PCB design resulted in incorrect drilled holes

Archive BatchPCB
1

Hi all,

I’ve had a great experience spinning my very first PCB. I received my batch today and noted, on some holes, the drilled holes are the same size as the outer outer diameter of the pad. The result is a very thin plated pad around the oversized hole. I’m using DesignSparkPCB.

Example: a 5.1mm power jack has three 170thou pads, each with 118thou holes. The actual drilled holes are ~165thou with a very thin pad around the hole.

Most, if not all, of the round pads have oversize holes.

I attached the Gerber files for reference. I used GC-Preview to review the Gerbers but don’t know how to visualize the holes.

Thoughts?

Snapshot of the PCB Layout in DesignSparkPCB:

BatchPCB Boards. Note the overlarge triplet of holes for the power jack at lower-right edge.

Thanks for any feedback.

Michael

2

Looks like you’re right. The tool list in your drill file maxes out at 118.1 mils. I don’t know what Gold Phoenix’s rules are, but PCB houses generally will make substitutions within reason (like, actual drill sizes may be multiples of 5 mils so you get the closest one, etc.) to keep the # of different drills per panel from being insane. It’s possible that the rules relax even more as holes get larger, but 39.8% over seems crazy. So you might get somewhere by complaining.

That said, I’ve found it pays to be conservative with larger holes (i.e. increase the # mils by which the pad is bigger than the hole), even if they get it right – a big flamboyant pad is way easier to solder than a demure little ring.

3

I suspect the error is the substitution of the pad diameter for the hole size and thus the drilled holes. All of the round holes suffer from this issue, not just the big ones. Interestingly, the vias seem alright. But they’re supposed to have thin pads surrounding the hole anyway.

A manual check of the .drl file reveals the hole diameters are properly specified by the gerber output. So, the error appears to be in the manufacture, not the design.

M

4

On an unrelated note, the optoisolator isn’t providing any protection in this circuit, since both sides share a common power supply and ground.

/mike

5

n1ist:
On an unrelated note, the optoisolator isn’t providing any protection in this circuit, since both sides share a common power supply and ground.

/mike

Meh. Of course.

I’m only trying to isolate the DMX input lines which are much more exposed to transients since it may be hooked up to gear over some distance.

Hmm. I’ll do some reading.

Thanks!

6

Over at DIYC and DLA (www.doityourselfchristmas.com and www.diylightanimation.com). we tend to use the NMR100C as an isolated converter to power the DMX side

/mike

7

Thanks Mike. I actually have a couple EtherDongles and Lynx Express. I’ll review the datasheet.

So, as I understand, the 485 chip should be on the isolated side of the DC/DC convertor. Is that right?

Is there any value in powering everything BUT the 485 with the convertor?

Michael

8

You are better off powering the RS485 chip off the isolated converter as it will draw less current than the rest of the circuit and the converter is rather limited on output power. That’s what RJ did on the etherdongle or even the USB dongle.

/mike

9

Excellent. So, the convertor isolates the power lines and the optoisolator handles the data lines. I’ll update my design accordingly.

Thanks!

(I love open source!)

Michael

10

Just for fun, here’s the updated version with isolation for both power and signal. Thanks, again, for the feedback.

Still don’t know how to solve the over-sized drilling. The SparkFun staff haven’t responded.