Re: [Pcb2gcode-devel] PCB Board Expectations

[Patrick B. <pbi...@us...>]

On Sun, 9 Dec 2012 21:05:11 -0600
"Ford Peterson" <fo...@hi...> wrote:

> Good evening,
> 
>  
> 
> I'm new to this forum.  I'm trying to diagnose some problems with this GCode
> conversion tool.
> 
>  
> 
> Using KiCAD for circuit board design and layout, I'm able to extract Gerber
> files.  Pcb2gcode converts them for running LinuxCNC under Ubuntu 10.04 LTS,
> which is the most recent stable version for running LinuxCNC.  My router
> table seems to be able to reliably hold a repeatable trace.  I have several
> concerns and am hoping this list can provide some reliable answers.
> 
>  
> 
> While the spindle on my CNC is light duty (a ball bearing dremel tool
> running a 0.1mm 60 degree V bit).  I have a new spindle on order and may not
> see it for another week or so.  Adjusting the Z axis to obtain a cut
> sufficient to remove copper, the narrowest insulation I can obtain with this
> 0.00397" V bit is about 0.030" using a micrometer.  Thinking this through,
> am I observing the run-out in the bearing at being roughly 0.026" ?
> Inspecting the spindle I simply cannot believe there is that much play.  I
> would certainly expect to visibly see 0.026" of play and I'm unable to
> observe ANY play in the spindle at rest.  While I was expecting some play, I
> was not expecting this much.  Are my expectations unreasonable?

I'm using a small proxxon rotary tool as well (quite successfully) and
a cheap, single-toothed bit. I noticed the isolation diameter changes
somewhat if the bit is turned in the collet, sometimes requiring
adjustment for boards with ICs in smaller packages (MLF).
Maybe the collet isn't that accurate, or the center axis of the ball
bearing is somewhat misaligned to the axis of the motor? Taking into
account how ball bearings work, the bit might even end up at a very
slight angle. I can't really tell as I'm not a mechanical engineer and
as long as it works reliably there's litte incentive to investigate.

>  
> 
> Using a test circuit called TraceTest I've installed nine SM1206 resistor
> footprints and connected them in series using trace widths ranging from
> 0.018" to 0.042".   I convert the Gerber to GCode with the following command
> line:
> 
>  
> 
> pcb2gcode --front TraceTest-Front.gbr --zsafe 0.08  --zchange 1 --mill-feed
> 6 --zwork -0.08 --offset 0.002 --mill-speed 20000
> 
>  
> 
> Even though the Gerber files were produced by specifying a 0.004" bit, a
> 0.004" insulating band, pcb2gcode insists on retracing the board a second
> time.  After a first pass, a 0.042" wide trace remains but only about 0.010"
> wide.  The second pass obliterates the traces and the pads.  I either do not
> understand the offset attribute or my spindle is simply a POS.
> 
>  
> 
> I do not see an attribute for bit size.  Only the -offset, which I've set at
> 0.002" (half the bit size)  It seems the second pass enlarges the insulation
> by taking part of the trace. I experimented a little with a negative offset
> and saw no real difference.

Do you have the remaining copper in the gerber file? The pcb2gcode
algorithm expects you to have only the actual traces in the file. It
takes your parameters (offset, extra-passes etc.) and tries to mill the
board as close to your specifications as possible in a best-effort
approach. If the remaining copper is in there as well, it will be
treated just like everything else.

I hope this helps, if not I can check your test file if you want.

-- Patrick