Welcome! Log In Create A New Profile

Advanced

Losing Calibration

Posted by Anonymous User 
Anonymous User
Losing Calibration
March 15, 2010 05:49AM
Hello,

First of all I would like to thank JJ for creating a very nice and pretty easy to understand plug-in, Thank you very much.

I would like to ask the people on this forum about the following. When I run a PCB with 1 pass there is no problem. When I cut multiple passes I appear to have the problem that my CNC machine is not cutting right next to the cut that's already there. This results in a PCB that's not useable at all because the machine sometimes cuts the existing traces.
Does anyone know to solve this problem? And is it an EMC2 problem or PCD-GCODE problem?

I use a 0,4mm V-bit on a CNC-Sable machine.

I could be that I should calibrate something in EMC2 but for no I just have no clue.

Thanks in advance,

Rob.
Attachments:
open | download - P1040146.JPG (2.87 MB)
Anonymous User
Re: Losing Calibration
March 15, 2010 11:56AM
Looks like you are cutting way too deep. Try cutting at about 0.1mm. Remember, you only have to cut deep enough to cut through the copper which is about 0.25mm deep.

It appears that everything hardware, and software, are running fine. After you reduce the cut depth, using PCBGcode, cut a new board, measure the width of the cut using a micrometer, then put that measurement into the 'tool width' in PCBGcode. They you should be fine.

See the below board I just made using a 60° V bit and a cut depth (Z-axis) of 0.1mm. The track is 0.15mm wide.
Attachments:
open | download - IMAG0082 (1).jpg (537.5 KB)
Anonymous User
Re: Losing Calibration
March 15, 2010 03:08PM
Thanks for the kind words Rob!

The easiest way to tell where the problem lies, is to split it in half. In this case that would be at the .nc file. You can view it in a simulator (like CNC Simulator or NC Plot) and look for the shifting you're seeing. (You can also use the previewer in pcb-gcode, although that isn't a true test, since it does not actually use the .nc file.)

In my experience, it is usually the machine losing steps. The stepper motor, say, is supposed to go 20 steps, but resonates and only actually moves 10, then everything is shifted by that amount. This can sometimes be corrected by going slower, cleaning and lubrication, accelerating slower, etc.

Let us know what you find, or post any more questions you have, here or on the Yahoo! group mailing list.
Anonymous User
Re: Losing Calibration
March 15, 2010 03:09PM
Nice looking board Steve!
Anonymous User
Re: Losing Calibration
March 16, 2010 01:51AM
Dear John, Dear Steve,

Thanks for the advice to cut less deep. I will have a look at that!

John, thanks for the suggestion to cut the process in half. And to look at the nc file with a viewer. I use EMC2 and it has its own viewer when I load the nc file. That appears to be perfect. EMC2 is giving an error about latency and I've been working on that yesterday afternoon. I now decided to try Mach3 instead of EMC2 because of this.

Like you describe, it looks like one of the motors is losing steps. I think it has something to do with the latency. I will look at that tomorrow.

Thank you very much for the information and I will post my findings here!

Rob
Anonymous User
Re: Losing Calibration
March 16, 2010 03:38PM
EMC latency may cause You to snap the tool or have less than perfect milling outcome.

There are a couple of fora ( forums ) which can help You with the latency issues.
One is the EMC forum on CNC zone the other is at LinuxCNC.

Aometimes the latency ussues can not be eliminated as they relate directly to the mother board performance.
Anonymous User
Re: Losing Calibration
March 17, 2010 03:45AM
Hi Miller,

Thank you very much for your Reply.

I found out that there are indeed a few problems that can occur when using EMC2. Do you, or anyone else, know how MACH3 handles this?

To my experience Windows handles multi-tasking pretty well so I expect to have better results with MACH3. But I can not really compare because I have no experience with Linux what so ever.

I'll keep posting.
Rob

EDIT:
Just installed MACH3 and made my first succesful PCB in a piece of MDF(for testing) Now running the second one which is the print EMC2 was not able to finish.
It looks like I have no delay problems while running MACH3 and that's very nice. I thank you guys a lot for your replies.



Edited 1 time(s). Last edit at 03/17/2010 08:30AM by Rob.
Anonymous User
Re: Losing Calibration
March 17, 2010 03:48PM
I am using EMC2. No experience whatsoever with MACH3.

I need to say that tothe best of my understandidng both EMC2 and MACH3 run the same basic NIST released public domain software.
MACH3 runs on top of Windows which has been modified in order to be able to run real time.
EMC2 runs on Linux kernel compiled with realtime library. The implication is that EMC2 is a real time application.

As such it needs to run on hardware which has low interrupt service latency. The lower the latency the closer the software runs to schedule if i can put it in these terms.

this departure from the exact timing can be influenced by peripherals attached to You mother board as well as BIOS settings.
Once configured for minmum latncy value the PC will serve You well with EMC2.

NOTE .. you must have the realt time instalation. Standard Ubuntu will not run correctly .
Anonymous User
Re: Losing Calibration
March 18, 2010 03:02AM
Hi Miller,

I've made a new PCB(2 actually) and still have the same problem. I made a picture of it. If you would take a look at it, I'd appreciate it. In the red circle in the picture you can see the problem that occurs. The line that are 'more to the left' are 'wrong'. Now the weird thing, in the green block you see a shape like a weird shaped '8'. This double-lined 8 shape is cut after the wrong lines in the red circle.

To me it looks like the machine looses caliration and than find it back again. I have no real clue. In pcb-gcode viewer it looks great, and also in the viewer of EMC2 and Windows MACH3 it looks cool.

I took a look at the latency/interupt thing. But what I wonder is the following; If there is a latency which makes my steppers loose steps. Why is the Yaxle perfect? The problem only occurs at the Xaxle.

If you have any idea, or someone else, I would be very thankfull.

Rob.
Attachments:
open | download - P1040148.JPG (2.51 MB)
Anonymous User
Re: Losing Calibration
March 18, 2010 08:48AM
Ok. It looks like my problem is solved.

I've changed two things.
-Turn of the screensaver and the 'harddrive to sleep' after 20 minutes.
-Adjusted the default and maximum setting in pcb-gcode-setup.

So I'm not sure what exactly was the problem. but it looks like one of these two.

I just made one correct PCB and will start to slowly add the features I did't use for this one again. Like spot holes and driling the PCB. That's all by hand at this moment.

If I find solutions for my probs, i'll post them here.

Rob.
Anonymous User
Re: Losing Calibration
March 19, 2010 02:57PM
I do not think You have a problem with either EMC2 or Mach3.

It seems like You may be loosing steps ( i assume you are running steppers ) or You may have a loose or too compliant a coupling along the one axis.

What is the high speed traverse speed and what is the linear cutting speed ?

Also as a check locate the gcode optimiser ( in this forum) and download the optimising software.

Optimise the g code file ( of the above graphic) and run it with old settings.

You may find that thedouble lines whicha re now mis aligned disapear or move elsewhere, The effect of optimisation is to cut spatially adjacent data . so any loss of step or slack between motor and screw may be minimised although it might mean getting artwork which is somewhat elongated or compressed along the one axis.
Anonymous User
Re: Losing Calibration
March 22, 2010 05:50AM
Dear Miller,

I think you're right. I'm pretty sure you're right! I am running steppers and it looks like I'm loosing steps. The odd thing that's happening is that I only seem to lose steps on the x axis.

I cannot find the gcode optimiser that you mention but I'm very curious.

I made a succesfull PCB last week. We need 2 more of exactly the same PCB so I used the exact same file. Last week I used a feedrate of 150mm which was rather fast so in Mach3 I set the feedrate to 70%. This made a mess of the whole process, I destroyed a bit and a blank PCB board.

So up to 100% again I got the exact problem you mention:
You may find that thedouble lines whicha re now mis aligned disapear or move elsewhere, The effect of optimisation is to cut spatially adjacent data . so any loss of step or slack between motor and screw may be minimised although it might mean getting artwork which is somewhat elongated or compressed along the one axis.

Do you, or anyone else have an idea of how to solve this?
Is it about getting a different PC. New Windows installation, optimize Windows to be more dedicated to Mach3?? I have no clue... maybe it's the macine settings in Mach3. Would it help if I put a screenshot up here, so you can have a look?

I put up a photo as well so you can see what happens exactly. The red line indicates the loss I have at the moment.

Thank you very much!

Rob
Attachments:
open | download - failedPCB.jpg (311.2 KB)
Anonymous User
Re: Losing Calibration
March 25, 2010 01:52AM
Rob,

First of all let us look at the 70% issue, the 150mm( /second ?) feed rate issue and what is good for Your process.

Most FR4 materials are rated at 180 metres /second cutting speed. This figure is the property of the material. This is the speed of a blade moving in a linear fashion just scraping the surface of the material.

This figure is then used to calculate the infeed rate of a rotaing tool given the chip load ( another concept ), the radius of the tool and the spindle speed of the tool. Chip load is a notional depth of "scrape" of a cutting edge of the rotating tool. Some manufacturere specify it as a percentage ratio of the tool diameter others sate a figure for example 0.001" or some such equivalent metric format figure.
Suffice it to say there are tables and charts of tool diameter versus spindle speed verus linear feed of the rotating tool.

There is a fellow who is active in these spaces who sell a very fine twin flute ( two cutting edges) V groove tool for not much money ( and so far he has not seen too much money from me because he sell a tool whihc seems to last) and the tool seems to like to run at about 40 000 RPM at a linear feed rate of about 20 to 30 mm/second.

ROB if Yor spindle runs slower than 40 000 RPM then PROPROTIONATELY reduce the 30 mm/second figure. If spindle does 20 000 RPM set Your feed rate to 15 mm/second.

You can not run Your machine feed rate at 100 mm / sec if spindel does 10 000 RPM.

NOTE here I speak of the coordinated motion feed rate or in G code parlance the G01 and G02 feed rates.

Now to the lost steps issue.

Stepper motors are problematic if not driven correctly or if driven too fast. They will loose steps and like someone doing a silent but deadly in a lift will simply not apologise.

Servo motors are better behaved in that respect.

You may have to play around with stepper motor settups in both EMC2 ( there is a stepper motor settup wizard whihc asks for stepper driver characteristics so as to configure maximumstep rates ) and i am sure MACH3 also puts its pants on one leg at a time.

Now different axes have different loads to move about and therefore the maximum step rates need to be determined where various loads will not cause lost steps.

I would suggest you reduce maximum step rate for high speed traverse movements to something a bit slower and retry the milling proces. High speed traverse steps are UNCOORDINATED and in G code speak theya re moves associated with G00 moves.

You will lose in terms of air cutting speed ( ie slower G00 ) but You will reclaim repeatability of product Youmake.


So to recap:

set coordinated moves speed in line with spindle and cutter capability ( G01 G02 G03 )
set uncoordinated moves in line with machine drive capability ( G00 )
Anonymous User
Re: Losing Calibration
March 25, 2010 01:57AM
Rob,

check this out [pcbgcode.org]
Anonymous User
Re: Losing Calibration
March 29, 2010 02:56AM
Hi Miller,

WOW! Thank you very much, even if this information doesn't solve my problems, I'm expanding my knowledge on this subject.

For now I'll first reply this thread, then run my PCB's again and come back to this forum.

Let me see if I get this right.Generally, I'm trying to cut to fast. I'll put my feedrate's down to the numbers you suggest. As I understand I'll lose a lot of time with air-cutting because of this. The gcode optimiser you mention will bring this time back by reducing 'air-time'.

I contacted my supplier about my problems and asked his opinion. He said I should try another computer because that could be the problem. He also sended me a new steppermotor. This is because I only have a problem on the X axle. I'm not sure abbout this, but it's worth trying. My supplier runs exactly the same machine so I asked him for his settings and they're the same as he provided in the papers that came with the machine.

After the testing with the steppermotor replaced is not succesfull. I'll propably buy another PC. One that my supplier delivers and which is guaranteed to work with my machine and Mach3.

Thanks again,
Rob
Anonymous User
Re: Losing Calibration
March 30, 2010 04:05PM
Rob,


In the earlir post I only mentioned speed.

You also need to check the acceleration.
Anonymous User
Re: Losing Calibration
April 07, 2010 03:51AM
Hi Miller,

A lot happend since my last post. I switched 3 computers because my machine as well as my gcode looked alright. I had the same error on all the machines. I first thought it was because of latency but my error only appeared on the x axle. I contacted my supplier and after a long search he changed the motordriver.

When I moved the spindle from left to right at a high speed I could hear the sound(high tone) sometimes interrupt. With the new motordriver this is gone. So it all appeared to be a hardware problem.

Thank you very much for the information which was usefull in anyway!
Rob.

EDIT: Do you know where I can buy good, reasonable priced, V bits?



Edited 1 time(s). Last edit at 04/07/2010 04:10AM by Rob.
Anonymous User
Re: Losing Calibration
April 07, 2010 04:56PM
CNC Zone has a section speciffically devoted to PCB milling.


Here is link to one of the discussions on which is the best bit... [www.cnczone.com]


So far the "best" I could find is LPKF but it is prohibitively expensive. Next best is a tool supplied by precise bits ithink.. model number EM2E8-0625-60V at under USD10.00 a broken tip smiling smiley
Sorry, only registered users may post in this forum.

Click here to login