PCB Trace Isolation - Minimum Track and Isolation

[ahoodlum]

So, I'm trying to make a PCB with Eagle CAD and pcb-gcode. 10bulls had posted some information earlier on his results with this technique. I have had equally as good results...with large traces.

A project I am working on requires that I use some TQFP packages (0.5mm pitch and 0.28mm spacing). I bought a drill mill from McMaster with a 90 degree angle and I'm having some trouble isolating 10mil traces. Most of them end up being more like 2-3 mil. I really don't like the idea of having to rely on depth of the cutter to dictate my trace width. A mil too much in the -Z direction can almost wipe out your trace. In the meantime I have ordered some 0.005" and 0.006" carbide end mills and I hope those will work a bit more reliably.

Anyway, I wanted to know everyone else's experience with using a milling setup to isolate REALLY small traces. What's the minimum track and isolation you have achieved? What kind of cutter did you use?

Thanks!

[fyffe555]

You've probably already spotted this but the pcb-gcode ulp produces gcode path following the edge of the trace in eagle. Theres no tool offset. That is, the centre of the cutter follows the track edge. If you're isolation milling and your cutter is removing 6mils total width of cut then that will reduce the width of your trace set in eagle by 3mils each side.... So , if you want to leave 10 mils and your cutter is removing 6mils then your trace needs to be set to 16mils in eagle.... The problem then becomes routing and pad spacings...

[ViperTX]

Well...that is why most people etch...

[ahoodlum]

fyffe555: Good point. I could add some gcode to turn on cutter compensation, but that won't particularily help my problem when my isolation is on the order of my cutter. That would only work when the isolation is significantly larger than the cutter. I will implement this though when I get my 0.005" end mills. Ugghhh, I hate waiting.

ViperTX: Since it appears those mills won't be here until tomorrow I think I am going to have to go ahead and etch these boards and move ahead.

Anyone else want to chime in?!

[fyffe555]

Three further options;

First: there's a ULP to convert the tracks to dxf and use that to produce gcode with offset - problem is the offset often puts the cutter into another track if the tracks are small and closely spaced.

Second: In a cad program you can import the ulp produced dxf of tracks and then in a second layer manualy draw a single line to make the necessary isolations. Then convert that layer to gcode giving a single tool pass to isolate two or more traces and removing the least aount of material possible to make th eisolation. You'd end up with a series of single lines breaking up the copper, not obvious traces. It's tedious, especially if you're doing more complex stuff and the end product looks a bit amateur.

Last option is etching and that's what I now use after going through what you're doing. Mainly because the cutters don't last long milling copper off FP4 etc and it gets expensive and the precision is difficult to reliably achieve isolation milling with <= 10mil traces. I drill the unetched board first using the drill gcode from the Eagle PCB ulp and cnc cut the outline using the outline gcode. Two simple passes, no need to move or realign the board. Then I'll etch, tonertransfer for simple stuff, photo for more complex and small trace stuff and smd. It works well because the predrilling allows accurate alignment for top and bottom layers. With toner transer the holes are filled with toner so avoiding etching from inside the hole. Can go from a finished eagle design to a board in about thirty minutes, depending on the number of drill sizes and whether I tin it ...

Which is probably just as well since I seem to go through a lot of prototypes!

Andrew

[ahoodlum]

Alright, so I got the 0.005" and 0.006" end mills today and PROMPTLY proceeded to break them after milling about 0.100" of a trace in the PCB. I've never seen $25 disapear so quickly. Quite disheartening. Then I went back to making my boards with toner transfer and etching. Ahhhh...MUCH easier and quicker.

I think I'll leave PCB milling to big DIP designs and do toner transfer for SMD stuff. I have one good 0.005" carbide end mill so if anyone wants it I'll sell it for a steep discount.

[bobJandal]

Hey ahoodlum.

I had to mill some traces for a SOP chip that were 0.5mm lead spacing, and 0.5mm lead width (don't know what that is in mils, sorry !).

I got this conical tipped diamond coated bit, and its tip was 0.6mm at the end, so I didn't even bother trying.

I solved it by milling a mounting block that was at 45 degrees, then mounting the copper on this, and cutting using the outer edge of a small cylindrical cutter.

This worked, but only in small areas, like for one or two chips (which was all I needed) - if I tried anything bigger, board warpage caused the line quality to decrease significantly. With more fiddling with clamps, etc, something better might have resulted.

It did what I wanted, and I hope I never have to deal with smd chips in this way ever again !

So, its doable, but, like someone once said, just because you can push a pea up a mountain with your nose, doesn't mean its a good idea to do it.

Cheers,

Tom.

[pminmo]

I'm pushing the same envelope. After busting a couple of expensive bits, ($30 for me in 2 bits) I'm playing with grinding my own. After some trial and error, I think I'm getting there. Key was a $2.99 harborfreight investment(happened upon them at their local store). http://www.harborfreight.com/cpi/cta...emnumber=31501
and a stash of cheap carbide drill bits. I took some #78 carbide drills and cut them so there was very little spirial remaining, then ground (by eye and hand, and believe me neither are very good!) a high angle point. I've also been experimenting with grinding them to an engraving profile. I'm not to a point of saying 100% I'm satisfied, but have had enough success to feel certain I will. At 35 cents a drill bit it's a lot less frustrating!

Those little diamond wheels are surprising, both in sturdyness and how easily they cut and grind the carbide.

[samco]

I have had really good luck with these (as have others)
http://www.thinktink.com/stack/volum...e/mechmill.htm

http://www.electronicsam.com/images/KandT/cbottom.JPG

sam

[pminmo]

thinktink has a $50 minimum order...

[samco]

I have actally done this also. I would put the broken drill in one dremel - and the diamond wheel in another dremel. I would sharpen the 1/8 carbide shank to a point. Then I ground the pint in half and relieve it around the diameter.

If that made sense.

Yes they have a minumum order but you need a few extra for opps.

sam

I'm pushing the same envelope. After busting a couple of expensive bits, ($30 for me in 2 bits) I'm playing with grinding my own. After some trial and error, I think I'm getting there. Key was a $2.99 harborfreight investment(happened upon them at their local store). http://www.harborfreight.com/cpi/cta...emnumber=31501
and a stash of cheap carbide drill bits. I took some #78 carbide drills and cut them so there was very little spirial remaining, then ground (by eye and hand, and believe me neither are very good!) a high angle point. I've also been experimenting with grinding them to an engraving profile. I'm not to a point of saying 100% I'm satisfied, but have had enough success to feel certain I will. At 35 cents a drill bit it's a lot less frustrating!

Those little diamond wheels are surprising, both in sturdyness and how easily they cut and grind the carbide.

[pastera]

SAMCO

Thanks for the link - I have an old T-Tech PCB router at work and the tool life is horrible. Next time I need cutters I'll order some from ThinkTink and see how they run, at least there a lot cheaper

Aaron