Greetings, I am a relatively new CNC user with an open builds 1515. I'm designing and preparing my gcode files in Fusion 360 and have been using this since January of this year. I've generally figured out cutting out templates and can get consistent results, though with occasionally excessive cuts into my spoilboard. I've mostly attributed that to learning the various heights issues. I'm for the first time attempting to cut plywood for some small cabinets. This requires some pockets to be cut. The x and y dimensions of my cuts are coming out perfectly, however the depth of my pockets is coming out too deep. In this case, I'm cutting 0.71 in plywood with 0.355 in deep pockets. Everything looks great in the Fusion 360 simulations, but when I actually run the job, the pockets come out at a 0.42 depth Attached is the gcode file I created to test this with a few different toolpath settings. All 4 pockets came out at a depth of 0.40 to 0.42. I'm wondering if this might be a Z axis calibration issue? Or if something else is going on. Would love any suggestions for resolving this as I've been blindly trying different things for several days now. Thank you, Russ For reference, these are the parameters I was testing with the 4 pockets in this file: CNC Testing Setup Origin Stock Point Test 1 - Cutting Feedrate - 120 Geometry - Bottom edge of pocket Bottom height - selected contours Multiple Depths: Maximum roughing step-down - 0.125 Stock to Leave - unchecked Feed Optimization: Reduced Feedrate - 80 Test 2 - Cutting Feedrate - 120 Geometry - Bottom edge of pocket Bottom height - selection - Selected bottom pocket face Multiple Depths: Maximum roughing step-down - 0.125 Stock to Leave - unchecked Feed Optimization: Reduced Feedrate - 80 Test 3 - Cutting Feedrate - 120 Geometry - Bottom face of pocket Bottom height - selected contours Multiple Depths: Maximum roughing step-down - 0.125 Stock to Leave - unchecked Feed Optimization: Reduced Feedrate - 80 Test 4 - Cutting Feedrate - 120 Geometry - Bottom face of pocket Bottom height - selection - Selected bottom pocket face Multiple Depths: Maximum roughing step-down - 0.125 Stock to Leave - unchecked Feed Optimization: Reduced Feedrate - 80
Hah! I actually did the calibration during setup back in February of this year. But, good thinking to try running it again. I'll do that and update here.
If it was calibrated before it might be mechanical (calibration doesn't drift, so unless you changed Grbl Settings afterwards is should be what it was). Check the Grubscrews on the Z axis shaft coupler too
I went back and re-watched the build video for the z axis assembly, and in looking at that, I believe one of the lock collars has come loose. I remember seeing some vertical play in the screw when moving the z axis height up and down as I was diagnosing the issue. I'll try that tomorrow and let you know if it resolves everything.
Update... I reran the z axis calibration in the wizards and it was actually spot on 50 mm. So no change in calibration. I loosened and then retightened the lock collars and the motor connector for the z axis. It's rock solid now with no play at all. I then reran the test cut file attached, which should make pockets that are .355" deep (1/2 a sheet of plywood). The pockets still came out at .41" deep. Which is almost identical to where we started. I did come up with a wonky work around that seemed to work. In the Fusion 360 setup, I changed the bottom height offset from 0 to 0.055 and that did successfully move the pocket height up to what I wanted. I'd rather not have to offset the height every time though if it's possible to fix the root problem. Any suggestions for other things to try?
Are you using an upcut bit? Are the pockets consistently the same amount too deep over their whole area? 1.4 mm is a bit much to be accounted for by the bit pulling itself into the workpiece (or pulling the workpiece up) but it could be a contributory factor. Are you using a palm router or a spindle - if the latter is it heavy? Whereabouts on the machine bed are you carrying out these tests - try in one corner of the machine where you have the greatest rigidity. Could you upload your Fusion file? - preferably the version before you changed the bottom offset? - I won't be able to look at until tomorrow now though (nearly bedtime here in the UK). Alex.
That's what that GCode file should do - the maximum Z depth is 0.355" The obvious question is whether you are referencing the same Z0 for both the machine and F360? (Looks like you are using the surface of the work, rather than the bed, in which case, stock thickness differences between F360 model and real life shouldn't matter.) A common mistake is to reference everything from stock top in F360, carry out a facing operation with one tool, and then to set the next tool Z0 on the cut surface - it will now cut deeper than it should by the depth of the facing cut. Another possibility (if you are using a Z probe block to set Z=0 ) is that the probe block thickness is set wrongly. As Alex said, it would be interesting to see the F360 file, too.
Hi Alex, Thank you for your questions... I'm using a 1/4" spiral upcut bit (Whiteside RU2100). I'm attaching the work piece to the spoilboard with double sided tape and have not seen any discernible movement during the cuts. I am not surfacing the plywood prior to cutting, and generally cut 2-4 pockets per test. The target depth is .355 and there has been some variance in depths in the test cuts. The depths range from .410 to .395. I'm unclear if this is due to issues with the plywood I'm cutting or something happening in the cut itself. I'm using a set of digital calipers to test the depths. I'm using the RoutER11 CNC that came with our 1515 kit. I have done 6 different test cuts now, and all of them have been in different parts of the spoilboard. The depths have come out from .410 to .395 regardless of where on the spoilboard I'm cutting. Fusion 360 file attached. Thank you so much for taking the time to look at this. I really appreciate the time and feedback. Russ
I think, with unsurfaced plywood, the only test I would trust is the one near the point where you set Z workplace zero. Plywood can vary enormously in thickness across it's width. Try a scrap of reasonably hard material (NOT mdf), surface it, change to the bit you are using, reset Z zero and cut just one pocket. Let us know the results. Alex.
Ok good points in here... a few notes: In F360, I have the model thickness and stock thickness set to 0.71. I have tested setting the F360 origin to both a stock box point and a model box point. Testing both yielded the same results. In all my tests, I've chosen an origin point on the surface of the stock piece in F360 and then used the "XYZ Touch Probe Plus" to set 0 on the actual piece. I have not been doing any facing operations, so I don't think that's causing the problem. Unless it's simply due to uneven plywood stock. Regarding the Z probe, I do not remember an option to set the probe block thickness during setup. I simply chose the model I have when running the probe utility. I believe that process is working correctly since when I go to 0,0,0 after probing, the z height and position looks correct. I.e. sitting right on the corner of my stock piece. That said, I'm happy to confirm this if I can find where to set it. Thanks again for the thoughtful response. Russ
Hi Alex, Great suggestion... I've got some very stable walnut scraps lying around and I'll run the test on that and report results. These have been jointed and planed to a consistent thickness, so we should be able to confirm if it's a stock issue with this test. Russ
Yeah, I've seen your F360 file since and, to me at least, it checks out. (There's no stock addition in the F360 setup so model box & stock box will give the same results.) I see for the XYZ Touch Probe Plus that the thickness is preset in OB Control. (I use a custom probe where I have to set the thickness.) Does the Z0 height of the machine also look correct at the end of the cut? (If you compared it against a fresh piece of stock, for example?) If not, the Z axis could be skipping steps / slipping, or it could be that the tool is pulling out of the collet slightly during the cut (but in that case, I might expect one pocket to be deeper than the other, and maybe to see some evidence of an unequal cutting depth in the bottom of the pockets). TL/DR: I can't see anything wrong!
Updates after testing today: 1. I cut a pocket in a piece of walnut that had been planed to a consistent thickness. Pocket depth came out at 0.395 - 0.405 on the 4 sides of the pocket. (Picture attached) 2. the collet on the router was tight, but not as tight as it could go... I fully tightened it before the walnut test. 3. after the test cut, I returned the spindle to 0,0,0 and it looks like it's spot on. (picture attachd) 4. The picture of the plywood illustrates the challenge, the tenon should be 1/2 the width of the plywood and as you can see it's coming out about 1/3 the width. Ok, any other thoughts?? Plywood sample cut (tenon coming out too thin) Pocket cut in walnut (depths were 0.395 - 0.405) Target depth 0.355 Spindle returned to 0,0,0 after the test cut.
Doesn't accommodate bowing, twisting, etc. How you clamp comes into play too (for example edge clamps cause sheets to bow upward) . After planing, surface it on the CNC and I bet you will still cut high spots. Now remember below that high spot is likely a hollow between the stock and the bed.
Hi Peter, Sorry for not being more clear, I actually jointed the backside, then planed the top and checked the top with a known reference that is flat to within 0.001". I'm as confident as I can be that it is not a stock issue or an issue with the spoilboard hollowing. I've run almost 20 test pocket cuts on 3 different pieces of stock in 5 different locations on the spoilboard and all of pockets have been in range of .385 - .41 instead of the target of .355. At this point, I feel like we've eliminated: - loose collet - loose drive screw in the z axis - stock variations - spoil board hollowing - Z axis callibration - thickness being off on the probe plate I'm running out of ideas to test... Russ
Difficult to see from a photo (never looks the same as the real thing - especially on a computer) but is that pocket in the walnut the same depth everywhere. It looks as though it might be deeper in the middle? Alex.
Hi Alex... the bottom does look flat and smooth. When I measure from each of the 4 sides, the calipers show between 0.395 - 0.405, and honestly that difference could just be measurement error. Visually it looks consistently deep. Russ
Hi Peter, I did have some play in the z axis at the beginning. One of the lock collars on the screw had come loose. For the last few tests though, everything is tight. I just tried moving it up and down again and there is not any play in the z axis. One other note from your comment about uneven clamping pressure... I'm actually attaching the workpieces to the spoil board with double sided tape. That's why you see a slight gap between the spoil board and work piece in the pictures. I'm referencing the top lower left corner of the work piece when setting 0,0,0 with the probe and am referencing the same point in the F360 setup. Russ
Two points, double sided tape is sometimes prone to movement - try using masking tape (good quality blue tape such as 3M) and superglue - tape on (dust free) spoilboard and bottom of workpiece, superglue the masking tape together. Second, from what you have said so far I feel Z calibration is the most likely culprit (of the three axes it's the hardest to do accurately) - can you describe step by step how you did this, what you used to measure distance travelled etc? Alex.
Ok, one last test today... I attached the work piece with double sided tape (this prior to your suggestion) and then surfaced the plywood. I then adjusted the test piece pocket depths to step down in increments of 0.1". I finally reset the 0,0,0 points with the probe and touch plate and ran the cut. Here are the results: Target | Actual Cut 0.10" | 0.12" 0.20" | 0.212" 0.30" | 0.32" 0.40" | 0.42" 0.50" | 0.52" So in then end, I think some of the variance was actually due to the plywood surface being uneven. Now it looks like I'm consistently 0.02" too deep. Unless anyone has other suggestions... I can just adjust the bottom height in F360 setup for pockets where I need precise depth and make it work. I really appreciate all the feedback and support. Alex, to address your question about calibration process, I used the Z axis calibration wizard. Marked the first line with a marking knife, moved the z axis 50 mm, marked the second line with a marking knife and then measured the distance with a metric ruler (which seemed to be exactly 50 mm by eye). I used a marking knife with a flat back and rested the flat back against the plate on the Z gantry to mark the line. Russ
@rcasenhiser, that was a good test, and rules out calibration as the error is the same regardless of the overall depth of the pocket. Another correction you could do in Fusion is to leave 0.02" axial stock. Alex.
that could be play in the spindle bearings, when you probe the Z height the spindle has to push hard enough on the probe block to make electrical contact, which can be tricky with some non conductuive coatings. that pressure can then make the Z height too low by that 20 thou because the spindle pushed up when probed and sagged down when cutting. (1) I would try setting the surface of the material as stock top, and the model below that by at least 30 thou, then do a facing operation to create the model top, then pocket that and compare the depth. Yes, I know you don't want to be facing your plywood, but this experiment removes the probe height from the pocket depth measurement. (BTW you can expect short life from your bits, the glue in plywood is nearly as hard as glass and will blunt carbide quite quickly) (1) which is why I use an optoisloated low pressure Z probe Optoisolated Z probe
Hi Alex, Thank you for this suggestion! It's an easy fix and I did a test cut and everything came out perfectly. Still not sure what the source of the error is, but I can finally move on with an easy workaround. Everyone else, thank you so much for the help and support on this issue. Russ
