Not getting any adjustment?

[booski]

Hi there, Nigel here.

I've just started up my machine to route an adapter board for a 128x64 display.
Using Linux CNC 2.5.2, I'm getting a very strange cut.

For example sake, lets say I probe the front of the board and I call it Z=0mm. If I probe the middle of the board it is Z=-0.177mm and then probe the back of the board it is Z=0.1mm so the middle is lower than the front and back edge.

The autoleveller code adjusts the height between the front and back but not for the middle, so from front to back, the machine raises Z 0.1mm, but doesn't lower the tool for the middle.

This means it is cutting the front and back but not the middle which is what I thought this software was meant to do?

It's almost like the it can only adjust for small changes in height but not large changes.

[booski]

Just had another go and getting really frustrated with it now.

On the Y axis, the autoleveller only makes adjustments between the front and rear probe areas but not in the middle hence why it cut at the front and back of the board, but on the X axis it does make adjustments.

This is Y axis specific too, it's as if on Y axis movements the autoleveller completely ignores all the probed areas apart from the front and back.

Only when it had to cut round a pad where there was X axis movements did the autoleveller adjust.

Would explain why I've had a lot of trouble in the past with this too.

I would upload the G-code files but it won't allow .txt or .nc extensions

[daedelus]

Hmmm, this needs investigating ASAP. Thanks for pointing it out.

My initial thoughts are that the interpolation algorithm I am using might need changing.

I have enabled .txt and .nc extensions now and you should be able to attach

[booski]

ALTop.Part1 [Profile1].nc

AL File

(49.61 KiB) Downloaded 826 times

Top.Part1 [Profile1].nc

(7.38 KiB) Downloaded 744 times

Attached are the 2 files, original and the AL file.
You might want to add the NGC extension also because AL outputs NGC so that'd be helpful also

I don't know how the interpolation works, but it made no effort at all to cut the board at the very lowest part of it, simply traversed the 0.1mm from one edge to another.

By all means have a look at the codes but I suspect that they aren't going to tell you a great deal.

[daedelus]

Cheers, I shall check it out tomorrow. I have my suspicions about the problem and I know where to begin looking.

[booski]

Smashing, give me a heads up once you've got some news. It's the first board I've tried to machine in ages so it's frustrating! I've always wondered why previous boards didn't come out 100% as expected, especially larger simpler boards.

I've just ruined my CNC Computer anyway, tried upgrading from Ubuntu 8 to 10.04 and now it won't boot, so I'm a little stuffed anyway and will be back to strip board tomorrow

Cheers!

[booski]

Well, I've managed to get my Linux machine up and running again now, so, perhaps I'll give it another try but I'm not hopeful.

[daedelus]

I have checked the algorithm itself and it seems fine and I cannot see any problems with my code.

To be really thorough, I will need to do a full test on my machine (I have done this before many times but maybe I missed something). The machine is not currently setup since my last house move and I have spent most of today rigging everything up so I should be able use it tomorrow.

For the test, I will log everything at every stage of the mill so I can really see whats going on.

Writing code for and testing CNC programs is very challenging since testing usually involves milling something which is expensive in terms of materials and time. So I hope you can appreciate this and please be patient. If there is a problem, I will fix it.

thanks,

[booski]

No problem, I understand that it isn't easy, especially considering that you are offering the software for free.

I have however been a little unfaithful and have found this:

http://marcuswolschon.blogspot.co.uk/20 ... lling.html

Very similar little program, much more basic. Have to manually edit the output code but I've just gotten near perfect result from it.

[daedelus]

This looks very basic as you say but he has released the source code too, which will help me narrow down the problem no doubt.

thanks,