View Full Version : Setting rails straight, and keeping them straight - how?

11-13-2005, 08:04 AM
Hi All

One of the most fundamental points of CNC machines is that the axes are straight. But, with good measuring tools (eg. autocollimator (http://www.spectrum-metrology.co.uk/newsslide.htm)), we will see curvature, bumps, twists, dips, etc. in any rail. All of us have probably fiddled with at least our long x-rails and have gotten to the point where we decided, "this is straight enough".

Now that we have a very stiff gantry, we can see that our x-rails are not as straight as we had thought - at some spots, one flank of a v-roller loses contact with the rail. (I can slip a 0.002" feeler gauge in). A more flexible gantry simply bends to accommodate that small imperfection, and maybe the stiff gantry will cause the high spots to wear off....?

Has anyone got some ideas on getting really straight rails? Let's keep it to our SB-styled "sharpened" plate/angle edges, and not get into linear bearings. (Unless we look at the beds for linear bearings, which also have to be straight - the bearing just bends to fit the bed).

For, our gantry lengths, we can use a local guy that has a "bed way grinder" on which he repairs lathes and other machine tools, but our wide table won't fit on his grinder to do the x-rails. I am trying to think of a way of re-grinding the x-rails after they are bolted/welded down......

11-13-2005, 09:05 AM
I just had a picture in my mind of a cutter/grinder attached to the Y cart, grinding the rail as it travelled down it's path. The first 2.5 feet would have to be done manually...

Just a thought...probably garbage...


11-13-2005, 09:14 AM
No Chip, not garbage. We can start with rails that are too long, and cut off the "bad" points later. Snag is, how do you run the y cart down a straight path?

One could lay down temporary rails, provided that one can find straight temporary rails. Have thought of alu extrusions, but their specs aren't all that great either.

11-13-2005, 09:48 AM
Just to illustrate the problem of "straightness".....

Let's say we had a perfectly straight (theoretical) steel round bar 2.5" diameter, 8' long. When that bar is supported horizontally, held by its ends, the center droops about 0.02" under its own weight. Of course someone will say that the bar is not straight - put a trestle under the center of the bar and you can bend it 0.02" the other way. This illustrates the myth of the straightness of linear bearings - they are only as straight as the bed that you made for them to lie on.

11-13-2005, 11:32 AM
Well, A pair of "calibration" rails attached to the "working" rails would probably do the trick. If you used them only a few times a year, they would not be worn down appreciably for several years.

Retractable wheels could be mounted on the Y cart to run down the calibration rails, and lift the y cart off of the working rails.

Checking level and true would only require a string stretched tightly across the top.

There would have to be a solid, adjustable, metered mount for the grinder, that can be attached to each side of the Y cart.

What better way to ensure a true wheel path than by using the machine that runs on it?

You're right...the idea wasn't garbage...


11-13-2005, 12:21 PM
Gerald, just a thought here. For setting and checking our guitar necks for straightness we use an aluminum beam with a stand-off at each end and a dial indicator in the middle. In our case the standoffs are two 1/4" allen bolts with jam nuts to secure them. The dial indicator is zeroed on our flattest machine table.

Then to fine level our fingerboards we use an aluminum I-beam with 3M stickit grit.

The face of the I-beam was previously trued by rubbing it on the machine table using stickit as an abrasive.

Of course this would only take down the humps. For the dips, would it be possible to hold an anvil against the outside of the rail and lightly tap the inside of the rail with a ballpeen hammer to displace the metal a bit and force the edge upward?

11-13-2005, 12:40 PM
I have put hardened bwc rails on bright steel angle on my benchtop Y axis and the results were excellent. I havn't applied high tech measurement, but the movement is way superior and the rails much straighter visually and every other way I can devise. I did use the next size up though.

11-13-2005, 12:54 PM
There will be ways to force an already machined rail into position - hammer blows, shims, weld/heat shrinking. But this will probably leave smaller wobbles that are still best removed by a final skim.

A big part of the problem is to actually measure the straightness of the 2 individual flanks of the ^. I don't think that string is anywhere near to good enough for this. Sure, a string will ensure a straight enough line for a cut board, but the snag here is that a stiff gantry forces the roller to climb up the rail, throwing out the z-height. (The climbed-up roller also doesn't seat firmly and that gave us some of our "chatters")

rh, wow, you are blessed to have bright angle iron over there! Increasing the wheel size would reduce the wear that we see on our soft rails.

11-13-2005, 01:12 PM
To understand what a collimator does, imagine mounting a laser pointer on a short wheelbase car that ran down the rail, with the beam pointing like a headlight down the rail. But 30' away there is a target for the laser to shine at. If the rail was straight, the laser would stay steady on the target as the short little car rolled down the rail.

A collimator is a telescope looking at a mirror mounted on the short car. Looking in the telescope you see in the mirror the face of the telescope dead center, unless the mirror car has tipped or swayed. This is put very crudely to illustrate the principle only.

11-13-2005, 01:32 PM
Here's an earth moving idea:
Start with long rails, maybe 2 to 3x longer than what you finally want to end up with. Make a grinding jig with a long wheelbase to ride the rails. every pass should get better and better. On the same principal as a motorgrader used to finish grade moving soil. The reason they're so long is any humps the front wheels run over is a lot smaller hump transfered to the blade. The next pass you've eleminated the original hump and now have a small hump in a different location. You could even get a little high tech by using a laser to adjust your grinding wheel height just like a lot of motorgraders use.

11-13-2005, 01:48 PM
Our slightly flexible SB gantries and cars find themselves an average position between their 4 wheels. Expanding on Dirk's motorgrader (3 axle), our grinding gantry could have 4 or more slightly flexible wheels per side.....

11-13-2005, 02:37 PM
Here (http://www.windhoff.de/e/v1/sch/index.htm) is a big rail grinder. Seems like the line is given by "6 rubbing blocks".

11-13-2005, 03:30 PM
Gerald, don't you have high grade steel over there? What happened to those big factories the pigs are flying over on the Pink Floyd album covers?

But seriously... I think that a better grade rail like RH uses is the answer and I'm gonna look into that.

As for the string measurements and the wheels running up on the knife edge, are you sure that both motors on the Y cart are keeping up with each other? If they are, you may just have to do a little selective bending here and there.


11-13-2005, 04:21 PM
Hi Chip

The big factory on the Floyd album is Battersea power station in the centre of London...brings back many memories as I used to make their PA systems.

The fist modification I made to my earlier PR96 was to bolt bright steel to the existing Unistrut and then set the V bearings to run on the corner.....it worked extremely well for many years.



11-13-2005, 04:36 PM

Assuming that you get the rails perfectly aligned, what affect would temperature have on 'misaligning' them? My shop can have a 70-degree temperature change during the day, depending on the season. I would guess that the Shopbot doesn't expand and contract perfectly uniformly. A variance of only 0.002 seems pretty good to me.

Allegheny (Unregistered Guest)
11-13-2005, 05:16 PM
Look at: http://www.lmwatts.com/straightedge.html

11-13-2005, 10:59 PM

I would think it is very hard to figure out where the high spots are by observing deviations when the car is moving. For instance, where you observe the 0.002 space, that might not be a low spot in the rail below that wheel, but rather a cumulative effect of the low spots at other wheels raising the car at that point (considering your stiff car).

Perhaps, you can remove the gantry and once you get the level of the rails "close enough", perhaps tape "Articulating paper" on the rails from a Dental Supply House to show exactly where contact is being made. Dentists use this paper when repairing teeth. When patient bites down on the paper, the high spots are marked on the paper and the dentist grinds down until he sees uniform marks on the paper.

I would think that you could somehow rig a straight and level piece of aluminum to slowly lower down(while staying level) and tap the the top of the rails (with the bite paper sandwiched between). Then grind down where you see high spots marked on the impression paper. Repeat until you have uniform bite. (It may turn out that only one or two areas needs to be shimmed up).

I would think carbon-paper might also work.

I am not sure if more or less work that creating a second set of grinding rails.

Disclaimer: I have not tried this in the shop, however I did just have a broken tooth repaired........


11-14-2005, 05:23 AM
Lot's of good ideas to chew on......thanks!

I don't think that I am too worried about small bumps and wobbles, it only really worries me when the rollers don't sit down firmly onto the rails. The equi-distance between the rails is the thing that makes the wheels lift.

Also toying with only one rail having one v-top rail to give line while the other rail has a flat top to control level only. The flat-top rail could be set down lower and the gantry could have a taller leg on this side. Then it will be easier to load boards over the low rail. But, a tall leg gives the flex again, and so it could be a v-top rail, and so the argument goes in circles.....

(Brian, a discussion (http://groups.google.com/group/rec.crafts.metalworking/browse_thread/thread/c6a253a22004a4a2/f55d2e196dd577f0%23f55d2e196dd577f0?sa=X&oi=groups r&start=0&num=3) on blueing and scraping. It is a very old art and the patience of the folk amazes me!)

11-14-2005, 05:58 AM
I think that it is important to look at the hold down rollers here. SB's version is very crude and creates forces not(45deg) in the required direction (downwards). I think that some form of positive hold down is needed, especially if we are not dealing with an absolutely(?!***!?) precise rail arrangement. On my Y carriage I have tilted the hold down roller to 45 deg and run it along the bottom of the bright angle - makes it easy to adjust and pushing the right way. I like the idea of one v rail and one flat rail - but I also like the dust clearing properties of the V rail.

11-14-2005, 06:05 AM
rh, the hold-down system is an important issue, but it will side-track the rail straightness issue that I am trying to tackle in this thread. Let's assume that we are talking x-rails only......

11-14-2005, 06:09 AM
How about a v-groove in a solid bar of metal?
Are there machines out there which can v-groove metal with great accuracy both horizontally and vertically?

11-14-2005, 06:22 AM
Mike, a groove will get clogged up. A "bed way grinder" machine that can make long straight grooves will also be able to make the ^-tops. (as I have done for my shorter gantry, but the longest machine in Cape Town is limited to 3 meter and I need 3.4 for my x-rails)

(rh, here (http://www.talkshopbot.com/forum/cgi-bin/discus/show.cgi?tpc=312&post=14231#POST14231) was something about hold-down rollers)

11-14-2005, 08:39 AM
I think that adding more wheels is a good way to make a simple thing more complicated. The thing that I like about ShopBots is that you don't have to be a brain surgeon to operate and maintain one. (No offense intended, Ted!)

The machine has a very simple setup - two tracks - one for movement and one for guidance. To vector in another wheel, or another type of guidance system will make it harder to align and keep in alignment. What you will find yourself doing is spending more time trying to make the machine work than trying to figure out ways to make it profitable.

Now, having said that, improving what is there is a different story. Harder rails, or creating a way to grind the existing rails that takes a small amount of time is the way to go IMHO.

The wheels creeping up on the knife edge is a serious problem, but not insurmountable. One of the rails is bent. Figure out where the bend is and unbend it. It may mean taking the rail off , mounting one end in a vise and twisting it. Or it may mean heat and the application of heavy "manual instruction" (ie. the hammer).

One thing that I found that worked for me was to take a large metal file and using it to file the burrs off of the rails. That actually improved performance.

Making the rails equidistant means finding a calibrated rule of your own making or from the local metal shop and adjusting, bending or twisting.

It is what it is, no smoke and mirrors. The ShopBot is an inexpensive CNC machine originally designed for boat building. Everything else the thing makes is gravy, as far as I am concerned.

We are continually making improvemnts to our 'Bots. Just yesterday, Gene installed the blast deflectors and improved our dust collection by at least 90% - only a handful of sawdust was left on the table after carving a 12 X 12 piece of oak. The point is that we try to improve upon the existing design instead of re-inventing it and it saves us time and money.

My $0.02.



11-14-2005, 10:21 AM
WOW! Improved dust collection by 90%. Chip, how does the your blast deflectors work? Do you have any pictures of this? Mark

11-14-2005, 10:32 AM
Chip, the onus is on you to get this thread back on track.

11-14-2005, 10:36 AM
Gerald, try this site http://www.moglice.com/

I have not used this material, but have read that it works well.

One way to improve rail alignment is to attach the rail directly to the C channel, without the washers beneath the rails. This will give the rail better support and allow the use of shims at the edges (as opposed to the centre of the rail to correct height) of the rail to correct twist.

11-14-2005, 11:35 AM
OH! er...um...Right...

There is a new thread for the blast deflector thing. I'll be posting pics later today.

As for the rails... I don't know what else to say..