TIme for another alignment/calibration

Most of the jobs I’ve been doing on the X-Carve have been small < 300mm square or so and I’ve noticed no problems with level or alignment. The other day, I wanted to cut a much bigger piece, about 600mm square (enclosure for my 3d printer) so I added a bump stop by fastening down a square of Masonite and then cutting off the excess to make an L shaped corner piece. This is where I found out that the cut wasn’t at 90 °. Checking the spacing at the Y axis plates with the carriage fully forward showed that the gap was not equal - been there, done that, so a few tweaks and now it’s OK.

As it’s a wet afternoon and I’m stuck inside, rather than make sawdust, I thought I’d check Z axis level and travel in each axis (does it move 2.55mm when I tell it to move 2.55mm? sort of thing).
I found this thread by @JeremyJohnstone

so set up a dial gauge on a magnetic base to do the measurement. One problem is that X-Carve has no steel so no where to stick a magnet so I had to attach a small clamp to the router mount.

Procedure was fairly straight forward:

  • Home the system
  • Jog to the reference point on the bump stop and touch off in X and Y
  • Using the 123 block, jog down until the dial gauge was close to zero, adjust dial to suit, touch off in Z.
    Now it’s just a matter of moving Z to some positive value , say + 5mm, moving to the next XY coordinate you want to check and issuing a G0Z0 command (after moving the 123 block of course). Rinse and repeat.
    I can see that my spoil board is lowest in the top, left corner but pretty close at the front and across the middle. I’ll take a few more measurements when it’s a bit warmer. Hopefully I can fix this my resurfacing rather than any machine adjustment.

A few images of the setup and results can be found here. http://imgur.com/a/ooh2p

@RobertCanning You’re absolutely right. I’ve religiously checked v-wheels and belt tensions but apart from the initial setup and tuning, I’ve never really thought about ongoing calibration.

Now I have a procedure that only takes a few minutes, it will happen more frequently.

Thanks Phil, I’ll add that point to my pre-flight check list!

1 Like

Surfacing the bed with a large bit and very little stepover will not only help you level it out, it will show you if your spindle is out-of-square with the table. If you find any kind of a consistent sawtooth pattern in either axis (or both), you’re out of square and can easily fix it by either rotating or shimming.

Now if you find an inconsistent sawtooth pattern, it may indicate that your X-Carve has a lot of flexibility and you’re trying to go too fast or too deep with the large bit. It’s not much of a diagnostic indication, but it’s good to know.

Using the largest bit you reasonably can for checking out-of-square allows you to make a more precise adjustment since a sawtooth with a 3/4" bit will be magnified a whole lot more than using a 1/8" bit.

Ran a surface cut with a 22mm bit yesterday. Some slight sawtooth evident in the direction I did the cut (X). I’m planning to do the 30 minute stiffening mod today then I’ll run another surfacing pass to check. Still got a few more passes left before I have to replace the spoil board.

1 Like

You can always run a flattening pass on a piece of scrap before committing to your wasteboard, just make sure it’s clamped securely and it’s large enough to give you a good long cut, maybe 8"x8" or so.

If you want to see how I read the sawtooth, it’s at 9:32 in my Secondary Wasteboard video: https://youtu.be/5NKmmMKNpOE?t=9m32s

Looks like I’ll have to go through the whole process again tomorrow. I made the bits for the 30minute X mod today and thought I’d just need to loosen the Makerslide to get the bolts in. As I’m using a steel bar with threaded holes instead of separate nuts, I’ll need to disconnect Makerslide at one end to slide the steel bar into position on the bottom of the Makerslide.

Top piece is 12mm aluminium bar, drilled and countersunk for 8xM5 bolts. Lower part is 32mm mild steel bar drilled and tapped for the M5 bolts. A bit of a cleanup (maybe paint the steel) before installation tomorrow.


Update: Make that the 60 minute mod!


Went through the calibration process again today - I’m getting quicker at it!

Surfaced the spoil board (0.1mm only), still some sawtooth evident so a bit of shimming needed.
I tried re-cutting the t-square bump stop but it’s still not exactly at 90°, not sure why. The distance from the front of the Y carriages to the Y support plates is the same. As far as I can measure, the X axis Makerslides are at 90° to the Y axis motor plates, I did have to shim one end to get it so.

Repeatability of position is good, if I jog to extremes of X and Y and then tell it to go to X0Y0, it does within 0.1mm.

More fiddling with adjustments tomorrow.

I managed to surface the spoil board, no noticable sawtooth in either X or Y directions. A post surfacing level check shows that everythng is level in Z - that’s the good news.

Flushed with success, I screwed a new piece of Masonite down and ran my gcode to create a bump stop. This just cuts down 6mm to make an L shaped corner piece.
The bad news is that it’s not cutting at 90 degrees but a bit over.
I’ve checked:

  • belt tension
  • v-wheel tension
  • that both ends of the X gantry are equal distance from the Y axis plates when the gantry is fully forward.
  • that the X axis makerslide is at 90 degrees to the Y plates (I did have to shim this to make it so, it was done before I made the bump stop.

The only other thing I can think of is to change the wiring for the Y axis motors in case one is moving at a different rate to the other - faulty driver/motor.
In case anyone suggests adjusting the pots, I don’t have any. It’s running a Gecko G540 controller off Linuxcnc

1 Like

Hmmm. Gonna spitball a couple questions and ideas here.

How did you check belt tension? Is it possible that one Y belt is stretched more than the other? Have you inspected the belts or pulleys looking for damaged teeth?

If your 90 degree cut is more, you may try shutting down the X-Carve and manually pushing the right side back a little while holding the left side as is. If that fixes the problem, you may want to consider making a set of marks or bump stops to align the gantry by hand before you power the motors.

Did you check the diagonal measurements on your Y axis? The two may be just enough out of parallel that it’s causing a problem like this.

@RobertA_Rieke Belt tension checked with your method via a small scale - not sure exactly what the tension was As I it was cutting OK (no missed steps, circles were circular, etc) , I was more concerned with checking the the tension was the same.

I’ve been pulling the gantry forward to a set distance from the Y plates, a small piece of 25mm aluminium bar is used as a gauge. Setting an offset is a good idea to get around the problem - if I can’t find the cause.

I did check that the diagonal measurements were the same when I fastened everything down to the torsion box. That was a while ago so probably worth a check.

Swapping over the drivers made no difference, swapping the motors is a bigger task that I’ll tackle as a last resort.

How are the two Y axis motors wired regarding the step pulse signal? Are they wired independently, or does the same wire drive the step pulse for both motors?

@LarryM As I mentioned, I’m using Linuxcnc/Gecko G540. The two motors are slaved in software with one configured to rotate in the opposite direction to the other.
Step and direction pulses are separate and exit the parallel port on different pins. No common wiring between the two Y motors.

The problem I have in swapping the motors over is that my motors have 5 metre wires connected so there’s no terminal block near each motor. To swap them, I have to desolder the connector to the controller (DA9 style) and pull the wires out of the cable chain, rethread, re-solder etc. As I say, last resort :grinning:

I’m very interested in what you find out here. There is a potential problem using two motors with independent signals driving the same mechanics. Signal timing becomes critical. You happen to have an oscilloscope?

Given the number of Gecko G540/Linuxcnc users out there running similar configuration machines, I don’t think timing is an issue.

I’ve just run a quick check of the diagonals and it’s about 2-3mm out of square - in the direction that would cause this problem.
My trigonometry/geometry is not up to working out if this translates to the error I’m seeing so I’l just fix it instead …

1 Like

A 3mm diagonal difference with 1000mm sides equates to an angle of 90.12 degrees with the origin at your left front position of your machine. If this matches then the diagonal from the left back of the machine to the right front of the machine should be the longer diagonal.

Doesn’t match up with the drawing in your earlier post as it shows a deviation on the left side of -0.3mm. If both sets of measurements are taken without changes to the machine that would put your measurement of the deviation at about 286mm along the Y axis, but your drawing shows the deviation at somewhere near 800mm, assuming that the drawing is somewhat to scale.

My drawing as far as the X and Y measurements are concerned, may not be that accurate. I’m not sure that the method I use to measure this was valid. I’ll have a think about the best (correct) way to do this tonight and try it out tomorrow. I didn’t get chance to correct the diagonal error today, too nice to be indoors.

. It’s Spring! so I spent some time getting the seed raising production started. In the hothouse at the moment as we’ll still get frosts up to late October.

Problem solved!

As previously mentioned, when the gantry was fully forward, the distance between the Y plates was not equal. Following advice given, I was pulling the gantry square before turning on the motors so it would hold this position.
As the right side gap was larger than the left and this was the same direction that my squaring cut was out, did the cut once again, this time without pulling the gantry square. The cut is now 90°, or as close as I can measure it. A framing square sits in the bump stop without any play at all. :relaxed:

1 Like

Tried that but seems its not helping me… :disappointed_relieved: