The boards typically incorporate RF comms and ADCs, so breadboards/veroboard only go so far (though this is where I always begin).
I am not so much reticent as a bit choosy where I would use it; my main problem is that it takes quite a large investment in time. Boards to be milled have to be planned for in the layout stage and there is an extra step in converting Gerber files to G code (Coppercam or similar). The board then has to be milled - which normally has problems such as short circuits caused by copper swarf (not to mention operator error), as well as the extra steps of fitting shorting pins to replace the vias. The completed board has to be treated with extra care as there is no solder resist.
A simple PCB may take 4 hours of your time to make - compared with about £40 from PCB-Pool. Except for the time from dispatch to recieving the goods the commercial process is cheaper and better in every way.
Having said that, I routinely use PCB milling if it is for my hobby - and would at least consider using it if time is particularly critical.
Since I wrote that I have had occasion to make a PCB with very fine gaps and tracks. In this case it was a 10 pin DNF switch mode regulator IC with pads on 0.4mm pitch - the little Proxxon acquited itself admirably, although I did fit long springs to bias the X and Y axis and reduce backlash.
One point to note with the spade bits is that the 60 degree angle means that the cut width is very depth sensitive, you should ensure that the bed that you mount the PCB on is as true as you can get it or the variation in the cut width may make the job impossible. Outside of that, I think a LPFP100 should not be a problem.
As far as bit life goes, I could no more than hazard a guess that it is about twice the life of the common 30 degree engraving bits (sucks end of thumb possibly 10 meters with 0.2mm width)
That's very encouraging to hear. Many thanks for your quick and helpful reply.
I was thinking more in terms of building a 3-axis mill from scratch, but I must confess that finding a spindle with a runout spec suficiently good has been a problem. The Kress spec doesn't look as though it quite cuts the mustard for fine PCB work (unless somebody here has contrary experience?) In PCB we seem to be working at the bleeding edge of what's achievable. I have a cheap and Chinky 3.5" lathe and was thinking of having to make my own spindle.
But I confess I simply hadn't thought of using a Proxxon mill. It's true that their stuff is solidly made, and if it's as good as you say it may be a route to go. Which model do you have? (Have you posted piccys anywhere? It's always stimulating to gawp at other peoples' kit. ) I had set myself a design goal of milling Eurocard size boards (160 mm x 100 mm) but more from habit than actual need. Most of the PCBs I actually deal with these days are much smaller than that and could probably be milled at least four-up on a Eurocard.
I've spent the guts a few a months trying to get consistent pcb milling for reasonably fine pitch SMD....I actually only go need down to SOT23-6 compoinent size (which is about 0.6mm space between pins). I've got there in the end, but not without an astonishing amount of hair pulling & frustration.
Here's my general musings...
1. Level the machine bed - if you don't have a level bed, then you are up against it from the start - forget a V cutter unless your bed is level (in which case go with a small endmill instead)
2. Choose the right V bit for the board - for example if you intend going really fine - it likely has to be 20 degree V bit....careful though, these tips snap like buggery & are expensive!
3. All V bits are depth sensitive - the wider the angle, the more robust the tip is(less prone to snappig the tip), but with wider angle V bit, just taking the bit down a little in depth really make the milled cut a lot wider - personally I'd never now use anything wider angle tha a 45 deg.
4. Calibration, calibration calibration - eliminate backlash from your Z - I knocked up a simple g-code file to raise/lower the Z axis about 30 times....and made sure at the end of it my dial meter still showed 0.01mm or less error after 30 Z ascents/descents. Also, missed steps are the scourge of pcb milling (where incredible amounts of accuracy are needed - so make sure that your axis consistently move the distance they are meant to ie use a dial meter again)
5. If you are going really fine pitch, then seriously consider some form of Z autoleveller program- (this is where you first probe your copper board to pick up the height irreguarities (even if your bed is level, just clamping the copper down will raise the board slightly in the middle region) ...and then apply the probed Z inconsistencies into your g-code Z info prior to the cutting run. You can get a free autoleveller in pcb-gcode (an eagle plugin), but I went the way of CNC-USB which has this feature integrated (it's called warp - http://www.youtube.com/watch?v=0jGY92S8bxM )
6. Use a CAM program that imports gerbers...then you can really fine tune your V bit tip offset to suit...and see the projected cut before you go near the machine itself . I use CAMBAM (so create pcb in eagle ,export tracks as a gerber, import gerber into Cambam...group select all the pcb track outlines in there, tweak your machining parameters to suit....some say you should use an engrave mop for milling the tracks ...this is wrong...if you do, there is no offset applied (it cuts along the pcb track outlin line...and therefore you lose PCB track width = the radius width of the V tip at the actual cut depth ...you need to select a 'profile' cut (which then allows you to apply an offset away from te pcb outline- you want the maximum offset possible but without fear of then cutting into a neighbour pad when milling your finest pads - you can see the visulaizations within cambam - the trial is free )
it's been a long journey, but worth it....being able to knock out a consistently good proto pcb in short order...epanelled, mount holes in the right places etc (vs chemical etching) is a major win for me.
Last edited by HankMcSpank; 11-03-2012 at 01:30 PM.
Hank, that's very helpful. Z-axis resolution and accuracy have been bugging me a lot for the very reason you say. In a previous incarnation I was Chief Calibration Engineer of an MoD Main Contractor in gas detection where we cut big gas-tight threads in some of the nastiest stainless steels I never want to encounter again, and I don't believe ANY instrumentation - period! (I sometimes had to go back in after hours and check the measurements my staff were making during the day, which was a good test of the repeatability of the kit as well as the competence of the staff. Both passed I'm glad to say). We used a Mitutoyo Low Voltage Differential Transformer (LVDT) probe for examining the wear on thread gauges, and it was very sensitive and repeatable, so I have had it in mind to use LVDTs on each axis for absolute positioning, and to incorporate their signals into the controller before each cut.
Yes <sigh> I was afraid of narrow carbide bits snapping. I think I'd rather spend the money on getting good positioning accuracy in the first place, but in real life some wastage can't be avoided.
I hadn't even thought of missed steps being a problem! But as I'm actually a software engineer rather than a mechanical man I should be able to write some code which eliminates missed steps. And again, the feedback of an LVDT on each axis should help to eliminate that problem. And measuring the warp before each cut is a useful thought. Thank you.
Your point 6 is also a good one, and taken.
Like many of us in engineering, I'm a perfectionist; which means I go through many iterations of software and prototype PCBs before I'm satisfied that I've got something I can release. That's the main reason for needing to do PCBs myself rather than just sending the artwork off to Olimex or whoever, and I must have a system which turns out reliable boards quickly. I'm grateful for the benefit of the experience of all here.
If you want really quick prototype PCBs why not use the Press-n-Peel / etching method. You can easily get 10 thou tracks and spaces. You can then use cnc to drill the holes from your Excellon files.
What I found with chemical etching...it was a real chore, massive time sump, got erratic results (unless you use fresh batches of developer & ferric chloride each time - but then that starts getting expensive) then there's the temperature variations....make a chemically etched PCB on a cold day...you'll get differing results vs a chemically etched pcb on a warm day (my point being lots of variables) & fraught with problem getting the alignment right (oh what joy to see 1.5 hours chemically ecthing wasted because the CNC machine wasn't aligned 'quite' right when drilling holes out or cutting the board outline) ...and wait there a minute, if we are going to use the CNC to drill out the holes, we may as well get it to isolate the tracks in the first place!
Your mileage may vary, but I need rather intricate shaped PCBs...and therefore for me at least it's better just to run the whole sequence on the machine...at least this way you can be assured of everything being aligned.
My next target now is - after the tracks have been milled on my CNC machine - to get my CNC machine to dispense solder paste on all the track pads (I have all the 'process elements' in place...I'm just waiting on some solder paste of the right viscousity to arrive) ....another huge time win (because for protos, it's not worth making a solder stencil, and it's a real chore dispensing solder paste by hand each time!)
Last edited by HankMcSpank; 11-03-2012 at 05:28 PM.
By cropwell in forum General ElectronicsReplies: 2Last Post: 01-05-2014, 09:33 AM
By cropwell in forum Lathes, Lathe Rebuilding & ConversionsReplies: 44Last Post: 14-03-2014, 05:32 AM
By craigrobbo in forum Items For SaleReplies: 1Last Post: 28-07-2013, 06:54 PM
By Rockeyes in forum Milling Machines, Builds & ConversionsReplies: 0Last Post: 31-05-2012, 04:21 PM
By DGH in forum Milling Machines, Builds & ConversionsReplies: 0Last Post: 27-05-2009, 03:47 PM