Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: 2bits on April 18, 2009, 09:56:53 PM
-
Newbie using a home 'desktop' cnc. I've been at it hard for close to 2 months now and have been getting more advanced with some projects. This led me to my current problem and the software vendor says it's not their problem. I can cut Left, Right, on the line of circles, squares, whatever without issue and within a .002 tolerance all day long except for Pocketing with Island avoidance. In fact, take a simple 1" circle, put a .175 circle in the middle of that and tell it to pocket it and avoid the inner circle and I end up with .150 - .155, yet the outside of the pocket is 1.003 I can separately cut a .175 circle and be within .003
I have calibrated the system on X and Y it's within .001 - .002 (dialed in a 4" block measured at 4.279" and got 4.278"), backlash is the same, I can cut a 1" circle and measure all the way around and get 1.006 - 1.007. I've replaced the router with a better one and verified I do not have a runout issue with either the router or the bit. Again, this only happens on islands within pockets. My current workaround is to cut the raised island first with a slightly larger toolpath around it, then pocket specifying the larger toolpath for island avoidance to prevent it from getting whacked. If I do this same procedure without the larger toolpath, it will get cut down.
Anybody know why this is happening? Is it some configuration issue between the post processor and Mach3, or some setting in Mach 3? I'm new at gcode, but the simulation looks correct and the gcode looks correct, yet, it will eat into the island.
Appreciate any help on this.
Thanks,
Mark
-
Maybe if you attach your code someone will see if there is a problem with it.
Hood
-
Gcode file attached. Thanks
I should mention that I have tried this cutting with climb and conventional cutting and a finish pass with no difference in the outcome.
-
Try this g-code, it pockets a .175 island in a 1" dia
See if it makes things the correct size. Your sample code was cutting octagons not circles, was it meant to be like that.
Graham.
-
With so many files and trials going back and forth, that may have been the gcode tech support sent me to try, just in case, I regen'd the gcode from my file, like yours a 1" hole with a .175" center.
I'm afraid I'm really new and green at this and just starting to read and understand the gcode, but I have noticed there have been a couple differences in how I've seen 'circles' handled from a friend of mine (green like me) and tech support and the software itself. Octagons doesn't sound right, but I couldn't tell why it is.
I ran your gcode and ended up with .165-.168 measuring around the island, the outside of the pocket came in at .999
Thanks for looking at this,
Mark
-
what material are you cutting?
Graham
-
Cast Acrylic .250 and .125 thick
-
How much of the tool is sticking out of the collet/chuck, rule of thumb is 5 times dia max.
Graham
-
The bits have rings on them, I put them to the ring. .8" from the top of the ring to the end of the bit
-
I can only think its cutter rub that is causing the difference in size, have you tried running the feed and speed at a much higher rate?
Graham
-
I can try that, I've just run out of ideas and it doesn't help when 2 months ago I knew basically nothing about this.
I just tried one more thing hoping it's more accurate than what I tried before on calibration. Using the .1181 bit, I set it about .02 into the material, then ran the calibration utility and told X to move 5"
Entered that it moved 5", so I could get access to jog, and moved the gantry out of the way and measured the cut less the width of the bit. I got 5.108 less .1181 for a travel of 4.9899 I then reran the utility again and entered 4.9899. I repeated the same process for Y which was a net of 5.009 Even then this isn't off that much.
So I ran your code again and got .168 and .999, at 45 degrees on the outside of the pocket was 1.0
-
I just ran the gcode you sent again at 15 ipm, 3x faster from the 5 ipm, didn't make any difference other than quicker.
-
If your axis is .010" out over a 5" distance then that is a long way out in cnc terms and will cause sizes to be inconsistent, you need a 5" command to move 5.000" before anything will work right.
Graham
-
Without having a dial indicator handly available, is the method I used accurate? My thought was if I measure the cut with my digital calipers, less the dia of the bit, that should get me pretty close? I orginally thought for what I'm doing it's close enough, but since I'm making really small stuff, it's probably more important to get this right.
-
It is probably best to calculate rather than try to measure, doing the calculate steps by the button will only ever be as accurate as you can measure where if you calculate by pitch/motor steps etc you should get good results, unless that is the screws are not very accurate and then if that is the case you wont get it accurate by any means.
Hood
-
Hi.
You could try to increase RPM, 500 is very slow for a .079 end mill.
Just for the fun, try it at your MAX RPM and try to keep material as cool as possible, put some water on the cut and look if it's make any difference.
Jeff
-
The rpm's means little, I do have a variable speed router, but only 15k to 30k, currently running at 25k. I've had better results at this speed with chips.
I did the cut and measure and did it a couple of times and seems to be consistent, 5" gets me 5.002 to 5.004
Not sure on the calculate method how it's done. From the vendor, "The screw is 10 threads per inch, the stepper is 200 steps per revolution, you multiply one by the other and you get 2000. The controller can be set at 1/1, 1/2, 1/4, 1/8 and 1/16 step mode. Your controller is set at 1/2 step mode so you multiply 2000 by 2 and you get the the steps per inch"
So this means I'm at 4000 steps per inch? 1 step = .00025 So I just add or subtract however many steps to equal what I'm off?
I must have measured pretty good, and I learned a major lesson and alot about calibration today! I just recut the file, and I got 1" outer on the pocket and .174 on the inner island!
Close isn't good enough, it has to be correct. I thank everyone who helped me with this! I'm going to cut some other stuff and verify that they come out correctly now as well.
Mark
-
What the H********* is going on??? >:( >:(
Last night after recalibrating and cut the gcode correctly provided by Graham, I cut a couple others, everything was working great. Shut everything off, went to bed. Tonight when I get home, I can't cut anything correct again, not even Grahams gcode I cut last night. The outside of the pocket was 1.020 all the way around and the center island supposed to be .175 was .153. So based on my new knowledge, I figure if the circle is off by .018, then I'll manually change the steps in Motor Tuning accordingly. I should be at .00025 per step, and since the circle is off by .018, I should modify X and Y by that amount, so .018/.00025 = 72. I subtracted that amount from each X and Y and ran the job again. Now my circle is 1.000 - 1.001, but the damm island has dropped to .138 How can that be?
Because nothing else makes sense, I went back to the original values and added 72 and recut, the circle as expected was now 1.0028 and the island is .149 I've set it back to what I had last night, but I don't get what is wrong with this setup. Values are the same as what I had last night when it started cutting correctly. So what's changing? Extremely frustrated with this, been screwing around with this problem for a week now. Everything I read or advice I get makes sense, yet nothing works. Why can I adjust the outer pocket and yet cannot get the damm island to come out right?
-
Hi Mark sorry to hear your woes.
No one has mentioned your machine, if there are mechanical issues then that is what is going to determine the accuracy.
This is where a skilled operator on a manual machine comes in he/she can adjust on the fly.
Maybe some pics and specs of your machine would help
Phil
-
I'm not in the big league, just a home 'desktop' cnc. I purchased it new from a guy off ebay, for the most part it seems built well enough and I wanted a plug and play system rather than build one, at least for my first system.
Here's what specs I got:
24"x12"x 3" cutting area
- 282 oz/in steppers on all axis direct driving the lead screw
- 12 Precision linear ball bearings (4 per axis)
- Hardened shafts
- Anti-backlash wear compensating lead nuts
- 3 axis chopper microstepping controller
It came with a Rotozip Z1 router which I've replaced with a Rotozip Z20 because of runout and I wanted variable speed.
(http://www.markdidit.com/project45/masterscnc.jpg)
-
I'm not in the big league yet either that's why I'm here too.
I shall no doubt be shot down in flames but routers generally do not work at high precision, for that you need a milling machine.
Have you access to a DTI you will then be able to see by putting the DTI on the spindle and pushing on each axis how easy it is to move out of position.
THIS IS DONE WITH THE SPINDLE OFF.
The only way to confirm is to check this way.
The other test is to remove the motors disconnect the screws and manually push the axis this way you can feel for any binding/crabbing.
In my case i use a mill is is s****e in quality but by fettling it is now unbelievably accurate and i did know it would be s****e BTW
I believe all your troubles are mechanical
-
The hard part for me to understand if there is any 'slop' or binding, runout from the router, gantry etc, I should see the same margin of error on the outside of the pocket, not just the 'island'. It doesn't matter what bit size I use, and I can get the same error cutting .020 deep versus .08 deep on a single pass. I'm only cutting acrylic and a .020 deep pass with a brand new bit shouldn't put any strain on the motors, etc to cause this deviation especially when it seems to occur when pocketing.
Hell, I accidentally hit a clamp once and it almost ripped it out of the track. So the system has some torque behind it.
Last night, I ran a test, I cut a .5 circle, no pocketing, just cut it to a depth of .0625. it measured .5007. I then told it to pocket a 1" circle around it and avoid the .5 circle. It created a 1" circle and ground the island to .485. Why can I cut the circle accurately, but not pocket with island avoidance? It's like the system changes the radius and doesn't take in or drops out the additional distance for the bit to leave a .5" circle.
I don't know, that's why I'm posting this stuff here, because I'm a novice at this, I believe I understand how it's supposed to work, but it's beyond me as to why I'm experiencing what I am.
-
What happens if you cut say a 1" outside diameter circle in only one pass with say a 1/4" bit. Does the inside circle be exactly 1/2" smaller than the outside circle?
Next test, cut a 0.8" outside diameter circle in the exact same center as the first circle. This cut will of course be mostly air. Does the inside circle be exactly 0.2" smaller now than after the first cut was made?
-
I'll give it shot tonight when I get home and see what it does.
-
When you say "Pocketing with Island avoidance" do you have a program that is creating your G code by giving it the parameters for the pocket and island?
If so, what program is it?
-
Another question, your file is labeled "hole test conventional" does the conventional part mean conventional milling as compared to climb milling?
That is what it looks like to me running your program. What happens if you re post the program in climb mode?
-
Correct, it was to designate that it was configured as conventional. Climb the normal default, made no difference. Using a finish pass made no difference.
As far as creating the job, yes, I select a Area Clearance pocketing function, select the toolpath, bit, depth etc, optionally pick toolpaths of the 'islands' to avoid and generate the gcode.
-
Looking at holetest.tap and adding or subtracting .118 for the tool size your G code gives me .175" hole, a .5" hole with a .1752" island and a 1" hole with a .5" island.
The G code is correct. Do you have any numbers plugged into the tool window?
-
I agree, the gcode is correct and looks fine, just don't get how I can adjust and get a 1" outside dia as anticipated and have the island off.
If I understand your question, I only specify the bit, the depth, climb or conventional, an optional finish pass, inward or outward on pocketing (neither make a difference by the way) the clearplane height etc. Nothing that overrides the toolpath dimensions. If you mean the tool window in Mach3, no.
-
Ok, just for me, run both of these programs. If the sizes are still off, Uninstall Mach 3 and re install it.
-
I'm running behind tonight and will try and get to it. While I don't have a dial guage I got thinking I do have a digital calipers that seems to be pretty accurate. So I rigged up a portable vise and hooked it to the gantry and recalibrated again (5" runs). According to the calipers, it was way off, so much so I wonder if the specs for the machine I got are correct. Originally I was cutting while doing the mach3 wizard, measuring and entering that value and eventually did some manual calcs and entry for fine tuning. So anyway, I thought I'd try this (wish I had the change to buy some good precision tools), when done, I cut a 1" square and was with .010. From there I have tweaked manually and got a 1.0 square +-.002 (for now???) So now I'm currently recutting the file from Graham to see what that does, then I'll try your files. It's getting late, so will see how much time I got yet tonight. Thanks! By the way, for tweaking, is going off a 1" square correct?
-
Sorry, out of time tonight to get to the files you want me to cut. I'm beginning to wonder if this system can and will hold a calibration. Do I need to spend the money and get a dial guage with a block setup to dial this as close to 5.000 as possible? If I'm 5.006, is that enough to throw this off when cutting the island from .175 to .148? I'm picking 5" right now because my calipers only go to 6". I know the dial guage would be the most correct, I'm just wondering if I'm spinning my wheels, because I'm not getting an exact true measurement. Or, is there something wrong with my controller or motors where I'm skipping or gaining steps at random? Even with the calipers, it would vary +-008. Sunday after playing with it all day, I had finally successfully cut twice to spec, and the last two nights the calibration seems all over the place. I don't know if it's because I don't have the right tools or if there's something wrong with this unit. Once calibrated, should I expect the same measurement everytime or within say .003? At this point, I don't think there's any point in cutting test files until I can get the calibration stable.
-
I have an old Bridgeport and am happy with +-.0005. From what I can see of your machine, +-.002 or +-.003 should be reasonable.
A six inch caliper should be just fine at those tolerances. One thing to think about is when your Z axis is all the way down you will have the
most potential for deflection errors. Try mounting your material to be cut as close to Z zero as possible leaving just .2" clearance.
For testing Try this file. Remember to subtract the tool diameter from 5" for the measurement. The file cuts 1/2" around each corner.
-
If you are +-.003 over 5" then over the 24" travel of the machine you can be .040" out and dimensions will not be consistent. It must be and can be spot on.
Try putting the calculated figures in again. When you change a steps per setting you should restart Mach3 and then do your checks, if you are not doing that that may be why you are getting strange results.
Graham
-
Because of the design of this system and using 1.5" bits, there is not enough travel to as far as required, so I had to raise my sacrificial board 1.25" from the get go which puts me alot closer to Z zero.
Graham, thanks, I was not restarting Mach3 after changing steps, so I'll give that a try. Tonight I'm also going to install Mach3 on a newer, faster pc and try that as well just to rule out the pc I'm currently using. It's at the minimums and seems to run it ok, but it's 1 more thing to rule out since this seems to be such a strange issue.
-
Alright, grabbed new pc, downloaded and installed fresh copy of Mach 3. Got my portable vise and digital calipers out, ran calibration on each axis using the Mach 3 utility for 5" moves. I did this until I got 4.999 on X and 5.000 on Y and then I did it 5 more times getting consistently 5.0 +- .001 Even if my calipers aren't the high dollar ones, it consistently gave me close results. So at this point you would think everything is good to go. Restarted Mach 3 and cut Grahamns file, results are: Outer dimension of pocket; expected 1.0 got 1.025, Inner Island dimension; expected .175 got 1.50
It's .025 off, how can it be off .0125 on each side when i did sufficient calibration to get 5.001? I did another calibration check, 5.001, It's like calibration and cutting are two separate entities, neither one has anything to do with the other. What's unique about this .025? it's always been off in this area, .020 - .025. On the controller, if it's supposed to be 1/8 step, but it was strapped for 1/2, or vice versa, would this be real obvious, or could this be a reason? Or maybe the controller is just bad in some way? Pc is ruled out, software is ruled out (CAD/CAM and Mach 3), gcode issues ruled out, system appears to be able to calibrate, but fails with .025 overcut consistently. Motor to lead screws coupling is tight, can't change the threads per inch, no backlash noted especially when it's .025 over all the way around. New Router, less than 4 hours on it, new bits, no runout detected. This doesn't leave much but the motors and the controller. And it would be real nice if the builder would return my emails!
Mark
-
If you set Jog to 1" increment and hit X jog 5 times how far does it move?
-
It has probably been mentioned before, but the backclash compensation is turned off?
Have you tried the test I suggested in reply # 22?
That little test should say something about the hardwire and calibration issues.
Run the last cut in the same direction as you do when you'r pocketing the final small diameter cut.
-
SteinarN, backlash compensation is off and I did run that test, but it was off because of the same issue I'm having right now. Calibrate X and Y which appears to be correct, but when cutting it's off by .020-.025. I'm at work right now, so I don't have the numbers I recorded, I'll post the results when I get home tonight. I didn't get around to mentioning it earlier because I was ticked off it was still wrong after calibration and it basically ended up the same as all other test cuts.
MachineMaster, I'll check the 1" jog tonight as well when I get home. Sure glad this is a hobby venture, I'd be out of business by now trying to get this up and running to make a living.
-
2Bits, are you SURE the bit is .118 and NOT .143??
IF you have the machine calibrated correctly then The only way I can see "your " error happening is IF the bit is .025" larger than you think it is. That is the only way I know to cut oversize in both interior and exterior cuts on the same setup without tool comp being applied and there being NO runout on the bit.
AS a further test setup your caliper and do a simple 1" move at slow feedrate. did the caliper show it to be exactly 1".
What were the settings in motor tuning (steps per inch)
Just a thought, (;-) TP
-
I wish it were as simple as a misIdentified bit. I've measured it and it's .1181, I get the sames results with a different bit size anyway.
-
OK have you cut a simple straight slot (one width of the bit) and measured the slot?
Repeat the movement test of 1" move the axis in a direction to remove backlash Set the caliper and then move 1" how far did it move??
What are your settings for steps per unit?
(;-) TP
-
Will have to try that tonight when I get home. The builder finally got back to me, he's out of town. He thought my controller may be set at 1/8 step mode, but the config file is based on 1/2 step mode. If this were wrong, I'm not sure if it would be off by alot more or not. Says he's never had a controller go bad in 5 years or the motors, and agrees this is odd that I can calibrate but be off when cutting. Wants me to send the controller back. I'd like to pop the case apart and check the step setting, but I'm not sure how it comes apart and I don't want to void the warranty unless he tells me to go ahead and pop it apart to check.
-
This thread is done!
I might have solved my issue although I don’t trust the calibration because I had it correct last night and tonight when I got home it was off again (.016). While I checked runout by turning the spindle manually, my eyes are no longer as sharp as they used to be and my son said he could see a slight blur on the bit at speed but he wasn't sure. This is a brand new router, but the way things have been going, anything is possible. Anyway I recalibrated and replaced the collet and nut with a brand new set and I’ve cut 2 items out exact to spec. I cut Grahams file exact, the holetest_g3 whoever created that one to spec. Currently cutting a third to see if I can keep going. I’m actually quite surprised to see it cut out .175! So I should now be able start cutting again, barring this whole fiasco starts over again, but it’s looking better than it has since I discovered this.
I appreciate everyone's help in this and I hope I can return the courtesy to someone else in the future, I have a lot to learn.
Thanks again.
Mark
ps: 3rd part finished to spec .5" hole - .5"
I can see me purchasing a higher quality collet system in the future.
Mark