First pass with multiple depths enabled always twice the step that is set

[jemmicat]

I have seen where people have had similar issues but what was suggested didn’t change anything

i am using fusion to create toolpaths for cutting guitars

i am using a ttc450 pro cnc router so I have to do multiple depths for all cuts. It is always doubling the depth of the first cut and this the pocket is always deeper by this amount. So a 22mm pocket depth is 23.5mm with 1.5mm DOC. This is the same if depth of cut is 1.5 or 2 or 3. It doubles the very first cut but each successive cut is correct

but here is what I don’t understand. I zero out the bit and then set the height of the material for the bit position. 45mm in this case. So the position of the bit is at the correct position in the machine and reality. I then start the program cutting and the first cut should be at a depth of 43.5mm. The machine says it is at 43.5mm. But the actual cut is at 42mm - twice the depth of cut. Each successive step is 1.5mm and the 22mm pocket is 23.5mm deep. This behavior happens no matter what process - 3d adpative clearing, 2d contour, 2d pocket, etc. the first cut is always exactly twice the multiple depth setting. It was only today that I verified, though, that it looks like the depth of cut offset is also offsetting the actual z position by that amount

i have noticed this behavior with cuts that were too deep and that initial cut being wrong but today I had a perfectly flat piece of wood so I know it is not my wood that is the problems. The machine prior to running the program is at exactly where it says it is. But then the program runs, there is an offset that is introduced that is exactly equal to the depth of cut set in fusion. What I did not do and should have now that I thought of it is to move the z axis back touching the material surface and see what that reads

i am using z0 as the stock bottom. I have calibrated the z0 and it is nearly dead on. Each step after the first one is almost perfect. But the first step is too deep and the gcode somehow puts an offset into the z position

i’m attaching the file ii ran today along with the gcode created for just making the middle feature.

while it could be my machine, I don’t know how I could set the zero point then position it to a known height and it being correct and then that very first pass and every position afterwards reporting a mchine position off by exactly the distance as set in the multiple depths

i use Candle to send the gcode. I don’t know how this would matter

[programming2C78B]

The code is correct, so it's something youre doing on the controller. Are you losing steps? 
I also recommend you look at Adaptive toolpaths instead of super shallow pockets. All of your feeds + speeds look extremely conservative - is that machine really that under powered? 

programming2C78B_0-1776351548950.png

[jemmicat]

it maxes out at 800mm/min speed. it will do deeper cuts but I have had some issues doing that so I keep it conservative

While I know the code is correct as far as the z, what I do not understand is that the depths are always off EXACTLY by the amount set for the DOC. if it is 1.5mm, the first cut is 3mm and the pocket is 1.5mm too deep. if it is set at 2mm, the first cut is 4mm and the cut is 2mm too deep. 3mm and first cut is 6mm and cut is 3mm too deep. it is very consistent. but I've absolutely no clue what is causing it

in the picture on the top right, it was supposed to just have a bevel and not a ledge. I used 3D adaptive clearing for that bevel with 3mm depth of cut. because everything was 3mm deeper than it should be, there was a step where it met the top face instead of a blend. this is typical for how it is working. but why?

[jemmicat]

it also doesn't explain why the z value is being offset negative by exactly the amount of the depth of cut value

[programming2C78B]

Well, there's something 100% controller or machine related that's causing it. I'd look for EMF issue and lost steps to start. Run these in scrap/air and confirm the Z heights throughout the program. Are you using a candle/GRBL post processor? Have you tried another controller like UGS? 

[jemmicat]

But again, why is it consistent and repeatable with lost steps? It makes no sense. Why would the exact error in the first cut correspond exactly with a doubling of the depth of cut? Also, why would it ONLY lose steps on the very first cut as every other cut is exact-ish? I’m not ruling out a controller issue, but the logic doesn’t support that. If it was losing steps it would be random and not ALWAYS on the very first cut

it doesn’t appear to be an emi issue. Perhaps. I am using candle. I want to use gsender but it doesn’t detect my z probe. I’ve not tried UGS as I do not like the UI and while I like candle, it is a bit antiquated

[jemmicat]

I had it set as stock top for everything in the heights and changed that to model top and model bottom as I read that might be contributing. It will be a few days before I try again as I have to wait for the epoxy in that sun area to cure. My plan is to check it on one cut of the outer shape and then verify that the z position i checked before running is the same after the cut as before. Since it is consistent, I could offset my z probe by the depth of cut I think and resolve it. But this is a band aid and not the right way to resolve it

[programming2C78B]

losing it in the same place DOES make sense because a) its the same code (ie lost steps due to acceleration to the first location and it's binding on one command)
b) EMI issues will happen at a certain position and voltage draw (of the spindle or drivers) 


You can try running it with the spindle off, no material mounted, and in another area of the bed and come back and verify your Z after. Do it again with the spindle on and see.
If thats not it, it may be a true load issue but you can 100% know that the code+fusion is not the problem if the gcode says Z23.

It may not make sense to you now, but I have over 10,000 hours of programming and running these things. Eventually the black magic has some logic behind it 😉

[jemmicat]

But this happens with multiple programs with different designs and different depth of cuts. Same issue. That is why the lost steps makes no sense to me. Maybe I am wrong. And again, it is a different number of steps for 1.5mm, 2mm and 3mm and has happened all over 460x360mm pieces of material. Just seems odd that the amount of lost steps with the 3 different depths of cuts would EXACTLY equal the amount that would cause the first cut to be EXACTLY double. Not impossible but just seems unlikely

and again, if EMI, why does it ONLY happen on the first cut which can occur anywhere on that 460x360 material and also happen whether the stock is 42mm thick or 45mm thick. Anything is possible but that just seems unlikely

my plan is to do my same z probe command and then verify the height. Send the program and then cancel before it starts cutting and then verify that the dimension is what is intended. This might help me narrow it down.

[programming2C78B]

As I said, voltage. The spindle is ramping up and drawing more then usual when the program is starting. This is just ONE option of what might be going wrong here. 

[jemmicat]

I accidentally hit accepted solution

the spindle is a makita trim router not controlled through the grbl controller at all. I will report back when i’m able to verify the dimension

[lauri_barnhart]

Hi @jemmicat, thank you for your question.

Did previous replies by @programming2C78B help clarity your question?

If this helps, feel free to accept it as the solution.

This helps other users benefit from the shared solution.

If the issue persists, let us know with an update so we can assist further.

Best,

Lauri | Community Manager

[jemmicat]

Yea and no. I isolated one problem with my testing… the spoilboard wasn’t completely flat. This was not helping matters. But after flattening the spoilboard, the first cut is still not the correct depth. This is most evident on 3d adaptive clearing of a bevel on the design. The bevel should be leaving .2mm from the top surface and simulation shows this. But it again is closer to that 1.5mm that I have been experiencing. I will continue testing