This is a very good article from Tweakie about the electronics (grounding) of CNC http://www.openbuilds.com/threads/electronics-101.693/

 

Ok, time for a few basic checks with tidbits I received from others here while trying to figure out my noisy/jerky steppers. Don't mind if some of these sound obvious, they just have to be mentioned / asked.

1) are you sure the jumper settings for micro-stepping are correct ? Don't forget the jumpers for the 4 axis (X, Y, Z and A - used for 2nd Y stepper by default) are around the CNC xPro board. So two of them are 'upside down'. Worth double checking.

To configure an axis of the CNC xPro controller to 8 microsteps, you should have jumpers 1 and 2 set, 3 empty. They are numbered (at least the version I have, before v2).

2) taken from the GRBL documentation online :

$100, $101 and $102 – [X,Y,Z] steps/mm
Grbl needs to know how far each step will take the tool in reality. To calculate steps/mm for an axis of your machine you need to know:

• The mm traveled per revolution of your stepper motor. This is dependent on your belt drive gears or lead screw pitch.
• The full steps per revolution of your steppers (typically 200)
• The microsteps per step of your controller (typically 1, 2, 4, 8, or 16).

The steps/mm can then be calculated like this: steps_per_mm = (steps_per_revolution*microsteps)/mm_per_rev

3) are the steppers wired correctly to the controller?

I had one of mine wired completely wrong (I blamed it on temporarily being colour blind) yet it seemed to work just fine. Call it luck ... it didn't even damage anything ! But how the thing managed to step was beyond me. No one ever explained it either. But it was real noisy ! It was not as noisy after that fix. It was probably more dragging along then stepping ...

Double and triple check your wiring connections. I eventually noticed one of the stepper's connections, just one wire, at the CNC xPro was loose ! So it would sometimes stop stepping for a fraction of a second and then go. Hardly noticed the effect on short move done manually (jog control).

4) Try to tweak the potentiometers (pots) for current to each stepper as "Under or over powering can cause noise". I had mine set too low for the NEMA23 I'm using (it varies between models/brands) But that was just as wrong as too much power. This fixed my noise problem ... thanks to ... Robert Hummel.

5) check the $N (startup block) of your controller. Just type $N, instead of $$, and GRBL will report back the current startup blocks stored in controller EEPROM. It should probably be blank (empty), but double check as it could be the needle in the haystack ... It could be running a G20 (switch to imperial mode, as opposed to G21 to go to metric mode) every time you start your controller or do reset. This without any indication since the Arduino based systems are stretching their processing power. So you might think 10 mm but your controller sees and does 10 inches ! To cheat with the other parameters will just bit you down the road. So double check ...

Also, along the same line, are you (or the sender being used) loading and possibly running some kind of job before you get to jogging?

If that 'job', which could be just to initialize 'things', does a G20, your controller will go into imperial and your 10 mm becomes 10" ...

I, , was puzzled for a while about this. I used a few different programs to generate G-code files. I was pretty sure I was working in metric within all of them, yet some generated G-code in imperial (!) The catch is that each time I would take control to job around, me though I was in metric and ... zoom ... my tool was flying off instead of moving a few mm at a time !

Man ... you pull that power cord real fast when you see your router head way too far way too fast for 'no reason' ! That's where the limit switches and emergency switch become important ...

Note, the jogging controls are usually in small steps. If you click multiple times, they add up into a longer move - normally smooth if you click fast enough (and depending on the 'sender' program being used - I use UniversalGcodeSender for Mac OSx). Just keep the tally in your head before you click too much and crash - been there, done that but still waiting for the t-shirt

6) if you send a G-code command like "G01 X10 F1000" as opposed to "G01 X10 F100", does your machine behave differently ? First command should move X axis 1o units 10x faster than 2nd. It may still be jerky, but it should move faster. It's like jogging, but you get to control feed speed ... Just don't forget to do negative moves from time to time.

7) What is your step pulse setting again ? That's the $01. It's the number of microseconds between pulses (steps). If it's too long, the steppers are probably going to freak out on you ...

8) check the set screws on your pinion and stepper shafts. I had a few that were or became loose with time. Not certain if it related to my noise problems, but it did wonders for precision.

That should be plenty to check for now. I really need to get my "do's and dont's" out, eh ? I think I did just about every error one can do building an OX ... yet there seems to be plenty of room for me to make more errors

 

That's why there will be some tweaking once you get close to right with the basic config. You will need to do the tweaking with long moves to get maximum precision (actual move to requested move).

But first, let's get your OX to move 'about' right ...

 

Serge, don't confuse the guy, now. Once he starts with "I'm still 0.15mm off, despite everything is set right" we'll get to that one.

 

Really gonna second all the replies from Serge and Prauk on this. It takes some time and a lot of patience to get the software locked in - but you really need to focus on the moving about right portion and then tackle the scaling. Serge has done a great outline of the most common issues (Thanks Serge!).
I would only add:
1) Confirm that the gantries move freely by hand when there is no power to the xpro (X and Y should be very easily and through the full ranges of motion). Look for any tight spots and/or click along the way. Also watch your belts on the pulleys here - your belt should stay nearly centered the entire time.
2) square the machine as much as possible
3) Check that your belts are not too tight - this is kind of a gray area as you want the belts tight to reduce backlash but with the belt and pinion design you are applying a radial load on the motor shaft as well the rotational load - in a standard rack and pinion you will only have the rotational load.
4) A print of your settings (send $$ command) would be very helpful in troubleshooting.

 

Some sound advice there serge, glad I'm not the only one who has had issues with the grub screws.

What slightly baffles me is why you run your nema 23's on low to avoid noise? The output from the xpro is around 2a so surely you need those cranked right up? I lost steps under cutting load running that low. Depending on the model mine are rated 4.6a so I run mine at 4.5. Now doesnt loose steps, jogs without clicking, runs quieter and strangely enough, cooler.

I think if your having to back off the current to avoid noise, you have another issue there.

 

I think the clicking is due to stepper current too high. You might need to back off the current on the xPRO board. Mine are set at about 50% or so now.

 

If your steppers are working and the current settings make no difference (which would be strange) most likely it has something to do with your software settings. Check if there is anything there that could cause this kind of behavior of the machine.

 

tryed just changing the acceleration? Also Can you test the voltage of your psu and tell us what it is? The xpro converts voltage to current so if the voltage is too low you will get similar issues.

You say at present you don't have a ground connection. Is there anywhere on the xpro you could connect one? Is your pc grounded? If so do you have a laptop you could try using without its power supply plugged in? Current and settings sound to me as they are not really having any effect on the issue. Could be that the current max is still too low for that not to have an effect or you have a wiring issue. Bad ground will have this effect, to rule this out you need to either add a ground or remove all grounds with use of a laptop.

Also What nm 23's do you have also. The xpro won't power the larger nm versions. And current = torque so the xpro output is 2a so really at full power you are only running the nema's at half capacity. You have got to match the current to the motor. People saying otherwise are mis-informed afraid to say.

 

Actually, I think that power = torque. Power is measured in watts, formula V*A=Watts. So, at 12V a current of 2A will give you 24 watts, while at 24V 2 A will give you 48 watts. Hence, the higher the voltage with a given current, the more power you can have out of it, i.e. more torque.

Of course, if you allow more current to flow through, you'll have more power (torque). Running the stepper motor on less than optimal power will, in this stage of testing without actual cutting, most likely not cause this run/stop/run/stop/run etc. behavior.

As being pointed out in Tweakie's excellent article about electronics, it is of paramount importance to use shielded cables to prevent interference in the system. The shields need to be connected to a common ground, as do all other equipment in contact with your OX. Especially signal cables, like the ones you use for limit switches, need to be shielded or they could trigger at random because of the interference of other electrical sources. The shields need to be grounded only on one side (inside the box where the electronics are is most convenient place, I would say).

In your situation, this could be the case. You can find out if this is true, by observing the behavior of the stepper while you move it. If the run/stop/run sequence looks random, most likely you have a shielding problem or bad connections. If the sequence is very orderly (like 2 sec run, 1 sec stop, 2 sec run etc.) it must have to do with a numeric source, most likely your software.

So, test and observe to get an answer for where to look.

 

Same thing let's be honest, current = torque.

But you are right voltage is needed to create current hense why I asked what his psu is outputting.

 

Does the new power supply have higher voltage, or is it the same voltage but just another one?

If you can find them, I have good experiences with ATX's on my OX. You can easily put in series 2 or 3 of them to get 24V or 36V. It also gives a kind of redundancy. If one fails, the other will keep supplying on lower voltage. You can keep working until the broken one is replaced. Their generally also easy to get in the local PC shop.

If all is ok now, maybe it's a good idea to tune in the machine by measuring the real travel distance and adjust the steps per unit (mm/inch) to get it more accurate. Also check if you have any backlash and try to compensate for it in the software (or adjusting the belts). If you do that before making some serious cuts, the result will be much nicer and make you feel even more proud!

 

Glad you got to the bottom of it. Persoverance pays with these things.

Now you may want to test for scale and lost steps, I mounted a marker and ran a program with squares, circles and diagonal lines to do this.

 

Just had to chime in on this "what is torque" issue.
The clue is in the units:
Torque is measured in ft.-lb., oz.-in., Nm.....
Power is measured in watts, horsepower.
Power = torque/sec

Motor torque is produced by magnetic flux.
Flux strength of the motor windings is proportional to current flow and number of turns in the winding.
The faster a motor turns, the more back emf it generates. Therefore you need higher supply voltage to push the same current.
Motor torque does not increase with speed if current is constant.
Power = torque * rpm.

 

So moral of this story is low voltage or wattage and subsequently low current also causes chattering.

 

Mine are rated much lower and are working much better with the toned down setting. They are configured as bipolar ( which means they can work at the lower current (2.12 A) then when they are configured unipolar (3 amps) - with some disadvantages elsewhere as some might point out (and I forget at the moment as I catch up on things at this end.)

As you say, it varies with each stepper model and how they are wired ...

 

No worries. I am in the computer industry for 30+ years and this DIY CNC stuff had me for a spin for as long, if not longer, than what you just gone through. In my case, it was the physical aspects which came in as a handy excuse for my head scratching over myOX.

 

The drivers on the controller are most likely over heating thus taking pauses to cool down at which point they start driving the steppers again. I lost quite a few hair over that one for longest of time .... Things worked great if I used the jog controls but sending G-code files started to give stuttering myOX. I was "slow", drivers had time to cool down.

In my case the problem had two causes : I had the drivers set at maximum power (amps set by pot) and I was not using heat sinks NOR a fan. It was middle of winter, the garage was cold enough to get me sick a few times over, but those tiny driver chips would of gone for thermal meltdown if the built-in safety didn't shut them down for me.

I used heat transfer grease, stuck heat sinks on each of the four drivers and, as I found one down on worktable and an other sliding off, I added just enough clear silicon in opposite corners to 'glue' each of them in place... The silicon seems to hold under the high temp so far. Ran a bunch of jobs running over 60+ minutes with no issues ... other than my work surface and X axis not being quite parallel to each other - 2 mm off over 1500 mm width.

 

It's crazy they don't sell them with the heatsinks, they obviously need them to work at all by the sounds of it. That the same issue with the xpro? And no warnings on them saying fit heatsinks before use?

 

For the cost of the heat sinks it should be a given with any controller.

I believe the xPro does have a little something about probably needing the heat sink if running full power ... I forget, been too long.

I paid something like 5$ for total of four with shipping a while back, but seen them at lower cost since (half !) Bulk price should be lower still.