Tramming With Taut Wires

[Kitwn]

If you can't build it precise, build it adjustable. Regular readers of this forum and anyone who has tried to build a machine using only a drill press and hand tools understands the significance of this. If you've welded up any of your machine out of steel box section scavenged off the local tip as I have then this is Design Parameter Number One.

The next trick is what to measure and how to measure it in order to make useful adjustments that will turn your whirring monstrosity into a precision instrument?

The most basic alignment is to get the fixed rails (X axis on my machine) straight and parallel so that they present a level, flat plane as the foundation for everything else. Whether you've done this by testing the base on a monster slab of known flat granite, used the epoxy leveling method which relies on the consistency of the Earth's gravity within the floor area of your shed, built a Devmonkey type optical laser alignment device or used a precision spirit level and adjustable feet (as I do for now) is up to you. Once this is done the next thing is to get the gantry aligned. That's what the rest of this blather is about.

For want of knowing any better at the time, I built my machine using 20mm continuously supported rails for the X axis which sit a little above the height of the machine bed. This allows me to use a pair of taut wires stretched across the rails as a reference for measurements. This has to be more accurate than messing about with pieces of glass and playing card shims and was certainly quicker to set up. My thanks go to John McNamara for inspiring this method. I appreciate that the sensible people don't have their rails and linear bearings sitting right where all the chips fly and might not be able to use the method described here, but your comments and suggestions will be very welcome. Raucous laughter will also be tolerated within limits.

As with all tramming methods an arm carrying some measuring device is needed to fit in the spindle. Mine is just a piece of wood with a 12mm rod sticking up at one end to fit in the spindle and a piece of brass bar at the
other facing down to act as a probe which makes contact with the wire. A 5V power supply and an LED indicate when contact is made. Precision is not required in the probe design but it must be sufficiently rigid to give accurate results.

The length of the arm is set to allow the wires to sit outside the footprint of the gantry feet linear bearings and the wires positioned so contact is made when the arm is aligned parallel to the X axis (90 degrees to the gantry). The Z axis drive is used to lower the probe until it makes contact and then raise it again until contact is only just lost. The Z axis position as displayed on the DRO is then the required measurement. The machine is used as it's own micrometer. The probe is rocked from side to side a little to ensure the lowest point of the probes rounded end always makes contact if it can. Measurements are made on both front and back wires at each end of the Z axis' travel along the gantry and at intermediate positions if required.

Ideally all measurements will be the same

More on processing the measured values later.


Stainless steel wire bought from the local fishing tackle shop and precision measurement probe.

Taut wires in place.

Making a measurement.

[Kitwn]

Here's a quick video showing how the measurements are made...




Any difference between measurements on the two wires means the gantry is leaning forward or backwards and shims are required on both the front or back mounting bolts of the gantry.

If the difference is different at each end of the Y axis (the rails mounted on the gantry) travel then the Y rails are not parallel and shims are neeeded under one of the rails themselves.

An average of each pair of measurements relates to the height of the spindle above the wires. If this average is different at the two ends of the gantry then the gantry is tilted to the left or right and shims are needed on both mounting bolts at one or other end of the gantry.

The DRO on LinuxCNC displays to the nearest micron. Somewhat optimistic since 800 steps/rev on a 16/10 ballscrew equals 12.5 microns per step, but I think the readings are going to be useful enough to get the machine rather better aligned than it is now.

[Kitwn]

Some results! Numbers are to 2dp to avoid going way off for calculations but in reality there is some interaction between the effects of different errors and I will need another set of measurements after the first round of shims go in. At least I have some ballpark figures for cutting the required shims.

There's a 1.78mm change of height between the two ends of the Y travel. Obviously skimming the base will hide this for some jobs but therewill still be a geometry error with the Z axis. A little over 2mm of shims will be required at one end of the gantry.

The measured diference between the front and back wire measurements varies between 0.44 and 2.23mm so the gantry is both leaning backwards and the rails are not parallel. 0.44mm over the 600mm between the wires equates to shims of about 0.1mm at the rear bolts of the gantry base.

The difference between the differences of the measured values calculates out to require shims totaling 0.42mm at one end of the lower Y rail. Since the whole rail will need some pretty thin shims at each fixing point I'll correct for the gantry leaning back here as well. This will be the longest and most tedious job but it will be possible to fit a shim, make a measurement, correct the shim, move the spindle, make a measurement, fit a shim..........

To put the twist in the Y rail plane into context, the measured error works out to cause the tool tip to make an unwanted move of about 1mm along the X axis when the Y axis is moved between it's two extremes. When squaring up my gantry using the draw-a-square-and-measure-the-diagonals method I have never been able to draw a square with the two diagonals equal in length to better than about .5mm. This is probably the reason.

Once the gantry is sorted out then the Z axis can be aligned as well.

Looking at the calculated thickness of shims required I'm actually quite impressed! The adjustments needed for my hand-built machine seem very small to me, especially as building this gantry was the first welding I ever tried and was done without the use of a proper welding table to jig all the bits together. I hope my experience will encourage other prospective first time builders to have a go. Just make sure that all your adjsutment points are designed in from the beginning and can be accessed easily once the machine is complete.

Kit

[diycnc]

Following this with interest.
I have just purchased a small 300x300 surface plate to use when calibrating my machines when its finished. Very similar build to yours, steel frame with supported round rails.

[Kitwn]

Following this with interest.
I have just purchased a small 300x300 surface plate to use when calibrating my machines when its finished. Very similar build to yours, steel frame with supported round rails.

I've seen you're looking into epoxy levlling. I didn't know about the right way to do epoxy leveling when I built my machine but would strongly recommend you look at other posts on the forum that describe using West Systems 105 resin with their very slow hardener type 209. This looks like the best way to get your rails as straight as possible from day one and my biggest problem, the twist in the Y axis rails, would not exist if I had done a better job of researching this forum before I began the build.

I look forward to seeing your finished machine.

[Kitwn]

This has been a day of measurements, fitting shims, making measurements, adjusting shims, making measurements... repeat and repeat!

I've never done a job like this before and it's been very interesting to go round and round slowly, but very surely, getting closer and closer to optimum each time. The measurement process using the Z axis and it's DRO as the measuring device is not as accurate as a dial gauge but has proved very effective and easy to use. Repeatability has been better than 0.05mm which is hardly world class engineering but is entirely adequate for this job and probably matches the accuracy available from the usual YouTube tramming methods involving leveling a piece of glass on the machine base or simply skimming the base and testing for ridges. We will see when I skim the base of this machine and see how smooth it is.

Having got the gantry rails as accurately aligned as I can it is possible to adjust the spindle bracket vertical. There is significant play in the mounting bolts but by positioning the gantry so that the spindle axis is directly above one of the wires it is possible to adjust the exact position of the bracket so that the probe just touches the wire on each side, indicating the spindle is vertical in this plane to within about .05mm in 600mm (twice the length of the probe). The wiggle room provided by the mounting and it's bolts allows a variation of approximately +- 10mm in the height of the probe relative to the taut wire.

The next job is to skim the machine base so that I can confirm the movement of the Z axis is accurately vertical.
l

[John McNamara]

Hi Kitwn

I like the way you use the spindle to rotate the sensor contact using the spindle itself. and the way you use reversing it on the same wire to establish that the spindle is at 90 degrees from that wire.

I am a little confused about relying on the method to align both Y (Along the table) axis bearing alignment wires,

In order to do this the carriage must be geometrically perfect, any twist making the carriage rails deviate from co-planar would mean that the long axis rail alignment wires would be adjusted to the error introduced in the carriage.

Taking a measurement from another measurement system is always difficult.

Ideally the long axis wire need to be aligned first with the lased method then using your method the carriage twist could be minimized using the y axis wires.

I hope I have understood your photos correctly.

Regards
John


Useful links:

Do a google search for wire alignment
https://www.google.com/search?q=Wire...hrome&ie=UTF-8

A useful and hard to find sag table
https://web.archive.org/web/20100602...wire-sag_1.htm

Sag calculator (Tick metric if you use metric it does both imperial and metric)
https://www.spaceagecontrol.com/calccabm.htm

[Kitwn]

I like the way you use the spindle to rotate the sensor contact using the spindle itself. and the way you use reversing it on the same wire to establish that the spindle is at 90 degrees from that wire.

I am a little confused about relying on the method to align both Y (Along the table) axis bearing alignment wires,

In order to do this the carriage must be geometrically perfect, any twist making the carriage rails deviate from co-planar would mean that the long axis rail alignment wires would be adjusted to the error introduced in the carriage.

Taking a measurement from another measurement system is always difficult.

Ideally the long axis wire need to be aligned first with the lased method then using your method the carriage twist could be minimized using the y axis wires.

I hope I have understood your photos correctly.

John,
Thanks for having a look at this, I have seen your PM.
I haven't used wires along the long, fixed axis of the machine (this is the X axis on my machine which might be some of the confusion) but have assumed these rails are straight enough. Checking they were coplanar was done using a precision spirit level and adjusting the feet of the frame which is not especially rigid. Sometime in the future I will make a suitable camera mount and use Joe's method to get these rails as perfectly straight as possible, but for now they will have to do as they are. My aim for this work was to get the gantry better aligned than it has been since I've always known this was the least accurate part of the machine.

Based on YouTube research, tramming the spindle with dial gauges usually involves some way of creating another plane using a piece of glass and shims that is coplanar with the fixed rails. This seemed overly complex and prone to error, hence going for the taut wires across the rails themselves. I think the video shows this best, though the opening still actually shows the gantry positioned for the final measurement I made for checking the spindle vertical as you mention.

In practice it has proved an effective method. Averaging front and back measurements allowed me to shim the gantry height at one end to remove any end-end tilt and taking the difference of the same measurements indicated a front-back lean. A difference in the differences indicates a twist in the plane of the rails on the gantry (Y axis). By taking measurements with the Z axis assembly positioned over each set of Y rail fixing screws it has been possible to shim the upper or lower rail at each point quickly and easily.

Some iteration was required but a day's work has got pairs of measurements all along the gantry to match within 0.3mm over a 600m measurement base. Hardly the highest engineering standards but more that an order of magnitude better than I started with. Using the Z axis DRO as the measuring device has limited accuracy but made measurements quick and simple at no extra cost!

I've since been able to square the gantry so that the diagonals of a 500mm square are exactly equal as far as I can tell using a set of made-for-purpose dividers. This is a noticeable improvement from the past when it has never been possible to get them within 0.5mm, presumably because of the distortions in the gantry set-up.

Kit

[devmonkey]

Are we talking about bringing the gantry into plane with the bed or tramming the spindle so it is perpendicular to an already planar gantry? They are different things right?

Assuming you have got your gantry in plane already (helped by ensuring your long rails are planar to start with), I then use an iterative method to tram the spindle itself.

1. Level the spoil board with a smallish (10mm) endmill. It will be slightly scalloped if your spindle is out of tram but this doesn't matter.
2. Put a 90 degree bent bar in the spindle with a DTI on the end and sweep it across the bed, I used about a 200mm diameter sweep. This will give you the tramming error, correct the spindle by rotating it left or right / forwards backwards and then goto (1). Note that left/right has to be done at the spindle not by jacking up one of the gantry carriages or you will take the gantry out of plane. If you can't adjust the nod (forwards backwards) n the spindle then you can jack up the back or front of both gantry carriages equally.

That is how I do it anyway, I do something very similar to tram my milling machine.

[John McNamara]

Hi All

If you are tramming a conventional milling machine one quite useful tool to assist in tramming a spindle is an old brake disk, car service centers usually have a skip full of them, you should be able to get one for free or almost free.

The disk is simply placed on the table center, its surface allows you to swing the dial indicator smoothly around the face of the disk without the risk of damaging the indicator tip on the table T slots if used directly.

Some disks have a deep boss while others are quite shallow. You will need a lathe big enough to face the old disk flat and parallel.

Regards
John