Page 1 of 2

Shark HD4 extended isn't cutting square...

Posted: Sun Jun 10, 2018 5:03 pm
by JerkMcLaughlin
When I cut a rectangle profile it does not come out square. I believe that the gantry is not square to the Y axis. This problem also showed itself when I drilled
four symmetrical reference holes for a two sided part; when i went to flip the part the holes did not line up correctly.
I ran a 3,4,5 square test and it revealed that it is cutting about 1/16" or so out of square on a 15"x20" rectangle (diagonal length should be 25" but on one side it's 24 15/16" and 25 1/16" on the other).
While inspecting the gantry I noticed that the 3 horizontal set screws on the right end of the gantry(where the gantry rides the rail) were very loose and the ones on the left end were just snug tight.

Does anyone know the correct procedure for squaring the gantry(x axis) to the y axis?

Any help would be appreciated.

Thanks!

Re: Shark HD4 extended isn't cutting square...

Posted: Sun Jun 10, 2018 6:14 pm
by Rando
JerkMcLaughlin:

Yikes! Sounds like a huge pain.

The under-bed slide rails are quite maneuverable, and the position of them is much more dependent on the rigidity of the cross-arm that goes under the bed. The gantry arm ships in a separate box, so because those are the bolts that have to be assembled, I suppose if they're gonna be loose anywhere, that's where it would be. ;)

Since I've removed and replaced the gantry on my machine several times, I can tell you to mostly not worry, but treat it gently. If feasible, remove the router to lower the weight.

Look underneath the bed at the cross-bar, and manually move it back and forth...or, try to. It should be quite rigid, but not infinitely so. If not, there could be bolts on the ends of the rail that are loose, or heaven-forbid, the center drive connection is loose :(. That means at least part of the bed comes off, unless you have very very thin arms or long tools...I'd still take it off if it looks actually truly loose under there. But, that comes pre-assembled from the factory, so I'm going to assume it's torqued better than I would have done.

Before tightening the six-per-side bolts, decide which set you want to be in. When the gantry is lower, it tends to give more accurate cuts and better surface finishes from the improved rigidity. But, that comes at a cost for Z-axis height. If you use MDF, of make thick parts, or like me use a Kurt vise, the top set of holes is where you want to be, sometimes even with the added flex from the effectively-longer gantry arm.

After you've decided which set of holes to use, lift each side into place and tighten the nuts until they're just barely tight. Then, like they do on cars, tighten the six in a star pattern or one form or another. Progressively. As you go, give the gantry arm a good tug to make sure it's properly seated. Then do the other side, and you're done.

To test if it's still not true, I recommend using a nine-hole pattern. That is, 4 on an outside square, then 4 on an inside square aligned to the outside one. The final hole is in the middle. Set your zero at part-center, then cut the pattern. If feasible, make it ~10-12" on a side for the outside. The reason you need to have the two squares is that will tell you if the problem is in the workpiece registration / alignment, or in the actual machine movement. Once you get the gantry tightened, all that machine error should go away. Hopefully.... :shock:

Hope that helps.

Regards,

Thom

Re: Shark HD4 extended isn't cutting square...

Posted: Sun Jun 10, 2018 7:07 pm
by JerkMcLaughlin
Thanks for the reply Thom.

The set screws that were loose are not the 6 that hold the gantry to the crossbar. It's the set screws that are on the slide rail car. Here's a pic of them:
Loose set screws
Loose set screws
It's the black set screws in the aluminum piece.

I have since tightened them and ran another test. No luck. Still cuts out of square. Here's a pic of the results I'm getting(left is as programmed, right is how it ends up cutting, exagerated for clarity):
Not square results
Not square results
As far as I an tell the crossbar under the table may not be aligned square(perpindicular to Y).

Re: Shark HD4 extended isn't cutting square...

Posted: Sun Jun 10, 2018 7:38 pm
by Rando
JerkMcLaughlin wrote:Thanks for the reply Thom.

The set screws that were loose are not the 6 that hold the gantry to the crossbar. It's the set screws that are on the slide rail car. Here's a pic of them:
IMG_2235.jpg
It's the black set screws in the aluminum piece.

I have since tightened them and ran another test. No luck. Still cuts out of square. Here's a pic of the results I'm getting(left is as programmed, right is how it ends up cutting, exagerated for clarity):
Not Square.jpg
As far as I an tell the crossbar under the table may not be aligned square(perpindicular to Y).
Those set screws are the pillow-block bearing retaining screws. Those being loose should not significantly affect movement unless they are preventing free movement of the gantry.

Can you run the test I discussed and measure the results with a ruler? An actual photograph showing the cuts and the error, while the piece is still on the bed, would be helpful.

Re: Shark HD4 extended isn't cutting square...

Posted: Sun Jun 10, 2018 8:09 pm
by JerkMcLaughlin
Thom,
9-hole test.jpg
Is this what you mean?

Re: Shark HD4 extended isn't cutting square...

Posted: Sun Jun 10, 2018 8:59 pm
by JerkMcLaughlin
I ran that test. The piece is still on the table, havent moved it.
IMG_2240.jpg
IMG_2237.jpg
IMG_2239.jpg

Re: Shark HD4 extended isn't cutting square...

Posted: Sun Jun 10, 2018 10:10 pm
by Rando
Okay, so are you saying those dimensions are a success? or a failure?

Offhand, I'd say that measuring 1/16" of an inch difference on 1/4" holes could be within normal tolerances in wood. Try a 1/32" bit if you want to measure 1/16" differences.

Also, the Sharks are notorious for the z-axis moving during plunges and hard cuts. So, if you watch carefully, you might see that the bit moves when it first touches.
With a 1/4" bit that long, it's entirely possible it's 1/16 off. In tests like these, plunge very slowly and at high rpms, and make the slightest impression you can still
measure against. You want as little torque on the router as is possible.

But, I'm honestly not seeing a significant parallelogram effect you were saying. If the diagonals go through the center holes, and are at right angles, then it would
seem the machine is square. For you to have a situation where the gantry moved closer to one side of the bed as it moves in your Y-direction, well you'd be able to jog and measure that with calipers. But, a LOT would have had to be out of tolerance for that to happen; just the fact that the bed members are so long, even if they are out of position length-wise, that's not going to give an angular offset. It would almost require some significant malformation of the table surface it's on, and it doesn't sound like that's the case.

The other thing to do is a sanity check on the GCode, just to convince yourself it's putting the holes where we think they're supposed to go. Also, kill virtual-Z, don't use it in these tests.

I suppose it's possible if it was damaged in freight, that the under-bed cross-member is mis-aligned, but that would be pretty unexpected. Is the machine new?

Also, one thing to remember: if you align your material with the travel of the actual cutting head, even if it's not exactly square to the t-slots, then your cuts will be square to the motion of the machine. I always do that tramming (to tram) of the vise, with a fold-down thousandths dial indicator, so the work is always aligned precisely to the movement of the machine. I guess I can honestly say I've never checked whether that was actually in alignment with the t-slots...hmmm.... :?

I wonder if the precise alignment of the MDF slabs in the t-slots could be causing the issue? Say if the MDF slabs at the y-negative end are pushed against the X-positive side of the t-slots, while the slabs at the Y-positive end of the bed are pulled to the X-negative side of the slots? Sounds crazy! :roll:

Thom

Re: Shark HD4 extended isn't cutting square...

Posted: Sun Jun 10, 2018 11:55 pm
by JerkMcLaughlin
Am I being too critical? The fact is I can’t machine two sided parts and I can’t cut a rectangle with 90deg corners. If I wanted to cut the parts for a simple jewelry box I couldn’t because they wouldn’t be square. Nothing would fit. Is 1/16” out of square in a 12” span acceptable? In my experience 1/16” out wouldn’t be acceptable on a 96x288. I would think that cutting a 12x12 square would be fully within the limitations of a $5000 cnc machine. Please correct me if I’m wrong.

Re: Shark HD4 extended isn't cutting square...

Posted: Mon Jun 11, 2018 12:02 am
by JerkMcLaughlin
Also, the mdf slats have been milled “flat” using a pocket hole program and a 3/4” end mill, 02” pass depth so I don’t use virtual Z.
I’ve had the machine for about two weeks.

Re: Shark HD4 extended isn't cutting square...

Posted: Mon Jun 11, 2018 8:18 am
by Rando
Nobody's saying you shouldn't expect accuracy; I expect mine to be within 0.001"
I'm saying using 1/4" holes to measure to 1/32" accuracy (required to establish
a 1/16" measurement) isn't actually an accurate measurement, given the amount
of flex in the shark mechanisms. Without knowing for sure whether the differences
are real or a measurement artifact, all diagnostic attempts amount to grasping for
straws.

If you want accurate measurements, use a spot drill and only make an 0.20-0.040
deep spot, and measure to the center-point of the spot.

The reality is that the ONLY way you could get x-axis offset relative to the bed
length is if the gantry is misaligned. Jog it where it just hits one extreme, and
measure whether there's any difference in position on the two sides, as measured
from the end. If there is no difference, then it's not misaligned, and the problem
is elsewhere...like in the measurement method. If it is, then align it.

Thom