Squaring the Y car to the X axis

[sagreen83]

Ok,

I've had my Alpha for a couple of months now. Everything was great until the other day I noticed that the Y car was not square to the X rails. To prove this I put 4 small holes (with a vbit) in the table 70 inches apart on the X axis, and 40 inches apart on the Y axis. Basically the 4 points of a square. After which, I measured from opposite corners in the square (triangulation) and found one of the measurements to be out by 1/8". So, over 40" my shopbot was out of square by 1/8".

I called support, and Brian told me how to loosen and square the carriage. Basically you loosen up the Y car so its kinda "floppy" after clamping down one of the sides to the X-rails so it doesnt move. Then you move the other side 1/2 the distance its out of square (in my case 1/16") to get it back into square. If you were using an older PRT, you could unhook one of the X motors, and let the bot move the otherside the amount you needed. Easy enough...

Roughly... this worked. Using the crude tape measure method, it squared up, and using a small square the resulting cuts are square. Now... I'm a total stickler for accuracy, and using a tape measure is ok, but in my book, no where near the accuracy that this machine is rated to perform at. One good idea that he gave me was to adjust the mechanical stops so that when square they both contact the pinions so if it goes out again all you have to do is run it to the stop on one side, then loosen and adjust the other side to the stop.

Does anyone have a better method than doing the tape measure method? A square would work if you could find someplace to square it, but no way it would give you accuracy over the full 48" travel of the Y car. I want dead perfect accuracy in squareness of this machine, and this method is not going to get it.

Thanx,
Scott...

[slendon]

Scott,

I'd be concerned to find out what changed to make the machine out of square, before I spent too much time readjusting.

I would also like my machine to be bang on square, so will watch this thread with interest.

Steve

[richards]

I fought the same problem with my Alpha. Somehow I just expected the machine to stay perfectly aligned. But, it didn't.

Now, I use a very simple method. I wrote a little program that positions the machine prior to turning it off (my 0,0 point is about 19 inches from the end of the X-axis rails - 120-60 machine used basically as a 96x60 machine). I bought two combination squares that are only used to check positioning. One combination square is set for the X-axis and one comgination square is set for the Y-axis. Every time I turn the machine on, I check that the Y-carriage is exactly the same distance from the end of each X-axis rail.I use a feeler guage to check the amount of error. If there is more than 0.010 inch error, I turn the power off, put a wooden clamp on both sides of the Y-carrage on one X-rail and then nudge the other end of the Y-carriage, turn the power back on and re-measure. So far, unless the machine is bumped with the power off, everything stays square. (A simple jig could be built with a dial indicator to replace the combination square and feeler guage.)

Mike

[fleinbach]

I find if you want something as square as possible using the trusty old A Squared + B Squared = C Squared you should make the A & B legs as long as possible and as close to equal as possible. Next when measuring use the hash marks on your tape to get as fine a line as possible for maximum accuracy.

I have a Prtalpha120 so the first thing I did was shaved off the edge of the X axis to make it absolutely parcel to the X rail. Next I made a mark about 1/4 inch from the "0" end of the shaved edge. Then I made a mark 60" down the X axis. This works best with 2 people. One person holds the tape on the 1" hash mark and the other makes a mark with a very sharp pencil at the 61" Hash mark.

Now one person lines up the 1" hash mark on the X axis "0" mark while the other person makes a small arc at the 61" mark down the Y axis. Now measure diagonal 84 7/8" and mark the 60" Arc made previously. The next step is to clamp a very accurate straight edge lining it up on the 2 marks made on the Y axis. Now as Scott mentioned above loosen all the bolts on both ends of the Y carriage. There are a total of 24 bolts to loosen, 4 on each end and 8 under the end rails. This should create excess free play in the Y carriage. Next with the shopbot turned off move the side of the bit in the router against the straight edge near the X axis. Clamp the Y carriage to the X rail closest to the router. Now push the router down the Y rail again holding it against the straight edge. Clamp this side to the other X rail. Now tighten all bolts. Don't remove the clamps yet. Hand push the router back and forth down the straight edge to make sure it is tracking properly. At this point you should be tracking fairly square.

Now the next step will assure you keep tracking square all the time. With the clamps still attached measure the approximate distance from the X motor gear to the mechanical stop. Cut a piece of hard wood to fit the space. It does not need to fit perfectly, just close, you will see why in the next step. Now loosen the mechanical stop, then hold the spacer against the gear and slide the stop against it. Tighten the mechanical stop. Now repeat for the other X axis. Once this is done you have made the stops for the Y carriage perfectly square to the X axis.

You need to do this because when the tool is turned off there is a possibility the Y carriage may relax slightly out of square. But even if that happens all you need to do is follow this procedure. Pull the Y carriage against the Mechanical stops and while holding against both stops (which are square to the X axis) turn on the shopbot control box energizing the motors. They are now in sync and square since they where energized while the Y carriage was being held square against the stop blocks.

[tim_in_mn]

I screwed up my PRT 96 a couple times, and came up with a solution to square my carriage, without cutting or drilling the table.

The theory is base on a 3, 4, 5 triangle.

First I made up a drawing of a 36" Y leg, and a 48" X leg triangle. I then made a 1" line at 90 degrees to the 60" ends of the triangle. Then I removed all but 2 inches on the X and the Y end of the triangle to save cutting. Then I created a cutting file to cut just to the table top on the X & Y cuts, and 0.2 inch deep on the 60" side and the 90 degree angle.

Clamp a scrap piece of wood on the edge of the table centered at 36"Y and 48"Z. Make the cuts, and measure the distance between the X and Y points 90 degree cuts and you should read 60". If not, I turned off the Shopbot clamped one end of the carriage and moved the other the distance of the error. rerun program and check again until accurate.

I do not have enough time to go into more detail, But I will have the cut file available if you need it. I will be gone for the next two weeks. If you have questions, let me know. Attached is a picture of the cut made at the 36"Y end. Hope this is clear enough, and is of help to you.

Tim


3281

[ron brown]

On my machine I squared using a compass (tramel beam) and intersecting arcs. I then scribed across the X rails. I transferred the measurments to the end where I could visually check "square" when "X" was "Zero". The X register device is on the "far side" of the table and the "site" is easy to check. If it is out, I will "MO" - Motor off - bump carriage, then re-zero and recheck - repeat if necessary.

I've considered mounting a cheap dial indicator to one or both stops where the deviation could be "read" but don't see much benifit with the work I do.

Ron

[gerald_d]

Back to basics....

How do we check an old-fashioned "square (http://images.meredith.com/bhg/images/products/m_wp_00238.jpg)" for square? We hold it against an edge, draw a line, flip it over, draw another line, error equals half the angle between the lines. This is not rocket science.

To check a SB, stack two thin sheets of ply on the table, trim them square together (with a through cut), flip one sheet over, and check the "match" with its partner below. Half of the mismatch, is the amount out of square. Also not rocket science.

The method we use to quickly detect and correct any error has been explained a number of times and I havn't got the energy to do it again.

[sagreen83]

Gerald,

You dont have to have the energy... Frank explained perfectly above. Frank, as always thanks for your help, and to everyone else as well.

Scott...

Scott...

[fleinbach]

I must admit I had actualy intended on including Geralds explination, in fact I had it typed out but ran out of time before I could explain the potential problem in doing this. Yes it is absolutly the best way to verifiy square. We have been doing it for years when framing additions to verify that our framing sguare was truely square. And as Gerold described cutting a sheet in half and flipping it will allow you to see how far out of square you are. The only thing I need to add to that is when making the cut you need to cut very slowly and don't take to much of a bite for maximum acuracy. If you take too large a bite, if you do the cutting forces will pull you off course slightly giving a less then acurate reading.

[fleinbach]

I tried to edit my post but was to late so I am adding the edit here.

I must admit I had actualy intended on including Geralds explination, in fact I had it typed out but ran out of time before I could explain the potential problem in doing this. Yes it is absolutly the best way to verifiy square.
We have been doing it for years when framing additions to verify that our framing sguare was truely square. And as Gerald described cutting a sheet in half and flipping it will allow you to see how far out of square you are. If everything is square the sheets will meet evenly there entire leght. If things are out of square one end will touch and the other end will have a gap. You will be out of square by one half the gap.

The only thing I need to add to that is when making the cut you need to cut very slowly and don't take to much of a bite for maximum acuracy. If you take too large a bite, or cut to fast, the cutting forces will pull you off course slightly giving a less then acurate reading.

[gerald_d]

Frank, I did not mean to cut the sheet in half - I meant 2 full-size sheets stacked one on top of the other. But yeah, cutting two "identical" half-sheets side-by-side and then flipping one over on to the other will also tell you the story of inaccuracies - including whether the rails are straight or bowed.

With full-sized (8'x4') sheets, the idea is to get the flipped sheet to match bottom sheet along the whole length of each of the edges. Flip the sheet around the y-axis as well as around the x-axis.

Some old threads:
Thread 1 (http://www.talkshopbot.com/forum/messages/26/558.html)
Thread 2 (http://www.talkshopbot.com/forum/messages/7/522.html)

[fleinbach]

Gerald,

You are right both methods will work just as well.

[Mike Anettts (Unregistered Guest)]

Hello Scott

What we did was to simply make a pass along the x axis about 1/16 of an inch deep and then align a full sheet of square plywood along the groove. (we checked it with the diaginol method) we then put a bit in he collet and using a feeler gauge put it alongside the end of the plywood near the front left corner. we then ran the router along the y axis and adjusted the carriage until the same distace between he bit and the edge of the wood was obtained. Once set we drilled four holes into the table at the corners and checked the diagonals. we then drilled holes and installed fairly thick self tapping screws into the carriage pieces to lock it together along with the bolts. I also took the advice of the forum and put bolts at the end of the x axis to "recalibrate if nessasary. that was a few years ago and it has never been out since. I don' know if there is a more accurate way of testing than doing the diagonols at the greatest distance possible.

Just my opinion
Mike

[sagreen83]

Mike, thanx for your comments...

Basically your full sheet of square plywood works as a huge square along your X axis.

Thanx,
Scott...

[cnc_works]

My bot is cutting a parallelogram about .010" out on the diagonal in a 12". I'm thinking of putting up the stop blocks to hold against while turning on the cnc as has been suggested.

Is there any reason that drilling 1/16" holes in my spoilboard at the extreme four corners of the table in the shape of a rectangle and using the diagonal measure method to check for square would not work as good as any method to check for square?

Donn

[gerald_d]

Donn, I have never had much luck with two people holding a tape measure trying to measure the distance between round holes. But, if you make dowels stick up out of about 1/4" holes, and use a batten/rod between the dowels with a feeler gauge, you are going to be as accurate as any other method.

However, you need to drill new holes after each adjustment, and your helper at the other end of the batten/rod/tape measure could be getting confused as to which hole.....

[cnc_works]

Well, I tried my method by drilling 1/32" holes in the four corners of my (hopefully) rectangle. Found it to be out by about .060". Clamped one side of the car, loosened bolts, shifted by a guestimate, tightened, drilled again (after x'ing out the previous holes - thanks, Gerald) and got about .015" out in my 4' X 8' rectangle.

I tend to do fairly precise work, but generally on a small scale, so I suspect I can quit having achieved that level of precision. Anybody think that it would be fruitful to pursue closer tolerances?

Donn

[Brady Watson]

Donn,
It is always to your benefit to get the machine zeroed out as much as possible...However, I suppose it all depends on what you are cutting/milling. Generally speaking, if you can get it within 1/64", you won't see a difference unless you are milling very small parts.

-Brady

[fleinbach]

Edited. Posted in wrong place