X rail not level

(Will James) #1

Hi all

I’ve just finished installing my SO3 onto a new stand and have squared and levelled each corner of the Y rails to within 0.05mm of each other in height. However, when I’ve then tried to square the X rail to the bed and level it, I find that the left-hand end is nearly 3mm higher than the right. Any suggestions please on how to rectify this before I dismantle the whole thing to make the holes in the plates wide enough to accommodate such adjustment?

Thanks in advance


1 Like

(William Adams) #2

The possible culprits are:

  • hole positions / powder coat on the end plates — sometimes they’ll be off in opposite directions on pairs of plates, and compounding that (of course) the powder coat will be thicker at the top of one hole, and bottom of the other
  • extrusion not cut squarely
  • V wheels are different diameters — this hasn’t been as much of an issue for a long while, mentioned only for completeness’ sake

Most likely it’s a perfect storm of each thing adding up — time was we’d suggest flipping the extrusion, but if you have the nifty decal, that’s not an option. Things to try:

  • clear the powder coat out of the (EDIT: top) holes (don’t enlarge them into the metal)
  • shim the ends of the extrusion
  • file / machine the ends of the extrusion square — this is easiest if one pulls all the rails, stacks them, and uses a large fine file to draw file them after marking the high points w/ a marker
  • swap the V wheels around

See https://www.shapeoko.com/wiki/index.php/Calibration_and_Squaring_the_Machine#Squaring_the_Machine


(Will James) #3

Thanks for such a quick and complete response. I’ll report back when I’ve tried all of this.


(Richard Cournoyer) #4

As a LONG time user, here is what I recommend:

  1. Please don’t assume that the bed is level and a good place to reference off (it’s not).
  2. The X rail should be level to the Y rail…THEN level the bed to the Y rails (washers/milling it flat)

(Will James) #5

I take your point, but surely the ideal is to have them all level with one another? Anyway, Y rails were parallel to the bed before addressing the X rail issue, so solution needs to be within the X rail. Removed the excess powder coating from the holes and brought the height difference down from 2.9mm down to 0.4mm differential. Much better but still too much out of square for my purposes. Could someone explain how shimming or filing the ends of the extrusions would have an impact on the height of the respective rail?


(William Adams) #6

Shimming or filing the ends will adjust it angularly up / down — but if you’re as close as you say, if you haven’t yet, my inclination would be to remove the powder coat from the holes for the extrusion on the high side if you haven’t already.


(Dan Nelson) #7

Have to agree with @RichCournoyer here, the MDF bed is likely the least flat of the whole machine unless you’ve done some work in making it flat. Y rails can be flat to the MDF, but the MDF could still sag in the middle (worse on an XXL, but I think you said regular SO3?). I’d first check to see if the rails are square, with a known good square (like machinist square). Also check as @WillAdams mentioned that the extrusions are cut square. 3mm (1/8") is a LOT to be off side to side on a small machine like the SO3, I’d check a few things before I’d start grinding or shimming (other than super powder coated holes like Will mentioned). Also make sure the TOP V wheels are tight, this will induce a bunch of unsquareness. Measuring and checking is reversible, cutting and grinding is not. Curious to see what the final findings will be, let us know!


1 Like

(mikep) #8

At .4mm, you’re well within what you should expect -that’s about 15 thousandths of an inch. To take the rest out, surface the table.

Here are my tips on how to get the axis sorted:
You -must- use a good level. I was amazed how bad my torpedo level (about ten years old, $100-when-new) was when I checked it. Put your level on something you think is “mostly level.” Turn it 180 degrees. It should read exactly the same. If it doesn’t, it’s out of adjustment. Many levels cannot be adjusted. If the level doesn’t read the same turned 180 degrees, don’t use it, you’re just going to make yourself nuts doing this.

A machinist level is the perfect tool for this.

Do be aware that a level is only accurate if it is vertical. If it is tipped to one side (sitting on the ridge on the rail, for example) it will give you a false reading and you’ll chase it all over without getting things any better.

If you run out of adjustment, add blocks under feet as necessary. My machine is on a table made from an old countertop in my garage. The garage floor slopes very strongly toward the door. It isn’t obvious how dramatic this is until you actually level something up. I have over an inch of blocking under the “downhill” end of my machine to bring it level.

You can get a machinist level pretty inexpensively on ebay (they can be adjusted) - a Starret 98-6 is a good place to start (98-12 is 12 inches, 98-6 is 6 inches). Used is fine, it just needs an unbroken vial (they can be replaced if you find a great deal though). I wouldn’t suggest a “master precision level” - they work GREAT for this, but good ones are very expensive and pretty limited use.

Here’s the process (it’s iterative). The trick here is to only move 3 feet, never the 4th (here, y2, rear). Do not attempt to level the wasteboard itself. It’s a waste of effort at this point.

  1. Level a Y rail, front to back. (Call this Y1)
  2. level the other y rail, front to back (call this Y2). If the world was perfect, you now only need to get the two Y rails level with each other now. It’s never that easy.
  3. Move the gantry all the way to the back of the machine.
  4. Level the gantry left and right by adjusting A back foot - for the sake of argument, the rear Y1 foot. If you adjust more than one of the back feet, you need to start over. That’s a hint. Don’t do what Donny Don’t does. Adjust one and only one foot :slight_smile:
  5. Put a level on Y1, front to back, it probably isn’t level. Level it, adjusting the front foot only.
  6. Put the level on Y2, front to back, it should still be level, but probably isn’t (you didn’t move the back foot or front foot…). If it isn’t, adjust the front foot only.
  7. Move the gantry all the way forward
  8. check the gantry for level left to right. It probably won’t be level, but it will be closer than last time. Adjust the Y1 front foot a touch, and it should be closer. Now go to step 1…and don’t touch the rear foot on Y2, and keep cycling through all this until it’s “close enough.”

Don’t be surprised if this takes an hour or more to do. I think I did 4 or 5 iterations before I was happy (I just did this yesterday) and it took quite a while. Be patient. Once you get close, it’s a little fiddly. You will almost certainly now be out of square with the table. Because…table. If you put a level on the table, you will almost certainly get readings that imply it’s twisted or bent. Let it go. It doesn’t matter. The table is not a good reference at this point. Now surface the table. After you surface the table, it will be parallel in all directions to the two axis. If the axis are both still level, the table will now be level. Pretty close, anyway. It isn’t super strong though, so you can make it deflect a few thousandths with pressure from your hand very easily. The first time i did this, it took a 20-30 thou cut (maybe more?) to get the table cleaned up.

At this point, check the tram on the Z axis. It’s probably off a little. left and right is easy to adjust with the eccentric cams, but front to rear is harder and requires loosening the screws holding the rail to the endplates and twisting the whole thing. Then, go and check everything for level again.

Now you’re ready to calibrate for belt stretch. That’s a whole can of worms in itself, but at least doesn’t require mechanically moving anything in the machine.

Be reasonable about your expectations though. My XL is off less than 2 thou front to back, and about 3 left to right. When I measure variation front to back and left to right between the table and the router (test indicator mounted to the router clamp) I measure something less than 1 thou variation with no weight on the table. It could probably be a little better, but much better would require blocking under the table, or maybe a torsion box.

If you move the machine, start over.

I’m no expert, but that’s my take. I’ve done this a few times now, and I’m happy with the result. There are probably better ways, but this works for me.


Squaring S3 XXL
Leveling Feet removal