Tuning slaved axis

[flyhigh]

Hi all,

I am trying to tune a gantry that is running on rack and pinion system under no load. I have identical motors, racks, and pinions on both sides. The A axis is slaved in mach to the x, and both motors are using the same steps per inch in mach. However, when I run the length of the table, I am off by a huge amount - about 3/8 of an inch -- one is too far, and the other not far enough. If I run even 5 inches and return, it is off almost the same amount -- . basically, the gantry skews as much as the hardware will allow it to. I am pretty certain from visuals and noise that the steps per inch is off on one or both motors, they thus bind, pull out of true, and start losing steps...

Since they are slaved and there is no way to run one motor without the other, how can I test the tpi and set it, or what else can I possibly do? In theory they should both have the same steps per inch, so I don't really understand the issues here. They "should" be at 200 steps per inch, but in fact they are set at 170 since that is what I measured using the mach tuning function (I am not using micro-stepping, since I am attempting to build something fast with like .2mm accuracy as ok for now).

Thanks in advance!

[lik2waterski]

I am new at this myself,but is there a chance your pinion gear is slipping on the motor?

[flyhigh]

I don't think so. I cranked them down, the motor shafts have a flat side for the screw, and I have seen both sides skip steps, which implies a tight pinion. The problem for me is that there are always a ton of variables, and it is virtually impossible to rule any of them out completely .

[PaulRowntree]

Does the gantry move smoothly by hand if the pinions are disengaged ?
Is there any evidence of a pinion jumping the teeth of the rack?
Is the error dependent on the velocity and/or the acceleration?
Switch the cables to X and A motors : is the problem the same or reversed?
Does it sound like hell?

[flyhigh]

I will switch the x and a and get back to you. My table is rather far from the house. as to the rest - negative to everything. I tuned the rails to 500 ipm, which isn't great for rack and pinion, but the problem persists even if I slow it down to 120 ipm. I was getting some noisy skipping on the A motor, it was close to a fluorescent light that I unplugged, which resolved the problem - or so I thought. All my wires are shielded (except for the last 18 inches where they connect to the motor wires).

[PaulRowntree]

If the fluorescent lights really do affect your steppers, then you have some very serious grounding/noise issues. If a light can make a difference, my guess is that any line noise would send your system off into never-never land. Check that all wires are well connected, that the grounds really are in place, and that the shielding is really well connected. How much volts & amps are you driving the motors with? What is the motor inductance?

Can you extend the pulse duration going from the PC to the BoB/G540? Try to make your machine as conservative as possible, not pushing performance envelopes, and see if working far from the bleeding edge helps you out.

The correlation with lights is a big flag. Many years ago I worked for 2 weeks to track down a noise source in a sensitive piece of equipment. It turned out that a faulty neon light of a power bar was being triggered on/off by the daylight and overhead fluorescent lights and this was infecting the whole room with sporadic electrical noise. That was a tough one, but I can't really see how it would affect your motors unless you also had severe grounding issues or marginal pulses driving the stepper driver.

[flyhigh]

I can't be 100% positive on the neon light (I adjusted the pinon at the same time), but I can verify it again as best possible. I don't even know what you mean by connecting the shielding. Is it supposed to be grounded? My table and gantry are both steel, the legs are wood. how critical is that, and should they be grounded at all costs? I am not even sure how I would do it, do I just run a line to the concrete floor, or should I tie into the ac ground? I thought of that, but didn't want to fry everything if there was a problem with the ac.

The oddest thing about this is that the table *seemed* to be running square after I dealt with that issue, and then "picked up" this problem. I kept my stepper wires pretty far from each other (there is some bundling with zip ties), etc, and tried to be conservative with everything. I am fairly unhappy with my hardware purchases (I bought a kit that didn't pan out -- it is really sloppy -- the fact that 3/8 error is even possible is just abysmal in my book), but this extreme error of instant out of true has to be the motors/electrical I think.

At any rate, from your comments, it seems that you are ruling out a simple issue of each motor actually working at different steps per inch. The step pulses are at a 4 out of 5 max in mach, I will set them to 5. The specs on the motors on the x, A and Y axes in bipolar parallel config is 2.8 amps, 1.5 ohms, and 7.5 mH inductance, but I bought all the electronics together from someone who sells a ton of them with many happy costumers, and he assured me they are all compatible. The numbers when I ran them left plenty of margin both in regard to power supply, BoB and driver board, etc.

I greatly appreciate the assistance. Thanks for everything!

[PaulRowntree]

I am always cautious about giving electrical advice, so please consider any and all advice very carefully... Also note that my comments on the driving voltage are based on GeckoDrives systems; other makers may do things much differently.

1) I would make sure that the steel of your machine is well grounded, if for no other reason than you have a powerful AC spindle motor attached to it, and if a wire breaks loose, you want to protect everyone who might be near. Some people provide a grounded metal box on the machine to plug the spindle into, and if done right this grounds the machine. Or you could run a separate wire to a pipe, or better, an electrical outlet. Ask for professional help if you are in the least bit uncertain about this.

2) When you say 'shielded' cables to the steppers, the shield must be connected to something that is at ground or else the 'shielding' is of limited value. I use a G540 driver, and I run the shield connection to one of the ground pins of the DB-9 connector on the G540. On the machine end I bring the wire out to a terminal that I can use or ignore. But the shielding itself is grounded.

3) Properly installed AC outlets have a ground that is independent of the AC itself. If the AC 'goes wrong' then the ground is there to save your skin while waiting for the breaker to (hopefully) pop.

4) I don't run slaved motors, so I don't know if it is possible to give slaved motors different 'pulses per inch' calibrations. That is why switching the motor cables would be interesting. If the switched motors now skew the other way, then it would seem that the stepper drivers are not delivering the pulses consistently. Anticipate that the gantry will move in the opposite direction to normal when the motors are switched.

*** is there any chance that the motors are not rotating in opposite directions when you are trying to move the gantry? Again, I don't run slaved motors so I don't know if mach3 takes care of this for you, but for a dual rack r&p gantry the motors must rotate in opposite directions or the motors will immediately try to rack the gantry (though I can't imagine that it would move very far down the table unless one motor was much stronger than the other).

5) 7.5 mH motors would need 32*sqrt(7.5)= 88 V to get maximum power, according to Gecko's std formulas, but this is of course limited by the driver's limits. Are you near this? What is the max voltage your drivers can work with? With such high inductances, you may be under powering your motors, such that they are prone to stalls etc.

6) Is there any chance that one set of coils on a motor has a bad connection to the driver? Disconnect the cable from the driver, and measure the resistances of the A-A and B-B coil leads at the driver end of the cable using a good ohm-meter. They should be pretty similar for all 4 sets, and all should have infinite resistances to ground, the cable shielding, and the frame of your machine.

Cheers!

[PaulRowntree]

4) I don't run slaved motors, so I don't know if it is possible to give slaved motors different 'pulses per inch' calibrations. That is why switching the motor cables would be interesting. If the switched motors now skew the other way, then it would seem that the stepper drivers are not delivering the pulses consistently. Anticipate that the gantry will move in the opposite direction to normal when the motors are switched.

*** is there any chance that the motors are not rotating in opposite directions when you are trying to move the gantry? Again, I don't run slaved motors so I don't know if mach3 takes care of this for you, but for a dual rack r&p gantry the motors must rotate in opposite directions or the motors will immediately try to rack the gantry (though I can't imagine that it would move very far down the table unless one motor was much stronger than the other).

I have qualify (4), since motor X pulling the left side of the gantry forward could look a lot like motor A pulling the right side backwards. You would have to measure the left and right movements to decide which (if either) motor or drive was working harder. Still not sure how Mach3 does slaving ...

[neilw20]

At the controller end...

SIGNAL and COMPUTER earth/common
All shields connect to ONE common earthing bolt for signal and computer earth.
All shields are ONLY connected at ONE end. This common point.
Don't run limit switch wires in the same cable as power wiring.
Don't daisy chain earths/ov/commons. All have separate wires to common earthing bolt.

If possible power the PC from the control box, and use the same common SINGLE earth point. Which one? Who knows.

If the PC is run from a separate power source, it is very possible that earth currents are flowing in the power lead.

The power wires for each stepper driver runs to a common point.
Don't daisy chain. Run direct from filter capacitor.
Use a single point 0v to common if you need to have one.

Run the negative and positive wires from rectifiers DIRECT to the filter capacitors. Do not connect any other wires to the rectifiers.
The rectifier wires carry large ripple currents.

ELECTRICAL EARTH/COMMON
Run an individual earth wire to ALL moving parts, gantries etc.
Connect your incoming power earth to this point.
Have ONE solid link wire to the SIGNAL common above.
Don't daisy chain common negatives to 'save' wires. Not smart in a CNC machine.

Do not use the metal of the machine to conduct any signals or power to ANYTHING. That always causes problems.

Stepper wires won't interfere with each other, but they can create havoc when near signal wires.

They may need shielding AT ONE END, and connect to one of the earths. Who knows which one is best. Depends how you wired it.

As little 100mm of wire wrongly placed in an earth system can cause severe problems.


Still got problems. Some photos of your wiring and how it all connects will invariably show up the problem.

[flyhigh]

Thanks everyone for the great info. I am working through the lists and actually progressing in terms of precision (increasing signal time in mach seemed to help), but I will hold off and give a run-down once I have fixed everything.

[ger21]

The step pulses are at a 4 out of 5 max in mach, I will set them to 5......

I believe the max is actually 15.