Project 1 from EmptyNester. Category: Custom Shop Tools - CNC Table

[EmptyNester]

Talked to Precision Plasma about the carriages. They said it's either the eccentric bearing adjustment or the carriage is out of square. I said I suspected the carriage and they suggested "percussive adjustment" with a vice and rubber mallet.

Instead of working on that, I wired the motors on the X-axis and gave it a try. Seemed to work okay. But, I'm assuming that the DRO reads in the units you set (I set inches). I must have done my calculation of steps per inch wrong because I was getting two inches of movement for every inch on the DRO. I'll have to redo the calculations and see if I can figure out what went wrong.

I was adjusting the direction of movement for the X-axis by changed the active hi/low of the direction signal. I accidently changed X and Y instead of X and A. It's pretty dramatic when the steppers go in opposite directions.

I'm curious if anyone has suggestions on maximum speed and acceleration rate...........

[Rambozo]

I'm curious if anyone has suggestions on maximum speed and acceleration rate...........

That all depends on the strength of your motors and the mass and friction of your machine. Ideally you want to be able to go as fast as possible without losing any steps. There are some motor tuning programs that will help you get in the ballpark. Also a lot depends on your step kernel rate in Mach. Run the drivertest to see what your system can support. Be aware that you can also have resonance problem at some speeds, there are tricks to fix that, if needed. One thing about steppers is that they have a wicked torque curve vs speed. So you need to test and tune from zero through your max feedrate, then at your rapidrate. If you have a resonance between max feed and rapid, you can live with that, but not if it happens below max feedrate. And of course you must never lose a step, no matter what. A general rule of thumb is to adjust velocity up until the motor stalls, then back down about 10%-20%. Then adjust acceleration as high as possible where they can still start from stationary. This will give the best position tracking. If you plan on using constant velocity, everything will have to be tuned really close, for exact stop you don't need to be as precise.

[acourtjester]

HI Empty
you may have the same problem I have with the "Z" axis, it look like your "Z" motor is a direct connect to rack via the motor gear.
What happens when you drop power to the motor the weight of the z axis will cause the motor to free wheel and the z drops to the limit.
Slam dunk kind-of-a-thing. I need to remember to prop it up before dropping power.

[SeanMurphy265]

This is an easy way to calibrate your steps per inch. I did mine with a tape measure because I do not have a dial caliper, but I'm ordering one soon. I'm not good at math, but this is an easy way to get it accurate.

[EmptyNester]

Thanks all for the advice.

---

Sean - Since I don't have a dial caliper I wanted to figure it out by the math. It turns out that I did the calculation right, but then just plugged in the wrong number (duh). Once I used the right number, it seemed as good as I could measure it without a dial caliper.

The math I used is:

Distance per revolution (circumference) = Pitch Diameter * PI
(the pitch diameter was specified on the web page at McMaster)

Distance per rev = 0.625 * PI

Distance per rev = 1.963 inches

With 200 steps/inch and using 1/16th microsteps - I have 3200 steps/rev

Steps/in = steps/rev / inches/rev
Steps/in = 3200 / 1.963
Steps/in = 1630.158

--

Ram - one gap in my understanding (of the may) is cutting speed. I had originally thought that would be based on the stepper speed, but then realized that would work right. I'm guessing that's something you set in Sheetcam?

--

Jester - thanks for the tip - I hadn't thought about that. The motors are pretty stiff when powered off, but I'm guessing that the floating head and the torch will add some significant weight.

--

I just finished "adjusting" the Y axis carriage. With a couple blocks of wood and a crowbar, I was able to get it square. It now runs perfectly on the tube.

Just a couple more wires and I should have all axis running.

[Rambozo]

The math is the right way to do it, for sure. As you can see by your steps per inch, your machine should be more accurate than any tool you have to measure it with. You may have some periodic error in your gear and gear rack can also have errors, but as long as you eliminate as much backlash as you can (you can program the control to correct for what is left) you should be able to split thousandths using a microstepping drive. If you get a dial indicator, I would check repeatability over large moves and basic accuracy over small moves, and go from there. Depending on what you plan to do, I would shoot for a basic accuracy of plus or minus .005" or better (ideal is .0005") over the whole table. The main reason is that often your gcode will be moving incrementally and any errors will accumulate fast.

Yes feedrate is set in your gcode (by SheetCAM or whatever you use) and the table can also rapid between cuts (G1 vs G0). The feedrate will have to be adjusted to the amps and material you are working with. Also different moves are often run at different feedrates, like the little loops to make sharp outside corners. Your CAM program will take care of all the heavy lifting, but it is very useful to be able to read and understand the gcode that it spits out. It will help you figure out what's up, when things do not go as planned. Also you may not like some things and make some manual changes from time to time. No CAM program generates the most efficient code, and sometime you may want to have more control over certain moves. Like if you have a clamp holding some thin material, you may tweak the rapid moves between parts so the torch doesn't slam into your clamp. CNC brings new meaning to software crashes.

[EmptyNester]

I do embedded control software. My favorite software bug of all time wasn't mine, but done by a guy I worked with.

Working on automated rail cars (ala airport shuttles), he was trying to tweak the acceleration rate and made a math error. It was a short test track and the test car is empty (just an opening in the floor to be able to scope the electronics, etc). The people watching said it took off so fast the rolled head over heals the length of the car (with the scopes chasing him).

The best part was that the deceleraiton used the same rate. They said he made it to the other end of the car on the stop

[SeanMurphy265]

You can test your ipm, acceleration in Mach3 by jogging the gantry with your hot keys. I had to slow mine down from what no sleep studio had posted on their forums. Then you can set them for cutting through sheetcam this is cool because you can slow down your cnc for holes.

[sportbike]

... I don't have a dial caliper

Come on man, get a caliper. Even if it is just a cheapo harbor freight 6" digital caliper. (I personally prefer mechanical dial calipers so the batteries don't crap out right when you need them...)

[Rambozo]

I did mine with a tape measure because I do not have a dial caliper, but I'm ordering one soon. I'm not good at math, but this is an easy way to get it accurate.

A tape measure? Gimme a break. You should get some calipers, but what you really need for this job is a dial indicator. Again, even cheapo from Horror Fright, will do fine. Depending on your steps per inch even an indicator may not show any difference for changes of a few microsteps, you should do the math, then use the indicator to confirm and check for problems and that the machine can return to the same place every time. Even an error of .001 in the half inch method you linked to will have your parts over 3/16" off at the end of your 96" sheet.
Like the old saying goes: Measure with a micrometer, mark with a crayon, cut with an axe!

BTW: About 25 years ago while doing some large structural steel beams, there was a fitting problem when one of them was being installed, it was too long to go in place. It prompted the foreman to do some checking as to how it happened. Everyone in the shop pulled out their tape measures from the edge of a large layout table to 20 or 25'. These were mostly name brand tapes and none were super old, but low and behold there was almost a half inch difference between the longest and shortest at that length. Was a real eyeopener for me.

Come on man, get a caliper. Even if it is just a cheapo harbor freight 6" digital caliper. (I personally prefer mechanical dial calipers so the batteries don't crap out right when you need them...)

The dial also lets you "read" between the lines and guesstimate a lot closer.

[EmptyNester]

Come on man, get a caliper. Even if it is just a cheapo harbor freight 6" digital caliper. (I personally prefer mechanical dial calipers so the batteries don't crap out right when you need them...)

I have an HF caliper, but its the normal 6" digital one - not a travel dial indicator. Just hadn't been on the priority list before now.

[EmptyNester]

I did a quick video of the test.

[Rambozo]

Looks good. The acceleration vs velocity seems strange, but maybe that's just a limitation of running at 25kHz.

[EmptyNester]

The acceleration vs velocity seems strange

Strange how?

[Rambozo]

Your velocity actually seems pretty good for only 25kHz pulse rate, but is seems really lazy on acceleration and braking. That may be a function of how the jogging is handled. It's hard to see what is happening when actual gcode is running because the part is rather small, and I can't tell when it's rapid vs feed. I'm sure once everything is tuned it will be fine. The real test is constant velocity. I would think a plasma would benefit a lot from being able to run in CV mode. The parts should be a lot more accurate and with a better surface finish.
Congrats on getting it all going.

BTW if your router causes the Z to fall without power, you could counterbalance it with a small gas cylinder or even a weight. Another option is to hang an electric brake on the shaft of the stepper, since I see you have dual shaft models.

[EmptyNester]

I taped a magic marker to the z-axis and ran the sample circle g-code. It looked very round, so I figured that I'm in good overall shape.

The big issue I have now is due to the warping of the table when I welded along the length of the frame. The cross-members that support the steel being cut aren't level with respect to the gantry (they are all at different heights). I'm wondering if there's a foolproof way to level them so I can do a test cut without a THC.

[Rambozo]

Can't you just put some shims under your slats to make everything good?

As to a method, just lock your Z an inch above the slats, then use a 1" test block as a feeler gauge between the Z and the slats. Shim under the slats until you fill the gap, check at each end and maybe the middle. Then move to the next slat, rinse and repeat. The 1" is arbitrary and you might change that gap depending on how much warpage you are talking about.

[SeanMurphy265]

Cut your slats according to what would make them level with the gantry. Make sure your gantry is level and then make a card board to pattern your slats. The crappy part is that if the warping is not the same you may have to cut each slat accordingly so it will be level. I'm sure that would be a pain, but I think it would be easier to work with a level cutting surface.

[EmptyNester]

Thanks for the ideas.

The table has a arch in both directions (length and width). So, every slat would be different (I'm more of a software/electronics guy than a steel guy). I've had lots of ideas but nothing that I figured would work.

I thought that using a 3 small pieces of angle running length wise (each side and middle) I would easily get a good level. I just don't think that I can use heavy enough angle so that it wouldn't bow but still bit in the water pan.

I'm afraid that if I try cutting custom templates for each slat I won't be able to get it precise enough to be worth the effort.

Short term, I think the shim method is probably the best bet. That seems like it would get me to test cuts the fastest (doesn't everybody want instant gratification?).

Once I can use the table to cut, would the following work?
1. Run a string along each side so that I can mark level at the end of each slat.
2. Use the table to cut each slat from one mark to the other.

[Rambozo]

Once I can use the table to cut, would the following work?
1. Run a string along each side so that I can mark level at the end of each slat.
2. Use the table to cut each slat from one mark to the other.

How about just use your marking pen, but mounted at an angle or sideways, then have the table mark all your slats for you. (I like machines to do all the work) Plus the idea of level is somewhat arbitrary, what you need is them to be in the same planes as your gantry, so having the gantry make the marks assures that. Are your slat slots accurately cut?

That seems like it would get me to test cuts the fastest (doesn't everybody want instant gratification?).

Of course, make some sparks! Once you have a THC my guess is that you won't need to have things very precise anyway in the Z.