A great deal of what we see on the Web, project-wise, is the ‘finished’ product. Every questionable decision has been ‘revisioned’, and what warts remain are carefully hidden, or presented as ‘features’. Yeah, I know this is a pretty cynical view, but unfortunately, nearer the truth, than even I would like to admit.
One of the great lessons I found, in a long career of software engineering, was to document every ‘failure’, every ‘good’ idea that turned out to be just plain bad, and attempt to make this collective wisdom available to the whole software group, so that when new ‘brooms’ came along, the same tedious business of making the same mistakes again was avoided. Well, it didn’t always work, but most of the time it proved very useful.
I could have documented my experiences alongside the Project Development posts, but I feel that would be the wrong place to do so, hence this separate diary.
Yesterday, I cut my first traces ‘in anger’ on a piece of copper-clad circuit-board. This was two copies of the bottom pads of the small HV_OP_AMP project I’ve been using as a test-set for the Proxxon MF70 CNC Conversion Project.
The results of this were encouraging.
This morning I cut the top trace of the same small board. Again the result is encouraging, but not without issues. So, rather than simply quietly sorting out these issues, I thought it might be a good idea to share these with others, contemplating similar projects.
1st take a look at the photo of the bottom pads, cut yesterday. (on a scrap piece of FR4 board)
And now, the bottom foil, together with a blow-up print-out of the foil as it should be.
Now, in my opinion, the upper photo shows excessive depth-of-cut, using the ‘default’ values in the pcb-gcode.ulp program. This sets the Z depth to -0.254mm, in Imperial units that’s 10mil. Considering FR4 board with 1oz copper is quoted to be around 2.08mil thick, 10mil seemed excessive. So some juggling required there. Also, as a result of excessive cutter depth, and the cone profile of the 0.5mm cutter I used, some loss of area can be noted around the pads – especially the small link pad. (The Mickey-Mouse ear of the pad 2nd-row, 4th from left) Despite these comments, these are matters which can clearly be cleared up with attention to the set-up.
Now moving to the 2nd photo, this too, has several issues. Ist of all, I reduced tool penetration to 0.102mm – or 4thou in Imperial measure. Removal of copper was still retained, but the setting showed up the sensitivity to a non-flat (slightly-convex) circuit board, in that the area closest to the left-hand clamp has not been completely cleared of copper where required. This led to short-circuits etc. Again, this is a matter of setup, and one I’m not unduly concerned about.
I marked the traces I was happy with on the blown-up print-out with small red ticks. As you will see a large percentage of the board I considered satisfactory.
What was a mystery were the two areas, both left and right, that I’ve surrounded with dotted lines and question marks. These two areas are supposed to contain SMD pads – instead they have been ignored.
Up until this point, I had trusted the output of the pcb-gcode program. I began to suspect that maybe that trust had been misplaced.
I downloaded and installed a trial version of VEGA DATA, and after a lot of fiddling managed to get the simulator to run. Yes, you’ve guessed – with exactly the same results, in that the two areas mentioned above are not machined as they should be.
So. My Milling software and CNC conversion works – leastways it follows the instructions given to it, without protest or error.
Before I go further I need to investigate why pcb-gcode is ignoring parts of my PCB layout. When I do know, then I’ll let you know too.
JWB 8th April 2012
The Schematic and Board I’m using for test can be found here: http://joebrown.org.uk/images/MillTable/Diary/HV_OpAmp_Discr.sch and here: http://joebrown.org.uk/images/MillTable/Diary/HV_OpAmp_Discr.brd
My top-foil gcode file (this is a txt file, you can read using Notepad): http://joebrown.org.uk/images/MillTable/Diary/t2.nct