A tiny prototype sensor break out

[Jay C]

I was contacted by my alma mater to mill a break out board for a tiny hall sensor. We are talking 1.1mm square with 4 .25mm diameter ball contacts underneath. The circuit was done in Eagle. The machine is a Widgitmaster wide with a Proxxon.

The board is 40 x 12 mm. The wide traces are .8mm, the traces to the sensor are .4mm. The 2 SMD capacitors are 805 size. It looks a bit rough as I think I need to adjust my PCB-Gcode settings. This is a fresh bit and I had it set for a worn one (.007" versus .005") also the depth is a bit too deep I think. However, it looks like it will work

The hard part will be mounting the sensor (holding it, and soldering it.).

I'll update when it's done.

Jay

[kn6za]

Jay,

Nice work! Cant wait to see the finished product.

How did you do the VDD and GND? Is it text on a certain level? I have tried this and the best I could do was the equivalent of engraved silkscreen on the non copper side.

[Jay C]

Originally Posted by kn6za How did you do the VDD and GND? Is it text on a certain level? I have tried this and the best I could do was the equivalent of engraved silkscreen on the non copper side.

I put the text on the top layer (or you can use bottom). I uses 16% on the ratio and the largest font size that would fit. I need to run some experiments to see how deep I should set my cut. The very close lines of the text don't leave any room for error

The blue is the mill layer. I used the "locking" 3-position deader from Sparkfun's library. That's why the holes look "off" ... they are supposed to be

Jay

[kn6za]

Jay,

Thanks, I got it to work! I had the font set to proportional, once I set it to vector it looked and worked like your example.

[pminmo]

Have you ever measured the runout of the Proxxon?

[Jay C]

Originally Posted by pminmo Have you ever measured the runout of the Proxxon?

Way back when I first got it I did though I can't remember what it was. I'll dig out my indicator and have a look.

I changed the settings on PCB-Gcode. I used .003" for the tool width, .003" for the DOC and 10ipm for the feed. The spindle was set to 11K RPM (thereabouts). I used a bit smaller drill, added a milling trace and widened the traces to the sensor to .024". Just waiting on the sensor now

Jay

[Jay C]

I did a line test to help determine the right settings for the PCB-Gcode. Just some 0805 SMD parts joined by straight traces. The thickest is .05" the thinnest is .01".

A: .007" tool, .005" depth, .005" step size. .005" is too deep, I can feel the "height of the copper and on these little parts, I don't want them to fall off into the trough. Feed was 8ipm. Note the left over whiskers, this tells me the tip is not .007".

B. .003" tool, .003" depth, .003" step. This one is perfect. I did shrink the isolation from .02" to .01 to speed up the run. Feed was 8ipm.

C. .004" tool, .004" step, .004" deep. Reset iso to .02" ran it at 18ipm the whiskers are there like on A, but the .1" line is impressive at that speed.

Also, I set up my indicator (though not very well) to get an idea of runout. I can hear the vibration in the Proxxon at lower speeds, and 11-13K RPM is the sweet spot as was told by the dial indicator. I need to try my Wolfgang spindle I suppose. Just need to make a adapter so it'll fit in the 20mm (.759") holder as it's only 19.01mm (.748").

FWIW,
JAy

[pminmo]

The left over wiskers are common, and it's not necessarily the tip size, but can be. When you get down to tiny remaining slices of copper it has a tendency to deform or move and leave those. If you set the minimum and maximum isolation to .1mil, pcb-gcode will generate one isolation cut. If you have a set of digital calipers or feeler gauges you can measure the width of that cut to help you determine what your actually cutting. Vibration, runout and backlash are huge issues when doing fine trace and isolation. They can and normally add mil's to the actual cut width. A $19 set of digital calipers from Harbor Freight is money well spent.

Ref: http://millpcbs.com/index.php?option...d=15&Itemid=61

[Jay C]

Oh yeah, I own a pair I measured the resulting trace widths.

They are smaller by .001" on all of them, tho that could be the operator