[scyan]

Hey guys, our cnc is doing flat spots when asked to re-create circles... ive attached a file to you can see what i mean.. its a crude representation but you guys will understand right away !

Thanks for any help in advance !

david

[mrainey]

Don't know if this applies to your situation, but ...

I used to see that when machining large radii on a Bridgeport Boss 5, which could only move in minimum .001 increments. When nearing a quadrant (0, 90, 180, 270 deg position), the tool had to move a whole bunch in one axis before it could make that .001 move in the other axis.

This showed up on the part as a pretty wide flat at the quadrant.

[scyan]

I hope this is not the case or that we can fix it, otherwise, the machine is pretty much unusable...

Anyone else ?

[keithorr]

Scyan,
You describe the result. What is the input? G2 or 3 circular interpolation? If linear code (small segments) please post.

[[email protected]]

scyan,
Looks like a good time to check your machine for lost motion.
When you reach the 12 3 6 or 9 oclock position on the circle, an axis has to reverse and go the other way. Looks on your diagram you have a clasic case of lost motion. Place an indicator from the spindle housing to a tee slot on the table. Set a zero on the indicator . go to incremental feed, .0001" then move the table in .0001 increments until the the axis moves. reset the zero on the indicator and also zero set the axis in question, increment the axis in the opposit direction until the axis moves. Note the reading on the readout. That reading should give you a feel for the amount of lost motion you have. This might be refered to a backlash.
If the backlash is too much, then the same procedure should be used but indicate from the casting of the machine to the end of the lead screw. Then from the slide in question to the thread of the lead screw.
What you are looking for is something about what you seen when you checked back lash in the first step.
It coiuld also be improperly programed reversal error.
Hope this -----o never mind. Regards Walt...

[higgrobot]

We had a similar problem on a large retrofitted borer, turned out the backlash in our ACME lead screws was just too much for doing this type of work. We found tightening the nuts up and using backlash compensation was helpfull but the problem was only realy solved when with switch to ballscrews.

[scyan]

the syil already has double nut ballscrews.. so im guessing they might be loose ! We'll have to check that out ! I noticed in mach 3 that there were numbers enterned in the backlash section but it was not turned on.. hmmm

Im not sure i follow the procedure about the backlash verification, ill have to be in front of the machine and try it out.

Ohh, and heres an example of the code with some circles in it..

N1G40
N2G80
N3G90
N4G54

N5M3

S3000


N6 G00 Z0.25
N7 X2.2687 Y2.0669
N8 G01 Z-0.05 F10.
N9 X1.2887 F5.
N10 G00
N11 G02 i0.001001001 J0.
N12 G00
N13 G01 Y1.5841
N14 G00
N15 G02 i0.001001 J0.0
N16 G00
N17 G02 X2.2687 i0.00149 J-0.3636
N18 G03 i0.001 J0.
N19 G01 Y2.0669
N20 G00
N21 G02 i0.001 J-0.0001
N22 G00 Z0.25
N23 G01 Z-0.1 F10.
N24 X1.2887 F5.
N25 G00
N26 G02 i0.001001001 J0.
N27 G00
N28 G01 Y1.5841
N29 G00
N30 G02 i0.001 J0.
N31 G00
N32 G02 X2.2687 i0.00149 J-0.3636
N33 G03 i0.001 J0.
N34 G01 Y2.0669
N35 G00
N36 G02 i0.001 J-0.0001
N37 G00 Z0.25
N38 G01 Z-0.15 F10.
N39 X1.2887 F5.
N40 G00
N41 G02 i0.001001001 J0.
N42 G00
N43 G01 Y1.5841
N44 G00
N45 G02 i0.001 J0.
N46 G00
N47 G02 X2.2687 i0.00149 J-0.3636
N48 G03 i0.001 J0.
N49 G01 Y2.0669
N50 G00
N51 G02 i0.001 J-0.0001
N52 G00 Z0.25
N53 G01 Z-0.2 F10.
N54 X1.2887 F5.
N55 G00
N56 G02 i0.001001001 J0.
N57 G00
N58 G01 Y1.5841
N59 G00
N60 G02 i0.001 J0.
N61 G00
N62 G02 X2.2687 i0.00149 J-0.3636
N63 G03 i0.001 J0.
N64 G01 Y2.0669
N65 G00
N66 G02 i0.001 J-0.0001
N67 G00 Z0.25
N68 G01 Z-0.25 F10.
N69 X1.2887 F5.
N70 G00
N71 G02 i0.001001001 J0.
N72 G00
N73 G01 Y1.5841
N74 G00
N75 G02 i0.001 J0.
N76 G00
N77 G02 X2.2687 i0.00149 J-0.3636
N78 G03 i0.001 J0.
N79 G01 Y2.0669
N80 G00
N81 G02 i0.001 J-0.0001
N82 G00 Z0.25
N83 G01 Z-0.3 F10.
N84 X1.2887 F5.
N85 G00
N86 G02 i0.001001001 J0.
N87 G00
N88 G01 Y1.5841
N89 G00
N90 G02 i0.001 J0.
N91 G00
N92 G02 X2.2687 i0.00149 J-0.3636
N93 G03 i0.001 J0.
N94 G01 Y2.0669
N95 G00
N96 G02 i0.001 J-0.0001
N97 G00 Z0.25
N98 G01 Z-0.35 F10.
N99 X1.2887 F5.
N100 G00
N101 G02 i0.001001001 J0.
N102 G00
N103 G01 Y1.5841
N104 G00
N105 G02 i0.001 J0.
N106 G00
N107 G02 X2.2687 i0.00149 J-0.3636
N108 G03 i0.001 J0.
N109 G01 Y2.0669
N110 G00
N111 G02 i0.001 J-0.0001
N112 G00 Z0.25
N113 G01 Z-0.4 F10.
N114 X1.2887 F5.
N115 G00
N116 G02 i0.001001001 J0.
N117 G00
N118 G01 Y1.5841
N119 G00
N120 G02 i0.001 J0.
N121 G00
N122 G02 X2.2687 i0.00149 J-0.3636
N123 G03 i0.001 J0.
N124 G01 Y2.0669
N125 G00
N126 G02 i0.001 J-0.0001
N127 G00 Z0.25
N128 G01 Z-0.45 F10.
N129 X1.2887 F5.
N130 G00
N131 G02 i0.001001001 J0.
N132 G00
N133 G01 Y1.5841
N134 G00
N135 G02 i0.001 J0.
N136 G00
N137 G02 X2.2687 i0.00149 J-0.3636
N138 G03 i0.001 J0.
N139 G01 Y2.0669
N140 G00
N141 G02 i0.001 J-0.0001


N142 G00 Z0.25
N143 X2.0032 Y1.2204
N144 G01 Z-0.05 F10.
N145 G03 I-0.2245 J0. F5.
N146 G00 Z0.25
N147 G01 Z-0.1 F10.
N148 G03 I-0.2245 J0. F5.
N149 G00 Z0.25
N150 G01 Z-0.15 F10.
N151 G03 I-0.2245 J0. F5.
N152 G00 Z0.25
N153 G01 Z-0.2 F10.
N154 G03 I-0.2245 J0. F5.
N155 G00 Z0.25
N156 G01 Z-0.25 F10.
N157 G03 I-0.2245 J0. F5.
N158 G00 Z0.25
N159 G01 Z-0.3 F10.
N160 G03 I-0.2245 J0. F5.
N161 G00 Z0.25
N162 G01 Z-0.35 F10.
N163 G03 I-0.2245 J0. F5.
N164 G00 Z0.25
N165 G01 Z-0.4 F10.
N166 G03 I-0.2245 J0. F5.
N167 G00 Z0.25
N168 G01 Z-0.45 F10.
N169 G03 I-0.2245 J0. F5.


N182 G00 Z0.25
N183 X2.2687 Y0.5314
N184 G01 Z-0.05 F10.
N185 X1.2887 F5.
N186 G02 i0.001 J0.0001
N187 G01 Y0.8568
N188 G02 i0.001001001 J0.
N189 G03 X2.2687 i0.00149 J0.3636
N190 G02 i0.001 J0.
N191 G01 Y0.5315
N192 G02 I-0.0001 J0.
N193 G01 Z-0.1 F10.
N194 X1.2887 F5.
N195 G02 i0.001 J0.0001
N196 G01 Y0.8568
N197 G02 i0.001001001 J0.
N198 G03 X2.2687 i0.00149 J0.3636
N199 G02 i0.001 J0.
N200 G01 Y0.5315
N201 G02 I-0.0001 J0.
N202 G01 Z-0.15 F10.
N203 X1.2887 F5.
N204 G02 i0.001 J0.0001
N205 G01 Y0.8568
N206 G02 i0.001001001 J0.
N207 G03 X2.2687 i0.00149 J0.3636
N208 G02 i0.001 J0.
N209 G01 Y0.5315
N210 G02 I-0.0001 J0.
N211 G01 Z-0.2 F10.
N212 X1.2887 F5.
N213 G02 i0.001 J0.0001
N214 G01 Y0.8568
N215 G02 i0.001001001 J0.
N216 G03 X2.2687 i0.00149 J0.3636
N217 G02 i0.001 J0.
N218 G01 Y0.5315
N219 G02 I-0.0001 J0.
N220 G01 Z-0.25 F10.
N221 X1.2887 F5.
N222 G02 i0.001 J0.0001
N223 G01 Y0.8568
N224 G02 i0.001001001 J0.
N225 G03 X2.2687 i0.00149 J0.3636
N226 G02 i0.001 J0.
N227 G01 Y0.5315
N228 G02 I-0.0001 J0.
N229 G01 Z-0.3 F10.
N230 X1.2887 F5.
N231 G02 i0.001 J0.0001
N232 G01 Y0.8568
N233 G02 i0.001001001 J0.
N234 G03 X2.2687 i0.00149 J0.3636
N235 G02 i0.001 J0.
N236 G01 Y0.5315
N237 G02 I-0.0001 J0.
N238 G01 Z-0.35 F10.
N239 X1.2887 F5.
N240 G02 i0.001 J0.0001
N241 G01 Y0.8568
N242 G02 i0.001001001 J0.
N243 G03 X2.2687 i0.00149 J0.3636
N244 G02 i0.001 J0.
N245 G01 Y0.5315
N246 G02 I-0.0001 J0.
N247 G01 Z-0.4 F10.
N248 X1.2887 F5.
N249 G02 i0.001 J0.0001
N250 G01 Y0.8568
N251 G02 i0.001001001 J0.
N252 G03 X2.2687 i0.00149 J0.3636
N253 G02 i0.001 J0.
N254 G01 Y0.5315
N255 G02 I-0.0001 J0.
N256 G01 Z-0.45 F10.
N257 X1.2887 F5.
N258 G02 i0.001 J0.0001
N259 G01 Y0.8568
N260 G02 i0.001001001 J0.
N261 G03 X2.2687 i0.00149 J0.3636
N262 G02 i0.001 J0.
N263 G01 Y0.5315
N264 G02 I-0.0001 J0.


N265 G00 Z0.25
N266 X2.0376 Y1.2204
N267 G01 Z-0.0394 F10.
N268 G03 I-0.2589 J0. F5.
N269 G00 Z0.25
N270 G01 Z-0.0788 F10.
N271 G03 I-0.2589 J0. F5.
N272 G00 Z0.25
N273 G01 Z-0.1181 F10.
N274 G03 I-0.2589 J0. F5.
N275 G00 Z0.25
N276 G01 Z-0.1575 F10.
N277 G03 I-0.2589 J0. F5.


N278M5

N279M30
%


THANKS !

[scyan]

Ohh, I forgot.. does anybody know if we have adjustable ballscrews ?

Sorry if its the wrong word.. hehe

[HuFlungDung]

The gcode looks a little rough.

I don't like the lower case 'i' in there and there seems to be zero occurance of G02 XYIJ or G03 XYIJ format commands, and some entries with 5 to 9 decimal place accuracy.

I would try a simple circle on the origin for a test of the machine and see how it looked.

Although it could be backlash settings, this would typically involve a bit of a dogleg jiggle at each reversal as the tool crosses a quadrant line. I don't see that in your pictorial depiction of the results.

[scyan]

What do you mean by a center hole from the origin ? Sorry, really newbie here...

Also, ill post a pic of an actual result in a second.. so you'll see exactly what it does...

[scyan]

Heres the pic of what we did... ill post the code later but it was a basic like 4 or 5 lines of code, just a circle done from bobcad.

Here's the code in question...

N1G40
N2G80
N3G90
N4G54

N5M3

S3000

N6 G00 Z0.25
N7 X0.4375
N8 G01 Z-0.05 F5.
N9 G03 I0. J0.

N10 G00 Z0.25

N11M5
N12M30

[[email protected]]

scyan,
I am going to out of town for a few days. Have to make enough money so I can play on these forums.
Try checking for "backlash" the way I mentioned.
If your confused, welcome to the club but get with HuFlung via EMail, may he can help you out. I will be able to help you out Wednesday night.
Wish I could say "Hope this helps" but we don't pos ta say that no more.
That is, that is what I have heard somewhere.
Try it once and see if it makes more sense when you are trying it in person.
You might want to bolt a 123 block to the table, then you can check both axies, I think you are using X and Y and showing the errors at reversal.

Regards Walt...

[SyilAmerica]

I once was a newby as well. I paid ALOT of money for a large CNC routing table to make all my aircraft parts. 6 months later I was about to use DYNAMITE and blast the dang thing into pieces.
I couldn’t get perfect circles! Nice egg shape with some flats. Always on the same side! I got lots of horrific advice on how my table could be mis-aligned, and my steppers where messed up……..
Then, someone asked me as simple question. Did you tune your stepper motors? Have you measured your backlash and input them in the backlash screen….. My answer…. Ah, ah, I was just getting to that!
Several hundred perfect circles later……..
When you have “flat” spots on the same side of a hole in is most commonly not a correct setting in one of the above categories.
The nuts are pre-loaded and tested for backlash when they leave. No adjustment is necessary. Most of the time the solutions are easy, but when we are getting started in something as complex as CNC machines we can often overwhelm ourselves. Makes for some palpitating moments I can tell you.

[SyilAmerica]

Looks like your drawing has flat spots as well.
The I and J are in absolute? I'm having trouble loading the code.

Did you draw it in BobCAD or import a file? When you import circles into BobCAD, sometimes you get some double lines and extra stuff that needs to be cleaned up a bit.

[GVlcek]

Based on the location of the flat spots (where each axis switches directions) I would suspect backlash in your system. If the flat spots are located directly after the quadrent is reached (0°, 90°, 180°, 270°) in the cutting direction.

It could also be tuning if you're using servo motors. Basically your proportional or integral gain is set too low and is allowing steady state following error that is amplified (doubled) when the axis switches direction. If that is your problem, you would notice the OD on the round sections would be smaller than programmed.

If you're using steppers, I wouldn't point to tuning as they are open loop and hardware tuned based soley on current and inductance.

I'd look at backlask first, and let us know what you find.

[Pres]

Accodring to my handy dandy NCPLOT proggie (www.ncplot.com) your start and end radius's differ by more than 0.001", beginning with block N17 then N32, N47, N62, N77, ... ah, too many!
Must be a Trig trigger
Pres

[scyan]

newbie here <-------

What does that mean exactly ? I know mach 3 gave us an error at first about some of the lines and someone suggested we use a I of 0.001 instead of 0.000 like it was in some areas... But i dont get that. Its incremental IJ...


Scy

[bmracing28]

i dont know if it will work, the software supports these codes.

X0 Y0 (CENTER OF ARC)
GO1 Z-DEPTH F5.
G13 I=DIA.-HALF OFTOOL F=FEED RATE D=DIA OF TOOL L=NUMBER OF TIMES TO RUN IN LOCATION
G0 Z0.1
MOVE TO NEXT LOCATION

NOW YOU DO ONE OF TWO TYPES OF PROGRANING, You program on line or by hand, where you have to put the Dia. of tool in tool regisitry. Or you use CAD?CAM software in which you describe the tool in it, then you would put value to bring it to cutting size. You might also try G64 continuse path.

[NC Cams]

First and foremost, how much backlash/lost motion are you ACTUALLY measuring when you do a direction change???

If you have ANY lost motion/backlash in the screws, nuts or support bearings, any input you're putting in with your motors is NOT getting applied to the table. When this happens, the CLASSIC location is EXACTLy at the 12, 3, and 9 o'clock direction change positions that you're experiencing.

If you have issues with path deviation at the 45 degree quadrants, you can usually blame that on servo/stepper tuning. These are the 4 points where the motor speed is at peak velocity and most sensitive to control input errors.

The solution is most likely in fixing the slop in the screws/nuts. We spent a LOTs of time and money to eliminate the same problem, albeit to a lesser degree in our Bridgeport based Extrak fittted with ground ball screws and preloaded screw support bearings. I doubt if you're looking for the level of perfection that we were BUT you'll need to jump through essentially the exact same hoops to eliminate the problem.

Do some "Extrak + ball screw" searches on the 'Zone and you should be able to find pertinent info on how to eliminate the problem (methods, procedures, part ideas, etc)

[Switcher]

What the !

N11 G02 i0.001001001 J0.

.