of the probe, for me .1". I then use my DRO to manually machine.

I know people like Gerry have very high end machines with retractable pins, sorry if I get some of this wrong Gerrry but i'm going from memory, at 0,0. And using such accurate machines their limit switches allow for homing to the same point everytime, which is very useful.

I've just done some testing on my machine with a dial indicator and it turns out my machine is quite accurate, but I'm using very simple snap action switches. I would love to add some limit switches that are accurate enough so I can home everytime I turn on the CNC then go to the exact coordinates, if anyone knows how I can do that without spending a ton of money I'd love to hear about it.

Now what I think a lot of us less experienced people don't know is that the big expensive machines and companies don't worry about alignment to a very specific exactness. Then use a much large piece of material or a whole sheet of MDF to cut so they just need to get close and not exact positioning. If I'm wrong I'm sorry, that's just my impression.

What I'm dealing with now is the inaccuracy within myself. I make guitars and have been for 15 or so years. I'm now trying to do this with CNC. I noticed that since I'm resetting up parts mulitple times I'm getting some inaccuracies. I then realized that I was the problem not the machine. I take a neck blank and setup with a center line I've drawn on the face and back at different points. And even though I'm making the centerlines on the face and back at the same time with the same adjustable square and going from the same side of the neck I'm getting slight variances. I also get some inconsistencies from the fact that the line can't be absoultly straight as I use mohagany and it has big pores and my pencil line jumps, not much but in the .001 range. Then trying to identify the laser on dark wood on a dark line is very hard. Then do that with a fingerboard and see if all that lines up. I use registation pins made out of wood in the fingerboard to line up with the neck so when gluing nothing will slip. SO it's a daunting task trying to work with my own inaccuracies.

What I've come up with is a pretty good solution. I made an aluminum vacuum jig that will do all my operations. It uses the same registration pins that the parts have. SO I just line the laser up with the centerlines I've scribe on the jig then slap the fingerboard blank on machine, cut the registration pins, take the part off put some wood pins in the jig and press the fingerboard on and cut inlay and so forth. This was the registration pockets will always be in line, as long as I line the laser up with the centerlines on the jig.

So if anyone wants to give their experiences on how they align a part or have a setup technique or how to resetup parts and maintain accuracy I'd love to hear from you.

[Richster]

Sounds like some good ideas there. Here’s a couple of pointers I would have, probably aimed at the lower end of the DIY machine world. Having a (relatively) cheap leadscrew driven machine, the main inaccuracies are caused by backlash in the leadnuts and general flexing within the machine frame itself.

Here’s a couple of pointers from my own experience on a DIY leadscrew-based machine

- For setting up stepper motion, first calculate the theoretical steps per unit using maths. Then, try this out on the machine over the maximum range of movement. On mine, for example, I found I was consistently getting 802mm rather than 800 on the x-axis, which was attributed to manufacturing tolerances on the threaded rod. This equates to a 0.25% error, so it was just a case of adjusting the steps per unit slightly to compensate for this.

- For gantry squareness, use the machine to make a mark in the four corners of the range of movement. Measure the diagonals between opposite corners – if they’re not the same, then things aren’t quite square. You can work out the difference using simple trigonometry, then it’s a case of adjusting the gantry accordingly. With a dual leadscrew machine, this can be done by backing off one side as necessary. Being built from wood, the flex inherent in the design accommodates this within reason, provided the gantry was fairly square in the first place.

I didn’t worry too much about homing switch accuracy, given the other inaccuracies in the machine, so just have bolt heads operating as cams on microswitches. Repeatability here seems perfectly adequate for woodworking purposes.

Just my thoughts...

[LeeWay]

My system is by no means optimal. I make lots and lots of the same parts with tolerances at about .004" to maybe .01". In this range is fine for most of my router work.
What I did once my gantry was trued and squared was to mill the base table flat in the range I wanted the machine to cut. This left about an 1/8" deep pocket.
Each of the different parts I make has it's own template that is set into this pocket and screwed down or clamped in. I ruin too many brass inserts, so I am moving to all toggle clamps. These templates register in the X and Y home position. I have a little more travel on this to give some leeway.
On each template, I have a hole drilled by the router at exactly +1" in both X and Y. If my home position gets a little off or I loose steps, I can use any template and the MDI screen to get it back pretty close to where it parks at.
This works fine for my setup on the router given the accuracy I need. I am sure there are better ways to handle it.

I was thinking of installing adjustable hard stops and just use MDI off those to get home similar to how I do now with the holes. If you approach the stop very slowly and have steppers, it should get you at least closer than cheap limit switches.

[JohnMcNamara]

Hi All

This is my first post here after watching for quite a while

This note is about getting things straight…….Very straight.

Some time ago an old but looked after Heidenreich and Harbeck lathe came my way. The bed length is about 2 Metres and weighing at three and a half tons, this is a serious lathe!

Over time there is slight wear on the bed and I am contemplating scraping her back to perfection.

One day I went into a bookshop and found a new copy in the second hand section of Connelly’s book for 20 bucks. Yep I grabbed it. For restoring old machine tools I know none better. I have already used it as a reference. Scraping is not that difficult just hard work. and common sense. I found a couple in the US here

Amazon.com: Machine tool reconditioning and applications of hand scraping: Edward F Connelly: Books

Amazon.com: Machine tool reconditioning and applications of hand scraping: Edward F Connelly: Books

One of the mysteries of alignment is using a tight wire as a straight line.

How to do this this proved to be a bit harder to track down

However the following sites have helped a lot to understand the theory.

http://www.slac.stanford.edu/cgi-wra...-pub-11465.pdf

http://www.physicsforums.com/archive.../t-220848.html

http://www.spaceagecontrol.com/calccabl.htm

I guess before the Egyptians they were using a string line

On a more 21st century note here is a link to a MIT
J. Paradiso is still there.

This would make an inexpensive electronic readout system for a tight wire alignment system. As used on linear accelerators. So it is quite accurate.

http://web.media.mit.edu/~joep/papers/WireReadout.pdf

There are other references on wire alignment on Google scholar

http://scholar.google.com.au/scholar...ment&hl=en&lr=


So now I have a plan!! The next bit is to get the time to do it, in the mean time. I would be most grateful if the CNC crowd had any further Ideas.

Cheers

Macka

[yanjue]

For homing, i use optical switch and fixed speed. The switch is NTE3100 (or ECG3100) with little experience in electronic, you have a very accurate home switch under 10$.

At my job, our machine don't have home switch. Servo motor use absolute encoder and the controler have battery to always know the position. I thing it's the best but for homing the machine, it take 4 hour.

Positionning..... the best way is with a touch probe.
Another good way is to use a jig pined in the table of the machine. So even if you remove the jig, you can replace them at the same position.

Yan

[VWSatOz]

Rather thasn the laser which seems to rely on your eyesight, why not use a dial indicator to align your part edge or jig edge, & then with the dial indicator mounted in the rotating spindle freely, clock around a reference pin or hole in the work or jig or machine table& take your XY zeros from this. This is how we do it in an engineering shop. Then if you also have accurate limit switches on the mc axis it should repeat from day to day after the mc references itself, the only repeat errors will be from thermal errors in the mc, sun shining on the column frame etc. If you build your jigs accurately with steel dowels pins &have similar accuracy with the holes in your wooden parts things should work out OK.

[Cartierusm]

Yanjue, how does the use of an optical switch ensure accuracy?

For everyone else what do big fancy CNC router machines use for homing, is it an electronic board in concert with hardware or just very accurate snap action switch? I would like to upgrade my machine but I'm not sure what to do next to ensure accuracy in homing.

[ger21]

They use a combination of a home switch, then an encoder to hoe to a specific encoder position. CNC Building Blocks used to sell a board that did this with mach3, but they recently shut down their website. I believe they still have some boards available. I have the email address, PM me if your interested.

[yanjue]

In response to Cartierusm

An optical switch will ALWAYS send their signal at the same position. A very accurate snap action switch will make the job but, maybe 6 month later, you will notice a difference in your home position. Optical are accurate year after year.

Yan

[Cartierusm]

Yan, how does dust affect the optical switch. Thanks.

[yanjue]

With large amount of dust (on the photo-sensor), the switch will not work. Clean it with a air blast and you have a brand new switch. So it's a good practice to clean the switch when you clean your machine.

At my job, one of our cnc use optical switch and the switch is cleaned each week. The machine cut and sand wood. (very dusty)

Yan

[Cartierusm]

Yan, will this switch work? http://www.jameco.com/webapp/wcs/sto...oductId=278011
I'm not sure how these switches work? Are there 2 parts to it, if the "U" shaped part is the switch what breaks the beam (connection)?