PDA

View Full Version : Random Z-Axis plunge



gomako
14-05-2012, 01:52 PM
Hi everyone,

I'm new to CNC machining and a friend and I have bought a chinese cnc machine from ebay ( like this one http://www.ebay.co.uk/itm/290704531819?clk_rvr_id=342579254146&afsrc=1#ht_2396wt_934 from the same seller) I don't really know exactly what model it is, there are no numbers on it or anything (I know, I know apologies for being a useless n00b)

I've been using the machine on and off to make various bits and pieces and am running into problems with the z axis, every now and again (but often enough to be problematic!) the Z axis will ram down by about 10mm, usually snapping the bit and wrecking the piece. I have run the same g-code and it has worked fine, so I am assuming it's not a problem there.

I had a look through the forum and found this post :
www.mycncuk.com/forums/showthread.php/1395-
Z-Problem but it's from 2009 and wasn't resolved. He was having the same problem as me.

I've tried tightening the coupler which seemed pretty tight anyway, so I am assuming the motor is over running for some reason.

I guess by looking around the forum that it's probably the controller? The problem is I really don't know what I am looking for to test/replace/fix it. I don't even know which one of the controllers it is! Is it a controller or a driver that I mean (there are 3 of them with dip switches on)?

Any help and general advice (links to other posts etc.) would be very gratefully received. I can't leave the machine to do its thing without hovering a few mm from the stop button!

Thanks.

JAZZCNC
14-05-2012, 04:25 PM
Ok first lets clarify the difference between controller and drive. The controller is the software that commands the drives which run the motors. Each motor will have a drive (The things with dip switch's on).

I'm pritty sure it won't be a Drive problem.! Drives usually just work or they don't. . . Same with motors.!

It's more likely 99% to be either a lose wire or electrical connection, partly broken wire etc.? . . .OR . . .It could be electrical interference "Noise" from some other device thats nearby.?

"Noise" can be do all sorts of crazy things to a machine,false E-stops, twitching motors over running motors etc and it can be a night mare tracking it down.
Often it's because of poor grounding or earthing practices within the controlbox or just crappy electrics in general.
It can come from simple things like the radio or washing machine or the actual CNC machine it's self. The VFD from the spindle is Num1 candicate for noise.

I wounce had an issue with noise tripping my machine at random times, it went on for months. I went for all the usual suspects, dodgy connections,signal wires running too close to power wires even replaced relays that looked Iffy but still it happened.???
Then one day my son noticed that it always seemed to happen the same time the compressor shut off, OFF not when starting.? The compressor was in a box outside but just the other side the wall to the CNC machine.! He was correct it was the contactor which was on it's way out and producing noise, changed it and problem went away.!!
It had to be noise because the compressor was on a totally different circuit because it was such a big compressor so needed lots of juice so it wasn't feed back thru wires but the machine was close enough to pick it up and trip the E-stop.

Now I'm not saying yours is noise becasuse I'm pritty sure you have a lose or partly broken wire, or trapped wire.!
My advise would be start with the wiring and check every wire on the Z axis drive,motor and BOB connections then along it's length for pinchs or crush marks, tight spots in the cable chain are favorite spots, entry exit points etc.

It won't be a drive problem I'm fairly confident of that and I'm sure it won't be a software issue if the codes run ok before. (Unless there's a known Bug in the control software.?)

IF still doing it after all that checks out ok then I'd look to the control box it's self and the wiring grounding conditions.! . .Things like:
Does it have shielded cable for all signal wires IE: to motors, limit switchs etc.?
Does it use Star grounding.?
Does it or the components even comply with EU regs.? Could be sub standard and poorly made easily interered with noise wise or even produce noise.?

Good luck and keep us posted.!

John S
14-05-2012, 04:41 PM
It could also be missing steps because the Z axis is too weak / tight or trying to move too fast.
What happens is you are 10mm off the work, everything zeroed and command a Z-10 move which should put it on top of the work.
Then you command Z0.0 which should go 10 mm above the work but because of missing steps it finishes up say 5mm above the work.
When you then command the next Z-10 move it bungs the cutter 5 mm into the work.

Check the Z axis for freedom and binding and also reduce the acceleration and velocity in motor tuning.

JAZZCNC
14-05-2012, 04:54 PM
Check the Z axis for freedom and binding and also reduce the acceleration and velocity in motor tuning.

I think it's been working ok on same piece of code then just happens John so I kind of ruled missed step out has it would happen pritty much all the time if overtuned.? Or at least move some random amount.?

The little 3020 chinese bag of shit that I have here had a similiar issue in that it would change direction on its own and other crazy stuff like move down too far or not stop moving when expected. Turned out to be cheap nasty wire to Z axis motor changed it and it's sorted.

Seen has this is a chinese machine and possibly made in same place then it wouldn't surprise me if it hisn't a similiar issue.? Certainly worth a look or changing the wire for sake of £10-20.?

gomako
14-05-2012, 05:02 PM
Thanks for the quick replies guys! Thanks for all the info jazzcnc, I have a lot to learn.

I have followed your advice and I think there might have been a couple of solutions in both posts.

I had a look and some of the connections in the terminals of the drivers were looking a bit strained, so I pulled the cable back to relieve the stress. Hopefully that'll minimise any noise.

I also had a look at the movement speed that was set on the tools ( the guy that we got the machine from gave me this advice too after I'd posted here ) and the plunge and movement speed was set pretty high, and could well be causing the machine to miss steps as you suggested John. It does seem odd that it was pretty much exactly 10mm every time, so what you say makes sense.

Having done all that, the machine seems to be behaving itself! Thanks very much! I'm looking forward to being part of the forum.

gomako
14-05-2012, 05:11 PM
I think it's been working ok on same piece of code then just happens John so I kind of ruled missed step out has it would happen pritty much all the time if overtuned.? Or at least move some random amount.?

I assumed that too, but it's usually going wrong right at the end of the job when the tool is being lifted right up out of the workpiece, moving slightly and then zooming back down again (when we are cutting in several passes - we still have to get our techniques right!) so it's when it moves a up and down a (relatively) long way in a short space of time that it goes a bit strange.

Thanks again!

JAZZCNC
14-05-2012, 05:16 PM
Did it do it every time you ran the same code at the same point in G-code or occsionally and randomly with the same code.? If the later then I'd be surprised if it's missing steps.
Esp has it's exactly 10mm, missed steps from overtuning tend to be random amounts resulting in random offset distances.?

Seen has it's a constant figure It could also be backlash.? Backlash would give exact offset distances.!
Check all mechanical connections on the ballscrew, you should if built correctly have a fixed end where the ballscrew comes thru the thrust bearings and is fasten with a lock nut to stop end float. If this nut works lose(which does happen) it will have the same effect has backlash and give a constant movement amount when changing direction.!

EDit: The movement amount can be tiny but it accumlates with lots of moves resulting in an exact amount at a set point in the job so check carefully because even a tiny amount is bad news.!!

It's the constant 10mm amount that concerns me, missed steps thru overtune don't often give same error amounts.?

JAZZCNC
14-05-2012, 05:22 PM
I assumed that too, but it's usually going wrong right at the end of the job when the tool is being lifted right up out of the workpiece, moving slightly and then zooming back down again (when we are cutting in several passes - we still have to get our techniques right!) so it's when it moves a up and down a (relatively) long way in a short space of time that it goes a bit strange.

Thanks again!

Ahh ahh ok then you could well be overtuned or under powered on the Z axis like John said.

Either way if you have backed it off and it's working ok now then your sorted and happy days. . :)

Jonathan
14-05-2012, 07:19 PM
I'm sure I posted in this thread after the original post...must have not sent :(

If the error is occurring it the same point in the G-code, which would appear to be the case, then it's likely to be a tuning problem. The error would be constant since it's the same command causing it each time. Assuming the Z-axis stalls at the same point then the error will be repeatable. Try just running the few lines of G-code round where it stalls and see if it happens again.

I very much doubt it would be backlash since errors due to backlash are not cumulative and 10mm of backlash would be rather silly!

A couple of weeks ago I had a similar problem in that my machine wouldn't get past a certain line of code in a spot drilling pass. There didn't seem to be anything special about it, just a rapid move in X and Y. However after noticing that both X and Y would stall every time the machine got to that line I tried just running the same line and line before on it's own - sure enough it still stalled. Naturally checked all the wiring and the usual stuff, which made no difference. Reduced the rapid feeds on X and Y which worked only if I went to about 60% of the normal speed which is a bit pathetic. Tried running the code zeroed at different points and it still stalled so couldn't have been a mechanical fault.
In the end I decided to switch to LinuxCNC and a couple of hours later the machine was running just fine at full speed.

gomako
14-05-2012, 09:19 PM
Thanks again for all the info.

I don't think it was a specific line in the G-Code. When I was messing about with it trying to diagnose the problem, I made a small square cut-out with some depth clearance, bridges etc. Then I'd load the G-Code into the control program (NCStudio in my case), run the code. When it finished (usually with no problem), I'd manually move the bit and reset the origin and just run the code again so it would make the cuts again on another part of the scrap board and more often than not, but randomly, I would have the crazy plunging problem. It was roughly 10mm, not exactly 10mm every time too.

I couldn't re-create it and looking at the G-Code, it didn't look like there was anything obviously wrong ( I am a php programmer so am familiar with what code should look like ) there.

@Jonathan, could you tell me what backlash is, and what causes it/what to look for? Also, do I have to use the control program that came with the router, or can I use anything (like LinuxCNC) to control the machine? Is there a machine profile or something like that to set up the control program to work with the router? I don't really like NCStudio, and I think it's pretty buggy.

Jonathan
14-05-2012, 11:55 PM
@Jonathan, could you tell me what backlash is, and what causes it/what to look for?

Backlash is lost motion when the motor driving an axis changes direction. For the majority of machines there is some clearance between the nut (ball-nut or trapezoidal) and the screw, so when the screw changes direction the nut wont move until it has moved the distance of this clearance. Better explanation here:

http://www.machinetoolhelp.com/Repairing/What_is_backlash.html

If backlash is significant you'll for instance see it if you cut out a circle - at the limits of travel (so 4 points) there would be a slight blip. However on a ballscrew driven machine generally the backlash is so small (<0.05mm) that unless you're cutting metals it wont be noticeboard.


Also, do I have to use the control program that came with the router, or can I use anything (like LinuxCNC) to control the machine? Is there a machine profile or something like that to set up the control program to work with the router? I don't really like NCStudio, and I think it's pretty buggy.

It looks like NCStudio uses a PCI card to control the machine:

http://www.erel-cnc.de/html-uk-cnc/software-uk.html

So assuming that's what your machine uses, unless there are drivers for it for Mach3/LinuxCNC, which seems unlikely, I doubt you can use any other software. That's unless you remove the card and change to a parallel port system, but that's downgrading if anything. The PCI card should be better.

gomako
15-05-2012, 07:38 AM
Ahh, I see. Thanks!