Okay, so I was a bit of an idiot while designing my PCB. The specifications of my PCB fab were 6/6 mil trace/space, and the boards did not meet this. So, below is the design for the PCB:

Design file

The trace going to C16 (a ground) is way too close to the pad which is connected through a trace to an inductor (it's a switch mode buck converter.) This basically connects the +3.3V line next to the filter capacitor to GND, shorting it. I verified this with the continuity test on my multimeter.



I have already bought 10 PCBs, so I'd like to know if it's possible to hack these and repair them. I'm okay with having to provide them with an external 3.3V source; in future, I'd like to get the converter working, but it's the rest of the circuit which is important.

Obviously I won't make this mistake in the next version... I might actually run the DRC. This is only for development.

  • 7
    \$\begingroup\$ LOL @ "I might actually run the DRC"! Hard lessons are the best teachers, though! \$\endgroup\$ – tyblu Feb 12 '11 at 20:27
  • \$\begingroup\$ @tyblu, Thing is, I did, and it passed. However, this was a last-minute change before sending it off to production, and I didn't run the DRC after making that change. Stupid, I know. \$\endgroup\$ – Thomas O Feb 12 '11 at 20:30
  • 1
    \$\begingroup\$ I can't say I have every had a situation where I was able to get a gerber file sent in for a board fab and have not be forced to go through a DRC and a DFM. It is always nice to be forced to. \$\endgroup\$ – Kellenjb Feb 12 '11 at 20:37
  • \$\begingroup\$ @Kellenjb, the PCBs were cheap (about £28 for 10, including shipping), so they probably skimped on doing their own DRC. No big deal for me unless I forget to do it myself. \$\endgroup\$ – Thomas O Feb 12 '11 at 20:43
  • \$\begingroup\$ Now that is an interesting price. Do they ship to mainland Europe? :P \$\endgroup\$ – drxzcl Feb 13 '11 at 22:35

If I were you, I would try to burn the small copper bridge with a current surge. Using some cheap probes or solid wire, touch between the two points you don't want connected with a power source current limited to about 5A. If it doesn't work, increase current limit -- I wouldn't go past 50A, though. Note that the points you touch will oxidize immediately, and possibly also release magical smoke, so it'd be best if they were the edges of relatively large pads. Of course, everything between the two wires will be destroyed by 5A-50A, so don't miss any current paths...

Now, I don't mind blowing stuff up -- kinda' enjoy it, really -- but this method carries quite a bit of product risk. Don't get too mad at me when you burn your fingers, break your PSU, and melt some FR-4 on your desk. (You can get a little bit mad, though -- I understand.)

  • \$\begingroup\$ I have a 3A (CC/CV, short circuit protected) power supply, would this be able to do it? Unfortunately, trace resistance is going to be <0.1 ohms, and I^2*R tells us that even this would only just heat up the trace (0.9W.) Would 0.9W do it? \$\endgroup\$ – Thomas O Feb 12 '11 at 20:41
  • \$\begingroup\$ Heck if I know, @Thomas! S/C protection usually means voltage will drop upon short, though. AA batteries can surge quite a bit (use Alkaline so it doesn't wreck a rechargeable). I've never done this for such a tightly-spaced, small trace application. \$\endgroup\$ – tyblu Feb 12 '11 at 20:46
  • \$\begingroup\$ @tyblu, Yep, it drops to almost zero if it's shorted. As it's constant current, voltage drops if current increases beyond the set CC limit. \$\endgroup\$ – Thomas O Feb 12 '11 at 20:54
  • 1
    \$\begingroup\$ This is a useful fix when the short is underneath a component. I have used this when Vcc was shorted to ground. A standard ATX power supply puts out enough current on the +5V and +3.3V buses to clear some big shorts. And the voltage is limited so no components are damaged. Hopefully. \$\endgroup\$ – markrages Feb 13 '11 at 4:04
  • 3
    \$\begingroup\$ I actually used this with success, so it gets the accepted answer. 3A is more than enough to vaporise a trace. \$\endgroup\$ – Thomas O Feb 14 '11 at 0:12

When ever I have PCB errors that cause something to be shorted to something else, I pull out my handy dandy razor blade and go at it.

In your situation it looks like you might also have an issue with solder causing a short. If this is the case, find anything that you can place to add an insulator. The hard part is going to be finding something that wont melt while soldering. Maybe someone else has a suggestion with this, I would try electrical tape and hot air gun soldering.

  • \$\begingroup\$ This should probably be solution #1... \$\endgroup\$ – tyblu Feb 12 '11 at 20:48
  • \$\begingroup\$ Nothing you can use that won't melt/contaminate the joint. Better to solder the parts in place and then mechanically remove the resulting short (ie, cut away some of the solder). You can coat the repaired area with a conformal coating pen or ersatz coating (glue, etc) afterwards. \$\endgroup\$ – John Lopez Feb 12 '11 at 21:25
  • \$\begingroup\$ I was thinking more about shorts that were under parts. \$\endgroup\$ – Kellenjb Feb 13 '11 at 0:27
  • \$\begingroup\$ Um, don't do that? Seriously though, this is a hard one to solve. You can kludge air gaps under SMT parts by first putting a solder bump on the pad, then holding the part touching them and soldering. Still you're talking about sub-10 mil gaps in most cases. You can also 'billboard' chip components (mount them on edge) to reduce their X/Y coverage area. \$\endgroup\$ – John Lopez Feb 13 '11 at 3:44
  • 1
    \$\begingroup\$ You can add solder mask by painting on a bit of enamel paint. You can find small jars of red brush-on enamel paint in the cosmetics aisle. \$\endgroup\$ – markrages Feb 13 '11 at 4:02

It's always a good idea to have a rotary tool with a fine tip handy when dealing with PCB rework. I prefer a rotary tool to a utility knife, but have been known to resort to a blade at points (no pun intended).

It's quite acceptable to do PCB rework like this on small quantities, even in professional environments. Make your cuts however you can, use your meter to make sure the circuit is OK, then fix the artwork.


I'd get out my magnifying glass and scalpel, it doesn't have to be pretty so it's worth a try, just make light repeated scrateches next to the track, in a couple of minutes the grooves will ge deep enough to sever any contact. I've had mixed success with this method.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.