4 Attachment(s)
Rough Finish on 45 Degree wall
So i once thought that some of my rough finishes on angles might be do to my linear rails needing replaced on my Hurco Mill.(https://www.cnczone.com/forums/dynom...-software.html) But have recently replaced them. And now iam making an aluminum part that is octagon shape with 45 degree walls. All the walls that are inline with X or Y axis have a nice finish but the walls that are 45* have a rough finish, see pics.You can easily feel High and Low spots with finger. The walls are .425" tall and iam using a new carbide .5" variable-helix endmill.
Could this be my settings setup in the Trajectory Planner Screen?
Here is Gcode.
Code:
%
O1001 (ENCODERMOUNTPLATEOP1C)
(T5 D=0.5 CR=0.005 - ZMIN=-0.425 - BULLNOSE END MILL)
N1 G90 G94 G17 G49 G40 G80
N2 G20
N3 G53 G00 Z0.
(2D CONTOURFINISH)
N4 T5 M06
N5 S2500 M03
N6 G54
N7 M08
N8 G00 X0.05 Y-2.919
N9 G00 G43 Z1. H05
N10 G00 Z0.2
N11 G01 Z0.0394 F5.
N12 G01 Z-0.425
N13 G01 G41 X0.325 Y-2.869 D05 F20.
N14 G03 X0. Y-2.544 I-0.325 J0.
N15 G01 X-1.0538
N16 G01 X-2.544 Y-1.0538
N17 G01 Y1.0538
N18 G01 X-1.0538 Y2.544
N19 G01 X1.0538
N20 G01 X2.544 Y1.0538
N21 G01 Y-1.0538
N22 G01 X1.0538 Y-2.544
N23 G01 X0.
N24 G03 X-0.325 Y-2.869 I0. J-0.325
N25 G01 G40 X-0.05 Y-2.919
N26 G00 Z1.
N27 M09
N28 M09
N29 M05
N30 G53 G00 Z0.
N31 G49
N32 G53 G00 X11. Y0.
N33 M30
%
Troy
Re: Rough Finish on 45 Degree wall
I notice this as well. Not that bad, but I do get slightly better finish in X and Y moves then on interpolated moves. Im guessing its most likely to be minor oscillations in the servo tuning. Might be worth trying to same test with less P/I gain or more D gain. Another thing that could be working against you is servo deadband. I found turning deadband off made noticeable improvement in surface finish on interpolated moves. I hear a tiny amount of dithering without deadband so now I get my M3 and M5 programs to turn servo deadband on only while stopped.
Could possibly be a rough thrust bearings or maybe even ball screw. My Y axis thrust bearing has very slight rumble in it. I wonder if this could be causing the problem on my end. Hasn't been enough of a problem for me to dig into it yet.
Re: Rough Finish on 45 Degree wall
I had 0 for my Deadband except for X axis it had a Deadband range of 1, dont remember doing that. Made it 0 like Y and Z but still no change in finish. P and D are both 0 with I at .003.
Dont think its my ballscrews(could be wrong) as they are double nut with .0002 backlash at most. I also checked end play of ball screw when i replaced rails and got 0. Would think if it was mechanical like this that there would be similar(not exact) finish on surfaces inline with axis also, especially in harder materials??
Is your drives +/- 10V analog? Mine is still DMM-Tech's Step/Dir. I have plans to try the analog again now that the lathe is working in analog mode.
Troy
Re: Rough Finish on 45 Degree wall
Yes im using +/- 10v analog torque mode. This is with the original Fadal DC servos with feedback to Kflop only, no tach feedback to servo drives. Since your using the DMM servos and have lots of options, might be interesting to try and come up with a good tune in both torque mode and velocity mode and try cutting in both modes and compare your results. I wonder if velocity mode may do a little better here?
Curious to see what Tom has to say about this and possible ways to improve it.
Re: Rough Finish on 45 Degree wall
Ive never been truly satisfied with DMM-Tech drive in Step/Dir. Could never get it as smooth as it should be. Its really noticeable when doing high speed toolpaths with short moves. The drives in this mode are always "tunning" themselves as they state, and seems like it is fighting Kflop. I can smooth it out some by changing my Trajectory Screen settings per gcode.
Can you share your tunnings and filter settings you are using? I used your settings from your 4th axis on my lathe and it worked well.
Thanks,
Troy
Re: Rough Finish on 45 Degree wall
Here are my settings for X axis. Y and Z are pretty much same.
ch0->InputMode=ENCODER_MODE;
ch0->OutputMode=DAC_SERVO_MODE;
ch0->Vel=1020000;
ch0->Accel=3500000;
ch0->Jerk=200000000;
ch0->P=2;
ch0->I=0.025;
ch0->D=0;
ch0->FFAccel=0;
ch0->FFVel=0;
ch0->MaxI=500;
ch0->MaxErr=300;
ch0->MaxOutput=1000;
ch0->DeadBandGain=0;
ch0->DeadBandRange=2;
ch0->InputChan0=0;
ch0->InputChan1=0;
ch0->OutputChan0=0;
ch0->OutputChan1=1;
ch0->MasterAxis=-1;
ch0->LimitSwitchOptions=0x10;
ch0->SoftLimitPos=2377440; // set limit +19.5 inches
ch0->SoftLimitNeg=-2377440; // set limit -19.5 inches
ch0->InputGain0=-1;
ch0->InputGain1=1;
ch0->InputOffset0=0;
ch0->InputOffset1=0;
ch0->OutputGain=-1;
ch0->OutputOffset=0;
ch0->SlaveGain=1;
ch0->BacklashMode=BACKLASH_LINEAR;
ch0->BacklashAmount=36; // 48 for 0.0003" 12.192 COUNTS PER TENTH
ch0->BacklashRate=2000;
ch0->invDistPerCycle=1;
ch0->Lead=0;
ch0->MaxFollowingError=5000;
ch0->StepperAmplitude=20;
//Pole zero 45, 180
ch0->iir[0].B0=14.849600;
ch0->iir[0].B1=-28.946699;
ch0->iir[0].B2=14.106600;
ch0->iir[0].A1=1.804700;
ch0->iir[0].A2=-0.814233;
ch0->iir[1].B0=1.000000;
ch0->iir[1].B1=0.000000;
ch0->iir[1].B2=0.000000;
ch0->iir[1].A1=0.000000;
ch0->iir[1].A2=0.000000;
// Low pass 500
ch0->iir[2].B0=0.123861;
ch0->iir[2].B1=0.123861;
ch0->iir[2].B2=0.000000;
ch0->iir[2].A1=0.752278;
ch0->iir[2].A2=0.000000;
I didnt come up with the pole zero settings. But they seem to work very well. I bought the machine used and previous owner had started the retrofit and had motor tuning pretty much done.
I have a DMM servo on my 4th axis (+/- 10V analog setup) and tried all sorts of tuning but never got great results either. Just for kicks I tried running torque mode and adding that pole zero filter, with no idea of what it does or what values I should use, and it worked awesome. And seemed to work for you too. Strange there isn't more people using this, and more discussion about it.
Re: Rough Finish on 45 Degree wall
Also one more thing to note with the DMM drives. Using analog torque mode bypasses all the DMM tuning, so you have full control with Kflop. No fighting with each other. I had much better results with my 4th axis in torque mode.
1 Attachment(s)
Re: Rough Finish on 45 Degree wall
Hi Troy,
I'd suggest capturing data while making that cut so we can analyze how well the servo is doing.
The attached program should capture 10,000 data points with a sample rate of 540us/sample (~5 seconds worth). You must have a C:\Temp directory. The file will be KFLOPdata.txt
Assign it to an MCode (Execute without any wait and some unused Thread).
Then place the MCode into the GCode just before the diagonal cut - between N15 and N16. Then if you send me the data I will plot and analyze it.
Please also post what your axes resolutions are.
Any guess at how much deviation there is causing the ridged pattern? Any way to measure it?
Re: Rough Finish on 45 Degree wall
After cutting, rotate your part 45 degrees so its inline with the X axis, then run a tenths indicator over it to see how much variation you have.
3 Attachment(s)
Re: Rough Finish on 45 Degree wall
Hi Tom,
Sorry late responding.Getting things ready to test Granite Devices drive with my 4th axis Fanuc servo.(They give 3 weeks to try it out.)
Attached is Data file, current test Gcode and screen shot of my current Trajectory Planner Screen.
I used a .0001" indicator with a .08" diameter probe tip on the 45* and was getting .0006" to .0008" high and low spots.If i had a smaller tip probe there would most likely be more than .0008". When i checked the 0* there was not even a .0001" tenth of movement.
Thanks much again,
Troy
3 Attachment(s)
Re: Rough Finish on 45 Degree wall
Hi Troy,
Never a rush from our end :)
Here is the data plotted:
Attachment 441254
Attachment 441256
Attachment 441258
Not sure what to say. The servo errors seem to be only several encoder counts so I wouldn't expect it ever to be better than that. The servo errors seem to be ~ 4X better than what you are measuring so even if they were zero I wouldn't expect much improvement. However there does seem to be a repeating pattern every ~ 0.2seconds. Maybe lead screw pitch/wobble or something. And I would expect the servo to correct errors that persist for ~0.1 seconds.
What is your lead screw pitch and feed rate again?
Re: Rough Finish on 45 Degree wall
Interesting stuff guys. If you look at the fact that the error is adding and subtracting, the total error is much closer to measured. For example at 3.35 seconds X is leading by 0.0003", then at 3.45 seconds its laging by 0.0002", that would be 0.0005 total variation which you would definitely see in the finish. Not sure how the exact total amount of error works out considering its at a 45 degree angle, its probably much more complicated then this.
Seems odd to me that both axes have the exact same repeating pattern, its even the exact same shape of waveform. This makes me think its in the servo or drive. Looks like your feeding at 20 IPM at 45 degrees, this works out to 14.1 IPM per axis, over 2 seconds this would be 0.47", and there are just under 5 cycles so looks like its repeating every 0.100". No way your ball screws are this fine. My guess is that you have 10mm pitch ballscrews (approximately 0.400"), and you have 4 pole servo motors, and this is a pattern created going from pole to pole of the servo motors.
Re: Rough Finish on 45 Degree wall
Troy, try doing the same test again, but do it in 4 different depths, and offset the Y axis 0.025" for each depth. I bet every offset will give you a different pattern, and one will give a very good finish. This will confirm my pole theory, or eliminate it so you can go back to looking at mechanical issues.
Re: Rough Finish on 45 Degree wall
Screws are 10mm pitch.
That is interesting murray, checking with indicator the pattern is about every .100".
Do you mean different Z depths?
Also, there has always been a different tone of the servo on my Z axis. So the other day when i was doing a spiral bore and Z was feeding about 10IPM, i placed my finger on the servo coupling(dont try this at home) and was feeling a pulse of motor maybe every second. Shouldn't it be smooth all the time?
Troy
Re: Rough Finish on 45 Degree wall
Yeah do 4 different depths. For example start with a piece of stock at least 0.550 above your vise and do your test 0.125 deep. Then shift Y -0.025 negative and Z -0.125 and do it again. When your done you have 4 faces to compare. This small shift in Y will cause a phase shift in the Y axis error in Toms graph. At some point, the waves should be close to on top of each other and you should have very little position error between the two. If they overlap, the cycling will just result in micro small variations in feedrate then, and not actually create any error in the cut.
Really looks to me that this is what your seeing. If it is, its disappointing to see this from the DMM drives. You think the drive manufacturer would have this figured out, and output would be perfectly smooth.
4 Attachment(s)
Re: Rough Finish on 45 Degree wall
So i did the different depths and offset in Y. I started at .525" deep and came up from there each time and moved Y in the plus direction each time.
You can see the pattern moving to the left but finish is the same. Attached is 4 plots, one for each depth. Think the final test would be with drives in Analog mode.
Troy
Re: Rough Finish on 45 Degree wall
Can you post a picture of the block? Curious what it looks like. I was really thinking you would see some cuts with a better finish then others.
3 Attachment(s)
Re: Rough Finish on 45 Degree wall
Hi murry,
Attached is X axis only and 2 of XY 45*.
Re: Rough Finish on 45 Degree wall
Hmmm, doesn't look to be a whole lot of difference, bottom one could possibly be a little smoother but might just be picture. Do they all look similar on the dial? They all seem to have some micro variations in the finish, but the larger 0.0005-0.0008 wave you were getting may be reduced. Doesn't look like you have any repeating mechanical patterns which is good.
Interesting fact... If you feed a half inch endmill 0.004 per flute on a finish pass, you can clearly see each tooth on the machined part if you look close. The height difference on the part from flute to flute is theoretically 0.000013" high. A hundredth of a thou. So unfortunately the bar is set pretty darn high to get absolutely perfect looking interpolated cuts.
You might still have room for improvement. Will be interesting to see if analog mode is any better/worse, and how different tuning affects it.
Re: Rough Finish on 45 Degree wall
Quote:
Originally Posted by
Need TECH Help!
Hi murry,
Attached is X axis only and 2 of XY 45*.
What are the servo motors and encoder PPR
It looks like a normal finish of a low count ( PPR ) encoder machine, The difference between the actual and command position is known as lag, following error or position error, add in the jerk factor and you get what you are seeing in your finish