Plasmacam dhc2 Y dimension error

[Jacampb2]

Hi folks, I'm here looking for help again with my second hand plasmacam 4x4 dhc2. I've had it up and "running" for some time, but only to know it will move the table and make cuts. I've been trying to find time to learn the 3.11 software so I could tackle some necessary projects.

Well, I started this week and if I had hair, I'd be pulling it out. It works pretty much flawlessly with the exception of occasionally getting an error on initialization that the carriage index pulse could not be found. I'm beginning to think this is related to my overall issue, and research shows that this is likely due to faulty wiring harness, which I am planning to order in a moment.

The issue in having is that parts cut out are dimensionally correct in the x direction, but under sized in the y direction. The error increase as dimensions do. I try to post a picture of some test coupons below. I've cut dozens of these coupons, and the error is very consistent, so I don't really think it's mechanical. If I rotate the coupon 45 degrees, the sides are then dimensionally accurate, but the corners aren't quite 90°s.

I thought the table was possibly out of square, even though we really took out time getting it dialed in when we set it up. I remeasured, and it's still dead nuts square. I could pretty easily fix this in cad, but that's a huge pita to have to calculate it for every part. I'm going to order a replacement harness, but it seems odd to me that even an intermittent encoder issue could cause such a repeatable error.

Thank you!
Jason

[adbuch]

If I rotate the coupon 45 degrees, the sides are then dimensionally accurate, but the corners aren't quite 90°s.

Do you mean that you rotate it in the drawing before you cut it? Or you are physically rotating the part after you cut it? Your photo shows one sample part that is cut out of square, and another with dimensions that do not match for x and y. Please provide further clarification.
David

[adbuch]

I mean that the cut path was rotated 45 degrees. The Cartesian axis are labeled as the coupon was aligned while it was on the machine. They are all the exact same cut path. I also did some rectangular cutouts at normal and 90°, and the error always stays in the machine y direction, regardless of how the cut path is orientated in software.

Thanks,
Jason

This sounds more like a 3.11 setting problem than a mechanical issue with the table. As Robert suggested - "With 3.11 the resolution can be customized. Make sure the standard 4x4 setting is checked on the System Setup page."

David

[Jacampb2]

I replied to both of you on the owners community, but in case it will help others, here it is below.

I mean that the cut path was rotated 45 degrees. The Cartesian axis are labeled as the coupon was aligned while it was on the machine. They are all the exact same cut path. I also did some rectangular cutouts at normal and 90°, and the error always stays in the machine y direction, regardless of how the cut path is orientated in software.

4x4 table is selected. Advanced table settings are normally not available, but with a brief hex edit I was able to access them. The resolution is set the same for both axis. I figured I'd start tinkering with that setting tomorrow.

I did order a new wiring harness, so we will see if it helps.

Thanks,
Jason

[adbuch]

Thanks for the update Jason! Keep us updated on your progress.
David

[Francisco]

Quite likely a worn timing pulley on the y servomotor's shaft. As the aluminum pulley is worn down by the belt, the effective stepdown ratio changes. As the encoder is mounted on the motor shaft, the control thinks the torch had moved 4" when, in reality, it has only moved 3-13/16".

[adbuch]

Quite likely a worn timing pulley on the y servomotor's shaft. As the aluminum pulley is worn down by the belt, the effective stepdown ratio changes. As the encoder is mounted on the motor shaft, the control thinks the torch had moved 4" when, in reality, it has only moved 3-13/16".

That is a good call! Definitely something to check out.
David

[Jacampb2]

That's not really how timing belts work. It's a tooth count ratio, similar to rigid gears, or even more closely, chain and sprocket, not the pulley diameter. Unless it's worn enough to skip teeth, only backlash will increase, the ratio will not change with wearing diameter. That said, I did count the teeth to make sure no one changed it, but they are both 14 toothed motor pulleys.

In the end, I've figured out what is causing it, I'm just not really sure if there is a way to fix it since the actual motion control settings that need tuned aren't exposed to the end user. See below for my whole saga from today.

Well, I have an update, and I think I've found the cause. How to fix it is another issue all together.

I started the day by swapping x&y motors. My thoughts were that if it were an encoder issue, it would follow the motor. It did not.

I decided to try to cut the coupons at about half the speed to see what the error did. I'm practicing on a scrap sheet of 22g aluminum so I'm running at ~400ipm. At half that speed the arc is to erratic for the thc, so it was inconclusive. I got a scrap piece of steel a bit thicker, verified I got the same results in the steel as aluminum, and the error was still there. So, I cut the speed in half, reduced my cutting current to reduce heat in the part and the part came out almost exactly dimensionally accurate, however I got a big zigzag where the lead in entered the part - - I forgot a picture, but basically the start of the cut path and end did not line up. I've read this is caused by corner acceleration settings.

That's when I had my big aha moment. The axis with the most mass is always where the error exists. When my cut path was on diagonals, the moves were interpolated between both axis.

So, I've been messing with the corner acceleration to try to dial this in. The problem there is that you don't have a setting for each axis, which is nuts considering the difference in mass. I can get the y moves about right, but it rings like a bell after a corner move, especially along the x path. I would hope the pid loop is tuned for the mass of each axis, but that's not exposed to the end user. I've eventually got to what I consider acceptable results by doing a lead in from the corner and a lot of speed and acceleration tuning. The corner lead-in isn't going to work for everything, but I'm hoping it will get me through my job.

Meanwhile the new harness is on its way. My intermittent faults are getting worse and wiggling the harness brings it back to life, I trust it enough for practice, but not enough to load $200 of material on it.

I went through and sprayed all the rolling contact points with slipplate dry film lube. I'm also going to order all new roller flowers in an attempt to get the fiction lower in the heavy axis. I mean nothing is moving hard or binding, but something must not be quite right of it can't be accurate at 397ipm.

Thanks for the help folks. I'm open to other thoughts on the matter.

Sincerely,
Jason

[adbuch]

Thanks for the update. I'm still not sure what is causing your Y axis problem, but it still looks like a mechanical issue to me - at least from what I can see with your photos above. 1/16" of slop is certainly not acceptable for cutting precision parts, although it would be fine for cutting yard art. If you have access to some 14 ga. steel, I would try cutting the same part using FineCut consumables (assuming you are using a Hyperterm with Duramax torch) at 190 ipm with 0.04 G corner acceleration setting. The factory default is 0.2 G for corner acceleration. I feel that your problem is specific to your machine, as there are literally thousands of Plasmacam 4x4 tables that are capable of holding very tight dimensional tolerances for the cut parts without having to independently adjust the corner acceleration for each axis.

I don't remember if you ever told us which plasma cutter you are using. I used to cut all my parts 11 ga. and under using the FineCut slow speed settings, which for 14 ga. steel would be 150 ipm at 45 amps - and I could cut a part like yours and it would be square and dimesionally correct. I was using the standard corner acceleration of 0.2G. I have since discovered (thanks to Robert) that I can run much faster if I lower the corner acceleration and still get accurate parts, but with much less slag and heat distortion with much better detail for the intricate small shapes in a part.

If you've got some sort of acceleration induced table slop, then slowing the cut speed down to 150 and at the same time reducing your corner acceleration should give you some idea. I would focus on cut speed and corner acceleration if it were me doing the testing.

David

[adbuch]

Since I don't have 3.11, I don't know what sort of diagnostics it has. With Design Edge, there is something called a "hold test" where you can lock the motors and then manually try to move the carriage or gantry to see if there is any movement/play/slop. It would appear that you may have as much as 1/16" of play/slop in the Y direction. Since you have already swapped the motors, this eliminates the motor gear being lose on the motor shaft. I would take a look at the gantry gears located at either end of the gantry which engage with the gear racks. These two gears are normally pressed onto the cross shaft from the factory, but there is always the possibility that one has become loose and/or worn. There is also the possibility of the aluminum timing gear located on the cross shaft just inboard of the gantry gear is loose or badly worn. I would also check that all the springs are in place and tight. I assume that you have already checked all these items, but I am bring them up just in case.

Instead of wasting so much metal doing test cuts, you might also consider installing the EasyScriber into your torch in place of your nozzle/electrode and scribing test rectangles instead. Set your machine control setting to cut "on path" so that your scribed rectangles will (or should) measure correctly and correspond to your drawing dimensions. You could also set your kerf width to zero to accomplish the same thing.

David

20210812_182422_resized.jpg

20210812_182505_resized.jpg

20210812_182619_resized.jpg

20210812_182629_resized.jpg

20210812_182643_resized.jpg

20210812_182649_resized.jpg

[djreiswig]

I would use a sharpie. Then you can just wipe the ink off and start over.

[adbuch]

I would use a sharpie. Then you can just wipe the ink off and start over.

That is true, but the sharpie would need to be the fine point and very securely fastened to its mount - since we are looking at 0.030" to 0.060" differences in the dimensions.
David

[adbuch]

Thanks for the help folks. I'm open to other thoughts on the matter.

Sincerely,
Jason

Jason, it probably goes without saying that you have checked your torch mount to make sure it is tight and has no play in the clamp/torch mount. The reason I recommended the EasyScriber over a sharpie is that the EasyScriber will fit very snugly into your torch with no play or slop. If you were using a pen mount for sharpie, there is always the possibility of the pen moving in the holder, or the holder having some play in its mount. The EasyScriber will make a very narrow sharp line, as compared to the sharpie (even the Ultra Fine tip version) line width.

David

[Jacampb2]

Thank you for the comprehensive replies. I'll try to answer all the questions.

First, I am using a Hypertherm Powermax 1000 G3. It has fine cut consumables loaded in it right now. I was cutting the junk aluminum at 397 ipm with the plasma set just under 30 amps. I've had the plasma cutter since I purchased new back near the turn of the century. I don't believe it has and issues. It never caused any trouble in hand operation.

Second, I checked all the timing pulleys yesterday and the gantry pinion while I had it apart, as well as the rack on the gantry which is always always exposed. None of the cam followers seem to be sticky, but I'm going to order new for it once I get a few second. Checking the dual pinions and racks in the table rails will be a bit more difficult, but I'll see if I can find a way without completely breaking the table down.

Third, I bought new spring tensioners from plasmacam for every location back when I bought the table. The torch mount is tight and one of the first things I did was ensure it was also perpendicular to the work surface.

Fourth, the reason I was doing actual cuts is because I wanted to get a feel for the amount of material warpage and dial in sites to minimize it. At first I was doing test cuts in scrap 22g aluminum of the actual part that will be on 16g aluminum. When I noticed the issue, I just kept cutting the aluminum. I considered rigging up a sharpie, but I honestly didn't think I'd still be investigating this 3 days later. It kind of snowballed.

Fifth, I'm pretty sure the software has a diagnostic console which will allow you to brake the servos. I'll look today, however, I've manually moved the gantry and carriage enough to know that there doesn't seem to be lost motion while the servos are trying to hold position.

In the 3.11 software, the default corner acceleration is 0.1g. I think I mentioned above, but at 200ipm and 0.1g, instead of 400ipm, it cuts very close to drawing dimensions, but with the lead in centered on one straight path, I get a zigzag where the cut turns the corner after lead-in and cut end line up. I have seen other threads where it's been explained that this zigzag is related to corner acceleration. I'll try to take some pictures of it today. I don't believe it to be uncommon though. Years ago I farmed these same parts I'm trying to cut out to another plasmacam owner. Every one has the zigzag where the cut path starts and ends. It's the main reason I'm trying to get it more dialed in. It took a lot of hand work to fix that in all the parts I had him cut.

I know it's difficult to troubleshoot over the intranet where none of us know the others skill sets and backgrounds. I've done a lot of work on motion control systems. Retrofits as well as troubleshooting and repairs. Servos can't come to a nearly instantaneous stop. A lot of stuff is obfuscated behind plasmacams clunky software, but I'm assuming the acceleration setting is what determines how quickly an axis "puts on the brakes" if I set it very high, there is obvious ringing after it turns the corner, it pretty clearly servo overshoot and then it's oscillating trying to return to its requested location. It's far worse in the y direction then in the x. I believe this is directly related to the mass of the entire gantry VS the mass of just the carriage. Normally one would tune the servo amp pid controller for these different masses, however plasmacam does not expose these settings to the user. Based on that, I'm assuming they have done the tuning correctly, and since there isn't mechanical drag on the motion system like there would be in a mechanical cutting machine, they only have to account for known weights and resistance in the system. Long story short, I think for some reason my gantry movement has more drag on it than it should have.

I coated all the bearing surfaces in slipplate dry film lubricant last night when I shut the shop down. That should at least show me minor improvement of I'm on the right path.

Thank you all for your thoughts and ideas.

Sincerely,
Jason

[Jacampb2]

Well, I went to the shop today happy that I had me settings that would at least get me by. I cut a few more test coupons to be sure things were about the same, and three of 4 came out almost perfect, the 3rd in the series was under sized again, then the 4th was back to usable so I figured 3 was an outlier. I loaded up another piece of scrap steel sheet metal and cut the part I need to cut, it came out about perfect, but this is at lower speeds than I should be running, so I loaded up the last of the 22g aluminum to see how bad it would warp at the lower speed. I ran the plasma all the way down at minimum current so the warpage wasn't too bad, but the part was under sized in y again. Not as bad, but way more than in the steel part.

So I got the genius idea to look in the few pdf chapters of the manual that I have to see if I could make sense of the diagnostics screen. The section on diagnostics refers me to a chapter that's only on the video manual, which I don't have and have been unable to source from anyone. I did have the video manual text captions from the "help" program, so I was able to at least get the gist of what they were saying. I found that if I pulled the gantry all the way towards the control panel, recorded the y axis reading from the software, and then just manually moved it forward and back as little as an inch, the y axis would continue to count further and further negative with every stroke. I managed to get a 6 inch error in about 1 minute of moving the gantry rapidly by hand.

I then swapped motor wiring harness and lo and behold the x axis would now exhibit the is behavior and the y axis was perfect. I'm irritated with myself that I only bought the main harness. I just didn't think it was all that likely that the short motor harness would have a fault. I'm hopefully going to see of I can rebuild the harness tomorrow, but I'll get a new one on order and I'll let you all know how it goes.

[adbuch]

It sounds like you may have found the source of your problem. It that original harness is 20 years old, then maybe time to replace it along with the others.
David

[Jacampb2]

Yes, I'm sure the issue is on the encoder wiring. I happen to stock the molex kk connectors and crimp terminals used for the encoder, the stain relief is still in perfect shape and it's actually 3 individual separate wire bundles inside one jacket, so I suspect I'll be able to replace just the encoder wiring for now. If the 12 pin connector is also molex, there is a good chance I have those crimp terminals as well. I stock a lot of molex stuff for my work. If I have those pins as well, I'll just remake the harnesses myself instead of spending more money with plasmacam. The only reason I ordered the main harness was strictly out of convenience.

I'll let you all know how it shakes out.

Thanks,
Jason

[Jacampb2]

My replacement harnesses were all here by last Friday. I installed them while I had a few hours over the weekend. The replacement motor harness did indeed solve the problem. I also tracked down what connectors and crimp terminals are in use, so I am also going to rebuild my original harnesses for spares.

If anyone needs and are already invested in the Molex ecosystem (crimpers and pinning tools etc...), here are the part numbers for the connectors:

Molex Part # 39033155, 39033157 Male and Female 15 pin connectors
Molex Part # 1560, 1561 Male and Female crimp terminals for 15 pin connectors.
Molex Part # 22013057 KK series 5 position connector for motor encoder
Molex Part # 2758 for KK series crimp terminals
Molex Part # 39012205 Mini-fit Jr. Male 20 pin connector from harness to controller.
Molex Part # 5556 for Mini-fit Jr. Crimp terminals

I do have a new question though, but I'll start a new thread so as not to confuse this particular topic.

Thank you all for your thoughts!

Jason

[weldguy]

Hey Jason, thanks for posting back with your results. Could be very helpful to someone else on the future and glad your frustrating issue is resolved!

[adbuch]

Happy to hear you have finally resolved your problem. Thanks for the Molex parts list - I do have the pins, sockets, connectors, and crimp tools from my past EE work - but I also keep spare wiring harnesses on hand. I've never needed one yet, but you never know.
David

[Jacampb2]

Thanks guys, and thank you for the help and for listening to me trying to sort everything out. Plasmacam had been less than helpful. They were great until I sent them a copy of my bill of sale to transfer ownership. I bought the table from a state university for $125. Once they saw that, they were complete jerks to me. It means a lot that there are groups like this who will try to help support the product.

Thank you again!

Sincerely,
Jason

[Joe Jones]

I wish PlasmaCam would offer either custom length wiring harnesses, or maybe make EXTENSIONS that can be added to the stock harness wire to extend the length.

Robert Johnson ... do you make EXTENSIONS?

Joe

[rdj357]

I wish PlasmaCam would offer either custom length wiring harnesses, or maybe make EXTENSIONS that can be added to the stock harness wire to extend the length.

Robert Johnson ... do you make EXTENSIONS?

Joe

I don't make extensions. I make the full harness for $125 plus $0.60 per inch of overall length (controller plug to longest branch). I just didn't like the thought of adding yet another pair of unshielded connectors.

[Jacampb2]

I don't make extensions. I make the full harness for $125 plus $0.60 per inch of overall length (controller plug to longest branch). I just didn't like the thought of adding yet another pair of unshielded connectors.

I wish I had known that you sold the harnesses. I'd have much rather bought them from you rather than plasmacam. At least I know now for the future.

[Joe Jones]

I don't make extensions. I make the full harness for $125 plus $0.60 per inch of overall length (controller plug to longest branch). I just didn't like the thought of adding yet another pair of unshielded connectors.

Have you heard about any issues with communication errors over (n) inches of cable? Is there a limit to the DEPENDABLE length? IOW, if I convert my 5x10 to a 5x20 table, will you make a cable track harness for it, and will it work?

I will be ordering a harness from you when I get rid of the ^%&*&^ boom and convert my 510 table to a cable track system! In fact, I will probably order several of them.

Joe