What causes Saw tooth pattern in cut

[ChrisO]

I occasionally get a saw tooth pattern from CNC plasma cutter. See pictures. Anyone have any experience with solving this? It seems to be random. The torch head is well mounted and there is no play in the gantry. This saw tooth pattern will appear on straight cuts or radius cuts. I can be cutting a straight line perfectly smooth, and half way through the cut start saw toothing.
I appreciate any thoughts you have to offer.

Retrofit CNC System:
Mach4 Hobby
Vital Systems Hicon Controller 7866 with 77ISO board and ArcPro Plasma Profile
DYN4 DYN4-H01A2-00 AC SERVO Drives
DMM 750 Watt NEMA 34 motors
Thermal Dynamics Pak Master PCM-100XL Machine Torch with floating head
Retrofit Table: 6’x24’ linear rail and belt drive. Water Table

IMG_3519.JPEG

IMG_3518.JPEG

IMG_3517.JPEG

[weldguy]

Could be a number of things but the first thing I would check is the THC, it can be a common cause of this and is one of the easiest things to check first. If your using a THC to control your height automatically try turning it off and see if the sawtooth pattern goes away and post back.

[jimcolt]

Cut a 4" rectangle, mark the sides with x+, X-,Y+ and Y -. Now place the rectangle you just cut on the table arranged to indicate each axis. Power down your machine and push / pull each axis by hand with the previously cut rectangle laying on the table. Often you will feel tightness or backlash in one or two of the axis. Determine if there is one axis that is a problem.....can be a dirty way, or rack gear....or a loose motor mount. If the sawtooth shows on all y directions on your rectangle....then it may be an issue with the z axis, loose or tight motion. clean, lube as necessary. If problem is still there, cut another rectangle with the height control active. Rest your hand on the z axis.....feel a pulsation? Try setting a lower z axis speed, also try to adjust acceleration or gain on the z axis. Jim Colt

[ChrisO]

Thank you for your insight. I am cleaning the rails, rack and pinion. I'll remove belts and focus on finding drag and tightness in the drive system, and try to remove any drag I might find. When I begin testing I'll also test with torch height control turned off. I appreciate you suggestions, and will post back when I have some results, Thanks

[weldguy]

Before going to all that work I would try the THC test first. So quick and easy and if that is it you will be glad you didn't tear everything apart.

A simple 24" straight line and toggle the THC on and off during the cut, or cut 1 line on and 1 with THC off and compare.

I mention this because sometimes THC's can get stuck in hysteresis as it samples the voltage and can't decide whether to go up or down and ends up oscillating very quickly causing what you see.

I notice the sawtooth seems to stop as it nears the corner for a direction change, this could be the THC shutting off for the corner slowdown, or could just be that your hardware is running rough as Jim mentioned and the slower corner speed runs smoother.

Either way the THC is fastest and easiest thing to cross off your list of possible problems.

[robertspark]

are you sure it's not the plasma cutter.... older type transformer plasma cutter may give this sort of a profile at a consistent feedrate..... it's to do with the 50 or 60hz supply voltage at high current.

cut a rectangle with the THC OFF..... then look at the plasma cutter.... cut a second rectangle at 80% feedrate and see if the profile is closer together..... indicating the torch.... if it is unchanged then it's the table.

[ChrisO]

Robertspark, that is interesting, as I have had a different pattern in the past. My issue is getting the machine to repeat this issue often enough to test as you suggest. Thank you for your input.

[ChrisO]

I finished cleaning X and Y's rails and rack, found a little drag and loosened up the Y axis rack and pinion. I began cutting with THC off 4" squares but because problem is so intermittent, I wanted to reduce material waste, and switched to cutting 4" strips in the Y and X directions. Out of about 100 cuts with THC off, I got 1 saw tooth pattern. see pics. This cut was 4" in the Y directions with a quick U turn in X to turn around. Note the saw tooth pattern didn't stop as the machine cut U shape pattern. Saw tooth pattern is all the way around.

My gut is telling me its not a drag issue. I ran the machine with the plasma cutter turned off, and my hand on the Z axis feeling for vibration. Its smooth until it abruptly rounds a corner, and then continues on smooth again. Also, I don't think the Servos are capable of moving the heavy gantry quick enough to create this saw tooth pattern in the X direction while traveling at 90ipm in Y direction. see pic of CNC table. (This thing is....kinda heavy.

Test material: Steel 14ga,
cut speed: 90ipm
cut height: .19
Plasma settings: 40A, Torch Tip: Small pn 7503
THC off

IMG_3562.JPEG

IMG_3552.JPEG

IMG_3551.JPEG

IMG_3547.JPEG

[tcaudle]

What ratio reduction gearheads do yo have on those smaller servos. If you can tape a marker and instead of wasting metal, draw out the toolpaths you can make some headway. If its ragged with just the marker its 99% mechanical. It easy to understand how a slanted line or a circle can have ragged marks because its constantly moving tow axis and ANY backlash will cause that effect. If it also occurs on straight horizontal or Vertical cuts then its some other issue. To get that pattern you have to have motion being delayed on one or more axis. Since its so evenly spaced it points to a mechanical issue where it either skipping a tooth or it slipping on each rev . Its Servos are neat virtually impossible to create that pattern from a linear g-code move.

Servos are neat but they require some extra attention . First they are LOW TORQUE, HIGH RPM devices (just the opposite of steppers) so the interface gearing (transmission) has a to be different. The PID settings need to b e such that the motors reacts in a smooth way as the load changes . Things like acceleration settings became important .
MOST important is to understand the servos monitor the motor shaft position NOT the table position. ANY backlash or looseness in the mechanics AFTER the motor can't be corrected no matter how good the tuning.

Th check of backlash do this: Turn on the table and the motors should lock. Go to each motor and try to move that joint by pushing and pulling on the gantry or that axis. Watch the motor shaft to make sure it does not move (mark it if you need to)
Then if the is ANY backlash find where it is and fix it.

If there is absolutely NO backlash or movement on any axis then consider that maybe the pinion drive gears may have a flat or bad tooth. The rack appears to be vertical so the moors have to have some form of down force to keep them seated in te teeth of the rack.

A final testis to remove the motors .gearheads and place them side by side on a bench and write a g-code file tha does linear moves back and forth at varying speeds Mark each shaft and start them all with he positron at zero. In the test file send the motors out then back to 0,0 multiple times. Put pauses int he code of several seconds to let you see the marks and that they are all true. If they never deviate and always s returns to zero by the shafts THEN run some of you code and see if its got any ragged motion . If both those test pass its NOT the motors or the code or electronics, its mechanical .

Good Luck.

[ChrisO]

tcaudle, Thank you. The ratio reduction on X & Y Axis: is 11.12 servo motor turns to 1 turn of final pinion drive and gantry moves 5.5014" Z axis is 15.5 motor turns to 1 turn of final pinion drive.
The use of a marking pen is a good idea, Ill run a bunch of testing with slanted lines and circles and see if the saw tooth pattern shows up. I do have this saw tooth pattern on a straight cut. The images of U shaped cuts I posted earlier were in the Y direction. Thank you for your insight.

[ChrisO]

Just following up with my progress. I reduced drag in the rack gear and pinon in the Y direction. I then replaced the plasma torch with a marker and began running tests to see if I would see any vibration in the markups. I specifically use slanted lines and circles ( spirals) and performed these tests in the area of the table that is used most often. I didn't see any issues and when holding my hand on the Z axis never felt any vibrations. I'm confident the z axis is running smooth when these intermittent saw tooth patterns emerge. Its possible that I have eliminated my saw tooth pattern issue... but my gut tells me I have not... so
I will move on to your first suggestion, of cutting with THC turned off and let you know what I find.
Thanks everyone for your suggestions!

IMG_3572.JPEG

IMG_3569.JPEG

[acourtjester]

Nice replies, keep them coming until a resolution is made, these help other in the future if they see a similar problem.

[weldguy]

Your marker tests look great, sure looks smooth, and looks like you've turned it into a giant printer lol

Curious to hear if it still runs smooth with the THC on. I and others appreciate your feedback working through this issue.

[tcaudle]

Its down to the Z axis and the THC motion (although that should be obvious frim the Z bouning up and down rapidly)
The final thing you are left with is plasma noise is getting into the motion electronics but that typically shows up randomly and usually as cut position errors.
I waffle effect typically comes form either ragged motion on an axis *and seldome presents on a stright line ) ANy of that with X or Y would ahve shown up on the mqarker test.

SO The act of troubleshooting is to find everything that IS working to eliminate that and what you are left with is the answer.
You know its not normal motion ( at least at the speeds the drawings are being done) so it a more narrow tofus.

If its not torch wobble (again not likely on a straight cut ) then SOMETHING is different between a non torch "cut" and one with the plasma on .

I thin you are real close to the answer. It up to going back to the torch and making cuts with no THC.

You Z axis is rack and pinion? approx 15:1 . what is the distance per motor turn? The XY should like over all if it moves over 5 inches per pinion turn the overall ratio is rather low.

All of that is less important than it seems to point of the Z . Either in the THC motion or BECAUSE of the THC doing up and down and perhaps overshooting. Its kinda scary but if you loosely grip the torch with finger tips while it cuts and close you eyes you should be able to feel any woggle or motion raggedness,

[ChrisO]

The Z axis is a rack and pinion drive type. 15.55 rotations of servo motor produces 1 rotation of final drive and gantry moves 5.495" (some rounding of values here as the rack and pinion are the same as the x and Y rack and pinion.) I checked THC z axis speed settings on the Hicon Controllers Arc Pro plasma settings, which they call "Adjust Speed %" which is set to 1% of max speed of Z axis (700ipm) which should mean 7 ipm during Torch Height Control operation.
I will begin testing with THC off with torch set to .19 off the top of the material and continue the same 90ipm cutting speed and see what I get. I'll do straights and circles and see what I get.
Thanks everyone, I really appreciate the feed back!

[ChrisO]

I began test cuts with THC off and made several cuts and the saw tooth pattern showed back up after several cuts. With THC Off (no Z movement) I feel pretty confident that THC is not a contributing factor.

I decided maybe dirty, oily contaminated air could be an issue, I turned on the air and blew onto a white rag and found discoloration began to appear after a couple of minutes. So I added a Motor Guard particulate filter to the supply line. I then disconnected the Plasma torch hose from the power supply, circulated 1 gal of detergent through the line. I then thoroughly dried everything out. I ran the white rag test again and it left no discoloration, it now blows clean.
So I ran some more Test Cuts and after several cuts the saw tooth pattern showed back up again.

So I am pretty confident that THC movement and dirty air is not a contributing factor in the saw tooth pattern I am experiencing.

Motor guard air filter and element.png

Dirty Air.JPEG

[ChrisO]

OMG, I think it's the Torch head causing the saw tooth pattern cut!
I decided to replace the Torch head with the hand torch, I rigged it on there and ran several test cuts, followed up by cut job that had about 40 ft of cutting distance and the cuts came out clean.
Looking at the offending torch head I can not see anything wrong, so I don't know why.

Temporary hand torch mount.jpg

IMG_4229.JPEG

IMG_4227.JPEG

[Deezl Smoke]

WOW, very interesting. It may be possible that the inside of the air supply hose in the machine torch has separated and once air gets behind the separated bit, it can plug the supply, then, depending on the way it has separated, can release the higher pressure air, then repeat,.... I've seen this in hydraulics fairly often. Real pita trying to diagnose.

[ChrisO]

Yes, I understand I have seen that on fuel supply lines. You might be correct, and maybe I’m not seeing any cut issues because I have reoriented the hose during my replacement of the torch, causing the hose to open back up and not show the issue. Thanks for pointing this out. Now if I could only figure out a way to test this theory….

[weldguy]

Very interesting and good find. Curious if you ever find out why the machine torch was behaving that way, please post back if you find out why.