Hi,
that is unusual. The only thing I can think of that would displace the hole is if Mach were displacing it by making an allowance for the diameter
of the tool.

The example board I posted is 100 x 80 mm. The outline is shown with a zero width line. The board mill path places the tooplath
outside of the line so that the finished dimension is 100 x 80 mm. I use a 1.5mm two flute endmill to mill the outline. The toolpath allows
for my standard etch tool of 0.5mm. Consequently my board will end up about 1mm undersize because I'm using the larger diameter tool
while the Gcode had made allowance for the smaller diameter tool.

When I drill the registration holes I use MDI and therefore can be absolutely assured that Mach has not made an allowance for the diameter of the drill,
if such allowance is required I do it. I do not offset the holes, I MDI to centres. For instance if I have just zeroed the X any Y axes at the lower left corner
and I MDI G0 Y-5 to drill the first registration hole its center is 0,-5   If I then MDI G0 X100 for the second registration hole its center is 100,-5 Thus the centers
are perfectly aligned.

I agree that the spindle would have to be badly off vertical for the displacement you describe. I think it unlikely.

May I suggest some experimentation. Either with some spare PCB blank or other suitably thin and stable plastic sheet or cardboard drill four holes in a rectangular
pattern with the drill going into the baseboard so you can try flipping it and see whether the holes line up. Its not inconceivable that the X and Y axes are not
90 degrees to each other. That will cause a nominally rectangular drill pattern to come out as a parallelogram. If you flip a parallelogram the holes won't line up.
Should say the X to Y axis be 89.5 degrees rather than 90 degrees the displacement of a hole would be 0.9mm over a 100mm rectangle. If you flipped it the
misalignment would be double at 1.8mm, and this occurs with a half a degree of error between the axes. In short the alignment between the upper and lower
traces of a PCB are very sensitive to the assumed 'squareness' of the axes.

Craig

Hi Craig!

Thanks for your kind recommendation and suggestion, and also for going out of your way to explain the procedure that I could try. I'll definitely follow that suggestion and try it out. I'll update you with the findings. Also, thanks for mentioning other possibilities, such as possible Mach 3 involvement. I'll explore that avenue.

I'll definitely be uploading some pics of my setup very soon. One thing that just came to mind is the relatively long 3 mm diameter drill bit that I use for drilling the registration holes. I think they're around 5cm long bits.  Maybe that could have something to do with it. I did order some shorter ones, so I will do some tests with shorter drill bits. I will definitely report on the findings with spare pcb blanks .... with drill holes at the corners of a square shaped pattern. Thanks again Craig. Much appreciated.

Kenny

Hi Craig!

Prior to starting those drill tests, I decided to see what would happen if I used a shorter 3mm diameter drill bit. In a earlier post in this same thread, I mentioned that I had ordered shorter length drill bots for drilling the 3 mm diameter registration holes. I had never used them yet. The drill bits that I had been using were precision 3mm ones, bought from a proper hard-ware store.

Total disaster observed yesterday after using the new shorter so-called '3mm' drill bits (ordered through an ebay-store). What I found was..... these short drill bits (clearly labelled 3.0) on the shank, with the box labelled 3mm drill bits ........ drilled holes that were very noticeably larger than the ones drilled with the 'precision' drill bit. I tested a whole bunch of these shorter '3 mm'  - ordered from various different ebay-store, all looking like coming from the same manufacturing supplier. They all drill holes larger than 3 mm ..... from observations and also measurement. I was thinking that if I used a shorter drill bit, then I could see if that would have an effect on the hole misalignment in the y-direction.

Unfortunately, these so-called '3 mm diameter' drill bits don't drill 3 mm holes. My precision 3 mm dowels (as well as the shanks of my precision 3 mm drill bits) just flop around loosely in those newly drilled holes. That makes these new bits unusable for registration. But that's ok. I'll just stick with the long precision drill bits for the moment.

I've uploaded one image of my milling setup. The drill bit mounted in the chuck is that short bit. This photo was taken just before drilling. The circuit board seen in the image is an old one that had bad misalignment issue on the top and bottom side, but the board is still good for doing these drill tests. The holes seen in the photo are old holes. I drilled new holes next to them (after taking the photo). That's when I noticed that the new '3mm' drill bit doesn't actually drill 3 mm holes! The holes just come out larger than 3 mm....even with really slow z-axis feed-rate. Disappointing, but at least I found out something about these drill bits.

Anyway, you can see in the photo that I make use of used water bottle tops for clamping down my circuit boards. They're a bit too short, so I need to later replace these with slightly taller support blocks. I also use suitable sized metal washers for clamping down the mid portions of the circuit board. Without those 'washer' clamps, the mid regions of the board often curve (bow) upwards a little. These clamps keeps the board relatively flat. And autoleveller software handles the rest.

Hi,
whats the bet you have 1/8 inch drill bits which are actually 3.175mm.

I've tried all manner of means to clamp the board down without flexure including the method you have shown.
When yo come to do bigger boards Autoleveller will fail to do the business with such clamping methods.
The best I've tried and currently use is double sided tape.

Apparently vacumm holding is better again but I haven't tried it yet.

Craig

Hi,
whats the bet you have 1/8 inch drill bits which are actually 3.175mm.

I've tried all manner of means to clamp the board down without flexure including the method you have shown.
When yo come to do bigger boards Autoleveller will fail to do the business with such clamping methods.
The best I've tried and currently use is double sided tape.

Apparently vacumm holding is better again but I haven't tried it yet.

Craig

Your bet was right on Craig. I measured those crazy bits, and it really is around 3.19 mm ....measured, which is right around the 3.175 value.

I haven't tried bigger boards yet, but certainly will work toward the bigger ones. At the moment, I've just been using relatively inexpensive waste-board, purchased from a hardware store. I needed to use the washer clamp held down with those lipped self-tapping screws, due to the bowing. I think the bowing is due to my bottle-top lids height being too short, so the clamps on the ends are exerting some force --- causing the bowing. This is just a guess right now. I will find some taller supports and see what happens.

The vacuum holding method does sound good. I will take a look at that method to see what's involved. At first thoughts, I'm guessing that there needs to be holes in a waste board in order to have some suction devices on the other end of holes to create suction forces to pull the circuit board down into the waste-board. I'm thinking that if there's enough suction applied across enough parts of the circuit board, then the circuit board would get held down quite nicely.

Thanks again Craig.....the '3mm' bits (says on the packets) are certainly 3.175 .... disappointing.


Kenny

Hi,
no messing around.....go to this outfit for PCB drills and tools:

https://www.ebay.com/str/carbideplus

Craig

Impressive site!! Thanks Craig. I will be looking for short 3mm diameter drill bits from there. That store is selling good quality items.

Yesterday, I decided to go ahead to try adding 0.6 mm offset to my y-axis (for curiosity) on the flip-side (bottom) mill. I hit the jackpot this time - nice alignment on the two sides of the board after pad hole drilling. The very nice thing is that the offset discrepancy is constant, so at least the pattern locations are repeatable. The four registration holes themselves have no issue, which is interesting. I flip the board over, and the 3mm dowels fit it snuggly - both sides of the board, and I can pull out each one after clamping without brute force method. There was a time (before I knew about the need to square-up the x and y axis) where the board registration holes (on the flip-side) didn't line up with the drilled holes in the waste-board --- due to the two axes not being square enough to each other.

Thanks again for link Craig!

Kenny

Hi,
yeah, drillman 1 is a good bloke. I have bought a bit of Chinese carbide, its cheap but that's about the only thing to recommend it.
The Koycera Tycom stuff is good and reasonably priced, not as cheap as Chinese, but price/performance much better than Chinese.

I got a 1/4inch Raptor by Destiny Tools from drillman 1 and it kicks anus in stainless steel, best I've ever used!

Some of my circuit boards are high current and so I have some very heavy copper board, 12 oz or 420um or 0.42mm thick!
You cant use ordinary engraving bits, the taper defeats you with such a thick copper layer. I use two flute endmills. My board is
surface mount with SIOC outlines, 0.6mm between pins. The endmills I'm using are:

https://www.ebay.com/itm/5-0-50mm-0197-2-FLUTE-MICRO-CARBIDE-ENDMILLS-Kyocera-1600-0197-079/151724697381?hash=item23537f2725:g:jwcAAOxy1VlRFPam

I got them on special at $2.65 each so I bought 30, wish I'd bought more now, I think drillman 1 had 100 at the special price at the time.

These little endmills are very easy to break. Two things which determine whether they work well is how much fiberglass you have to cut. If you use Autoleveller
to best advantage you can get away with cutting very little fiberglass and end up with a better job. I also use flood cooling, its more about washing away the chips
than cooling but it means I can get 8-10 hours out of a single tool whereas I can get about 1/2-1 hour otherwise. Also the cut edge is so much better.
I've now got into the habit of using flood cooling when cutting ordinary board (1 oz) with ordinary engraving bits, they last longer and the cut quality improves
so dramatically its worth the hassle.

On small to medium sized surface mount boards with 1 oz copper and 30 degree engraving bits with Autoleveller I set the cut depth to 0.05 to 0.06 mm, that is
50 to 60 um. Remember the copper is only 35um thick. For that to work the board can have NO FLEX.......NONE WHATEVER.

As I posted earlier I use double sided tape. Its not ideal, it can be hard to get it to release without bending or breaking the board and/or the flood coolant
can stuff up the adhesive. None the less I have had some great results. I've made boards with QuadFlatPacks with only 0.2mm between pins.