For more than thirty years I have been drilling PCB holes by eye alignment using some form of drill press.  Over the last ten years I have been using auxiliary magnification and lighting to improve accuracy with ever-decreasingly sized components and increased layout densities, but accurately locating  drill centre has never been consistently perfect at sub-millimetre dimensions due to parallax and lighting contrast issues.  I have thought about purchasing  a bottom-up PCB drill or using a CNC router/drill for some time but the costs are significant.

I have recently hit on the idea of using a low cost CCD camera with an overlaid target graphic, mounted below the drill station for easy optimal drill alignment.  Any swarf can be removed from the optical path with a stream of air from a low-cost fish tank bubble pump.  All I will need to do is get used to looking at an LCD computer monitor as opposed to staring at the board and guessing where the drill bit will make contact with the PCB.  This approach offers the feel of hand drill feed (as opposed to a Z direction forced stepper-motor drive) and improved accuracy provided that the intended hole centre can be viewed at the bottom of the work-piece.

Figure 1.  Drill Press with LEDs and CCD Camera Below

My first mission has been to purchase a low cost USB CCD microscope with the ability to incorporate an overlay on the displayed image.  Jaycar Electronics sell a USB Digital Microscope for around  NZ$50 complete with LED lighting.  Excellent!.  The accompanying software (a version of AMCad - and in my opinion not the greatest piece of software ever written) can do overlays for drill centre alignment.  As it turned out getting the overlay on the drill centre was simply painful.  It was easier to draw a simple X target on some Mylar film and tape this over the LCD monitor where it also protects the monitor from errant dust.

My next task was to true-up my antiquated Dremel press as best I could and remove as much slop as possible.

The most difficult job was making the camera bracket and fitting it to the bench.  I used an aluminium angle section and thermo-formed some Perspex over a piece of 38 mm brass rod for the camera clamp but an appropriately sized pipe clamp would do.  Make sure that the camera focus ring is not obscured by the clamp.  The bracket allows about 5 mm of movement at one end to form an an intersecting arc with the drill head.   This allows for rough alignment of the camera and drill centre to within a couple of mm.  The face of the camera is about 28 mm below the surface of the drill table, and just below the bench surface.  The hole in the bench top was 43 mm in diameter covered with a 6 mm Perspex plate.

Figure 2.  Camera Mount

Magnification with my set-up is about 15 times with the video running at 30 fps at a resolution of  640 x 480.  All of these parameters are readily adjusted either in software or by moving the camera relative to the plane of the drill surface.

Drill alignment is really easy.  Get the drill head and camera roughly aligned by rotating both the drill head and the camera bracket as necessary. Now place a piece of old circuit board on the drill table and adjust the camera focus and magnification.  Then mount a drill bit and turn the drill on.  (You must have the drill running for this procedure to get rid of any low speed eccentricity.  The reason for this is that the back-bearing on a Dremel rotor is mounted in a rubber boot.  It is only self-centring at speed.)   Lower the tip of the drill into the cameras focal plane.  Now drag the video application window on the monitor to position the tip of the drill bit on the Mylar X target.

An alternative procedure is to complete the rough alignment, drill a hole in a rigidly mounted piece of scrap (or the edge of your board) and then drag the video application window on the monitor to position the centre of the hole on the Mylar X target.

Figure 3. Target over 0.7 mm Hole

Movie 1.  The Drill in Action - Two 0.7 mm Holes On Dead Centre in 10 s
(You can’t see the target in the video because it’s on the Mylar taped to
the LCD monitor, not in the USB camera’s optical path.)

A word or two of warning.  Use the monitor for achieving alignment and then watch the drill during the feed to avoid making holes in your fingers.  Eye and hearing protection are not optional extras, but you didn’t need me to tell you this.

Now stick a test board on the table and practice drilling to perfect your technique before you destroy something important!  Note that you may get some image shake as the drill powers up depending on the sturdiness of your bench but this won’t affect the accuracy of the hole centre so long as you hold the board steady.  On technique, I turn the drill off between holes using an auxiliary switch on the power cord.  I don’t use the switch on the Dremel (set to full speed) as this risks changing the alignment of the drill head.

Remember to use sharp tungsten carbide drills at more than 30,000 RPM.  If you haven’t discovered it for yourself already, small diameter drills made from high speed steels will loose their cutting edges very very quickly with FRP boards, get jolly hot, and then ruin your board.

Recheck the drill alignment from time to time and whenever you change the drill bit or the position of the drill head (it takes just a second to lower the spinning drill into the focal plane and drag the application window back on centre), and drill a test hole at the edge of the board.  Also be aware that with two layer boards, poor layer registration may cause the drill bit to bend producing off-centre holes.  With extreme misalignment and small drills you also risk snapping drills.

The swarf problem I anticipated isn’t.  Most of the swarf goes up the drill  to the PCB surface and the few bits that land on the Perspex plate (protecting the camera from swarf and dropped drills) are so far from the focal point that they are an indistinguishable blur on the monitor.  A quick exhale from time to time clears the stuff nicely.

I’ve just finished drilling my first three boards using my new optical system.  Without effort I managed slightly more than 200 holes per hour - every one on centre and not a single broken drill bit.  I should have made this years ago.