Estimated reading time: 6 minutes

Drilling Deep

Why do my drilled holes not line up with the copper pads?

More specifically, why do the hole positions vary so much on the exit side of thick PCBs?

Whether you stack high or drill thick panels, the dynamics of drilling are similar. When you overlay the outer-layer artworks you may notice that the holes on the exit side of the panel or stack have a much higher level of positional variation than the entry side holes.

As designs get tighter and annuluses reduce in size this positional variation gets more and more significant. It can be a real headache. What was acceptable a few years ago, no longer stands up to inspection with finer pads and smaller pitches between holes.

I remember crossing this threshold when the design department in the factory I was then working in sent down a new design with a high-density through-hole connector in the middle of the panel. The PCB thickness was 8.2 mm and the drilled hole size was 0.6 mm. I know that some people have now gone much further than this in thickness and aspect ratio, but at the time I knew this would cause problems. The pitch of the holes was quite tight and therefore the innerlayer pads were small, so keeping the holes in the right place would be key to getting an acceptable result.

My first problem was more fundamental—the drill flutes on my 0.6 mm drills were not long enough to get to the exit side of the PCB. I had to make a few phone calls to my very supportive drill supplier. Many people in the UK will know “Storming” Norman Fergusson—a legend in the PCB industry who supplied me with drills and advice in return for cups of coffee. He managed to produce suitable long fluted tools and suggested some drill machine parameter settings which would give me the best chance of keeping the drill straight.

Before starting I made an audit of the drill department, picked the best machine based on previous tests, and went to town with new pressure feet inserts and checked all the foot pressures. I did everything I could to make sure we had the optimum working conditions for the equipment we had available.

We made panels and managed to yield enough good boards to fulfil the initial order but I could see enough positional variation on the exit side to know that we were on a knife edge. We needed some process improvement to get a decent process window to work in.

I did quite a lot of work with experimentation with hit counts but I must say this aspect was inconclusive. For panels over 6 mm, we were running quite low standard hit counts (600), and all the tests I made suggested that positional variation did not increase until the counts were much higher. The main reason to keep the hit counts low was to maintain hole wall quality—as the corners of the drill begin to wear, higher levels of roughness could be seen towards the bottom of the drilled hole.

I also tried deep hole cycling—drilling each hole in several steps and lifting the drill clear of the PCB between steps. Again, no detectable improvement in position but a definite improvement in hole wall quality. When drilling in FR-4 there is a definite advantage to peck drilling deep holes to help clear the swarf from the hole, which in turn keeps the temperatures lower and helps the drill to last longer.

When we experimented with drill parameters this gave the first real improvement in position. I added a larger delay (as per Mr. Fergusson’s advice) before the drill stroke. An extra 20 milliseconds, but the improvement was measurable. The drill machines we were using were quite old and I had already begun to wonder if we were just asking too much of them. Seeing an improvement spurred me on to keep trying. I no longer have the test data from this investigation as this is several jobs ago and the company I was working for closed more than a decade ago now. From memory I was sometimes seeing exit position variation of up to 12 thou [mils] or 0.3 mm. Adding the delay improved things by perhaps 10%. I needed to do better than this. I tried larger numbers for the delay but there was no further measurable improvement.

Drill entry and exit material trials next. I was using 0.2 mm solid aluminium for entry and melamine-faced fibreboard for exit and there were various alternative products available at the time, which all claimed to help with positional accuracy. I did not manage to measure any real improvement with the ones I tried. Perhaps there was a very marginal difference with the dual density aluminium (soft on top and hard underneath). I am not suggesting they may not be a benefit but I think my original material was pretty good. All the exit materials I tried worked well but there was nothing to choose between them, apart from price.

As I sat at my desk on a Friday afternoon I had a bit of a eureka moment. I had made a lot of measurements of exit position and quite a lot of measurements of the accuracy of the entry position. Up until now I had not really done much with the entry data. I had regarded it as less interesting because the values were much lower. They were always pretty close to the expected position—nothing like the variation seen on the exit side. When I superimposed the entry position over the top of the exit position for the same row of holes the coin dropped! Pretty much as soon as the drill touches the panel the gun is loaded. If the entry is off by a little it just keeps going in the same direction. Carbide tools are remarkably flexible under cutting conditions and the start point forms a bush which fixes the direction of the rest of the hole. I back-checked against a lot of data and the theory stood up.

This may go some way to explaining the improvement with an increased delay before the drill stroke. It was just enough time for any vibration to settle before the drill touched the entry material.

I will cut to the chase: No further improvement was going to come from tools, materials or drilling parameters. My background before PCB was in precision engineering and good hole position accuracy for metal cutting is usually helped by using a stubby centre drill. The idea is that the initial deflection is minimised as the small flute length and rigid shank keep the drill well-centered when it touches the material. I applied this to the PCB process, drilling the pattern with a special “stubby” tool before making a second pass with the normal drill, and this gave the required improvement.

I have known other companies to use similar processes and their exact method is often a closely guarded secret. Huge aspect ratios may be possible by drilling from both sides of the panel but the positional accuracy needs to be near perfect to meet up in the middle.

If you are getting close to your process knife edge…I hope this gives you some hints which will help.