'Hobby PCB CNC Drill/Mill - How to measure resoluti'
|The company I work for is just finishing the hardware and software for a
low cost CNC drill/mill machine for hobby use. It won't have quite the
resolution of more expensive units but should be more than adequate for
most hobby boards.
My problem is I have to measure what it's useful resolution is so
prospective buyers can decide if it's adequate. I imagine that for most
commercial systems you can calculate this by using the specifications for
all of the components and summing the worst case.
What I believe we need to do is somehow measure the resolution from test
boards, but I am unsure on how to do this.
The theoretical positioning accuracy when drilling is .001" +/- .0005
however I suspect in practice it is more like +/- .001
Routing is going to be much lower since the software can't compensate for
slack in the hardware as it can when drilling.
Any idea's would be appreciated. The first model has a capacity of 8" x 12"
and uses a PIC17C756 to control the steppers and translate the CNC codes.
(I did the PIC code)
KoolKits Electronics Ltd.
|My problem is I have to measure what it's useful resolution is so
|prospective buyers can decide if it's adequate. I imagine that for most
|commercial systems you can calculate this by using the specifications for
|all of the components and summing the worst case.
I'd suggest that if you build a few, you could try making some boards
with holes produced in a random sequence and then check their placement.
This may be easiest if you place the holes on a "rotated" grid [e.g. so
that the holes on a row are seperated by 0.06" in the X direction and
0.08" in the Y direction. This should net you a 0.1" grid].
|The theoretical positioning accuracy when drilling is .001" +/- .0005
|however I suspect in practice it is more like +/- .001
|Routing is going to be much lower since the software can't compensate for
|slack in the hardware as it can when drilling.
Why can't you compensate for backlash? It would seem you have two ways
to do it:
 Measure (or allow a user to enter) a "backlash" adjustment distance;
if this is set accurately, it should take care of most of the back-
lash, though there may be some uncorrectable slop.
 Always approach route "start points" from the same direction, and
always draw vertical and horizontal lines in that direction. For
diagonal lines which are in the right horizontal and vertical dir-
ections, proceed as expected; for those which have a "good" and
"bad" direction, split the line in half and approach both ends of
the line from the favorable direction.
Obviously still not 100% perfect, but I think you should be able to do
|Having spent some considerable time in the engineering industry both
using and fixing machine tools I will be very very suprised if a hobby
quality cnc router can achieve .001 +/- .0005. This kind of resolution
requires special clearance bearings, high quality backlash compensated
ball screws, attention to expansion compensation, extreme rigidity of
mouting etc. We had to set up comercial mills on special bases and use
laser levelling techiniques to achieve this with comercial mills.
We used to test mills with a calibrated dial indicators. One at each end
of each axis. We would zero the indicators and then run a cyclic test
pattern and watch the indicators.
The problem you will have in making test pcbs is how do you measure the
result to within .0005. I suspect the drils can't drill this this
accuracy. Have you measured the spindle runout ?
Even +/- .001 on a hobby machine, takung into account cutting loads etc
will be very impressive.
Bob Bullock wrote:
Peter Williamson | Phone: +61 15 898934
Waybeat Pty Ltd | Email: waybeat.com.aupeterw
I don't know why a 'hobby' machine would even need .001" tolerances.
I've drilled millions ;-) of holes myself over the years with about 50
thou tolerance sometimes, but the boards worked fine.
I even made a PCB drilling machine and plotter out of junked photo
copier parts, a hobbiest PCB drill and old C64 to drive it. (didn't know
about PICs then). That was much better than 50 thou but still nowhere
near +/-.0005, but for hobbiest use I couldn't complain. It was great
then to sit down and watch this little contraption do it's stuff.
Mind you, I don't think I would have sold many.
PicNPoke - Multimedia 16F84 Beginners PIC Tools.
*** FREE PCB OFFER ***
by the way, where we can get moe info about this specific machine you
are talking about..? prices and so on..
At 08:56 AM 99/03/30 +1000, you wrote:
>I don't know why a 'hobby' machine would even need .001" tolerances.
The .001" +/- .0005" figure simply refers to the theoretical mechanical
positioning accuracy of the linear drives. We know that the actual
realized resolution will not be anywhere near this, maybe as high as .005"
I have seen some drilled boards and can say they look pretty good, but that
doesn't seem like a very technical way to describe the PCB drill's resolution
to someone thinking of buying one :)
Maybe we are on the wrong track and a statement that simpley states the
feature limits of a PCB that it can handle is the best way to describe it.
KoolKits Electronics Ltd.
More... (looser matching)
- Last day of these posts
- In 1999
, 2000 only
- New search...