First off, what a great site and group of people. I finished my build about a month ago and have since been testing it out and getting used to the software. I based my build on Gerald's F117 build. I probably should post a build thread. Anyway it uses: Rack and gears for Y and X Acme for Z KL 5amp drivers KL 48vdc power supply KL Nema 23 570oz/in Steppers C10 BOB Mach3 I did the full axis calibrations multiple times. I used the longest practical units when calibrating, 46" on y and 36" on x. When cutting a project which was drawer dividers I used 1/2" baltic birch ply. I used calipers to measure the ply and used that in my design program (trying out Aspire). But the dado cuts were too narrow. I had to input a larger figure than what I calipered to get the correct fit. So that's what prompted this question. What accuracy should I expect from my machine? What error measurements are "acceptable"? I set up a test board. Two circles and two squares. You can see from the pictures what the results were. If you can't read it here are the results. 1" circle Pocket ---- 1.014" actual 3" circle Profile ----- 3.03" or 3.01" actual depending on which side I measure. 3" square Profile ---- 2.986" on x and 2.996 on Y actual using outside conventional toolpath 5" square Profile ----- 4.988" on x and 5.001 on Y actual using outside conventional toolpath I was expecting closer tolerances. Should I be? What would be "acceptable" ? One thing that really stumps me is the Y measurement on the two squares. The 3" square shows short and the 5" square shows long. By the way, this is a hobby machine, not for production. However I still want to get it as accurate as possible. So what do I do to improve it I believe the mechanical parts are tight. I cannot see any jerking, vibrating etc. I do not hear any missing steps. Will changing micro stepping help? Currently at 16. Calibrate at 3", 5", 20" etc? Increase / Decrease velocity and accel -- currently at 500 and 250 Any input is welcome
I think you have done a great work till now. You does not need any correction here. You are going very good.
Hi George, Y looks good. Try a smaller and larger square see if it gets worse or stays the same. Have you checked them with a square? Feed per rev on x looks like it could be tweaked a touch. I don't remember where that is in mach atm. tinker with set steps per unit in motor config maybe. I haven't tinkered with mach lately and I'm not at home. +- a couple of thousandths is good imo. Congrats on the build joe
Larger square is about the same. I had reduced the micro steps on Y - Will try the same on X. Running the axis calculations adjusts the steps for the motors. Rectangles do come out square. What I don't understand is if Y and X are pretty close, why do all the circles come out smaller that what is called out?
how are you generating the Gcode for the circles that come out small? if your software is using tool offsets (like G41) then having the correct tool selected will matter.
So if mechanical, do you have any suggestions as to where to look? Microsteps, axis calibration ( done many times), drivers??
Try the usual stress test. Bring the gantry to the front. Put a 1mm or similar bit in the chuck, and set the height just above the base board. Set Zero, say bottom left. Put a pencil dot at the exact tip of the bit on the base board. Program for a move to the extreme right and back. Check for Zero. Set it to repeat the movement say six times without stopping, travelling at a fast speed. Check Zero. Repeat from Zero up the board in the Y axis and back again. Try diagonal from the Zero, to stress both axis'. Try a large circle, fast, and back to Zero. In both directions! The object is to stress the mechanics of the machine to the extreme limits of direction. It's return to exact Zero is the only result you can accept. With a bit of luck you will find a deviation in one direction or another. This is a test which many should use periodically, to show any weakness or accumulated errors in the mechanical department. Give it a try and get back to us. Gray
Gray, I'm out of town currently. When I return I will try the stress test and report back. David, I did that right after I posted. I figured like you that it was best to have them the same.
So I took Grays advise. Set up a stress test on my machine. First I mounted a 60 degree 3 flute v bit. Sep up 0 near the front left corner. Lowered the bit into my bed just enough to see the 3 distinct lines in the bed. Then ran the machine at 500 in/min. 20 time on the x axis going 36". 20 times on the y axis going 41". 20 times on the diagnal to x36 y41. When all were completed and I lowered the bit it fit perfectly onto the lines I started with. To be honest, I was surprised to have it come out so exact. I then repeated the 20 runs with a 24" circle. Same result. At the end the bit fit perfectly into to original lines. So I'm not sure where that leaves me except to confident in the mechanics of the machine
What I was getting at was maybe your problem is related to the load your putting on the spindle. Lighten up your doc and see if the problem persist.
Joe, that is possible, although I am very conservative in my DOC - usually only .125". However I am in the process of building a new router mount so that may help.
Well if nothing else, it should have eliminated a whole lot of other potential problems. It must have been satisfying, as well as frustrating. As Joe says, the other stress point is the directional leverage caused by the amount bite you take on each pass. Gray
ALL machines flex under load, I can change the cut size on my 120kg cast iron lathe with one finger. Since you have accuracy when not under cutting loads, it must be flexing when cutting. Reduce your Depth of Cut (DOC)! 1/8" is quite 'large' relative to the bit size. if bit is 1/8" then that is 1 diameter deep, a very large cut for most 'not the size of a car' machines. This will help you figure out what your machine can do. http://www.precisebits.com/tutorials/calibrating_feeds_n_speeds.htm
David, thanks for the link. I have studied that page before because I purchased some tapered ball nose bits from them. Also the DOC, I misspoke in the post a few spots up.For my 1/8" bit I use a DOC of 1/16" - .0625. But even that may be too much. I will certainly keep that in mind when i re run the tests. As you and others have suggested, I think I am getting flex in my router. That is why I am currently making a new beefier router mount. When complete I will rerun the tests and hopefully see a difference.
