Homebrew PCBs

I've been working on mechanical etching of PCBs using V-bits from 
Think & tinker (http://www.thinktink.com/) and having some great 
success. I just uploaded two pics to the Ghidera2000 photo album 
(pending approval) that show 0.010 traces cut with a 45 degree bit. 
One is an actual board I'm using and the other was a test of SOP and 
QFP packages. The camera won't focus up close and I'm no photographer 
so they're somewhat blurry. Some parts look like the trace was rubbed 
out and others like there is no cut, but both metered perfectly, no 
shorts etc.

Anyway, this is working so well that I want to try out a 0.005" stub 
endmill but I'm uncertain about the feed speed to use. I'm hoping 
someone here has experience with very small bits. These things are 
over $30 each and I ain't rich 8(

Anyway, the IPT for hard copper with a 1/4 endmill is about 0.006" so 
a little proportional math would have the IPT at approx .006/.25 = 
0.024% of the tool diameter. For the 0.005" bit that makes 0.00012" 
IPT. I'm assuming hard copper because I really don't know what type 
is used on copper clad boards.

My spindle runs at 24,000 RPM so 24,000 RPM x 2 flutes x 0.00012 IPT 
gives 5.76 IPM feed speed max.

Does this sound correct?

--- ghidera2000 <ghidera2000@...> wrote:

> I want to try out a 0.005" stub endmill but I'm uncertain about 
> the feed speed to use. 

I use a Quickcircuit at work, and I've never had any luck running
stub endmills any smaller than 8 mils into 1oz board, no matter what 
the feed rate.  I usually try to route my boards to accomodate a
12 mil or larger bits, only running the 8 mil for those sections that 
absolutely need it like fine pitch part pads.  The 12 mil bit I run 
at 3 inches per minute, and the 8 mil at 1 inch per minute.  Perhaps 
someone else has tips on the science of how to properly feed these 
small end mills but this is what I've learned empirically.  

Mark B.
Albuquerque, NM

--- In Homebrew_PCBs@yahoogroups.com, Mark Brueggemann <qrq_cw@...> 
wrote:
> I use a Quickcircuit at work, and I've never had any luck running
> stub endmills any smaller than 8 mils into 1oz board, no matter what 
> the feed rate. 
> Mark B.
> Albuquerque, NM
>

Could you expand on that? What kind of troubles did you have?

I could probably do evetything with a 30 degree v-bit that I need but 
if the boards aren't perfect (few are) then the really fine pitch stuff 
is hit and miss. A tiny bit of a warp in the board can cause over or 
undercutting with a V-bit. The endmill is appealing in that trace width 
doesn't vary with depth.

--- ghidera2000 <ghidera2000@...> wrote:

> Could you expand on that? What kind of troubles did you have?

Sure. The bits bust, usually in the first 10 seconds of routing.
Milling copper, PCB or otherwise is tricky business.  It's very 
soft and bits tend to 'mush' through it rather than cut.  Even 
spinning the bit as fast as it will go, and slowing the feed rate 
to almost nothing doesn't help.  I'm guessing the ratio of
the thickness of copper to the size of the cutting face of the 
smaller sized bits just doesn't work.  A possible solution is to 
go to thinner copper like 1/2 or 1/4 oz, but I don't have any 
stock of that on hand.


> A tiny bit of a warp in the board can cause over or undercutting 
> with a V-bit. 

Another problem with V-bits is if you set them deep enough to 
remove a lot of material, it removes a lot of underlying substrate.
I gave up on 'em pretty early on in the process, and use endmills
for everything.  One of the techniques in the quickcircuit user
guide is to do all your routing with the V-bit initially (fast),
which removes most of the copper, then following it up with an 
endmill.  I'm not sure if this really saves any time, what I 
usually do is get it running and let it go, checking on it 
periodically.  A complex board can take several hours per bit 
to run, but since it's unattended it kinda doesn't matter.


> The endmill is appealing in that trace width doesn't vary with 
> depth.

Yes.  I usually route with 2, sometimes three sizes.  A large-ish
bit like 20 mils to do the brunt of the isolation, 12 mils for 
whatever features the 20 can't cut, and 50 mil if I have a lot
of copper to remove.  In my application I do a lot of microwave 
striplines and the quickcircuit has pretty good repeatability
and accuracy.  The downside is you don't want to bust too many
bits in the process, or your costs go up quickly.  Once I got
used to using it, then creating the right drill and route layers
in my layout software to accomodate it, I can make some pretty
nice double side boards with only one flip of the material and
a couple bit changes, and even create routed snap-off panel edges 
for our productions' automated assembly.  The quickcircuit is a 
nice tool but it's expensive to buy and finicky to operate, so 
I've ruled one out for my home business PCB jobs.  Toner transfer
is a bit more practical for most stuff I do.

Mark B.
Albuquerque, NM