It works with discrete wiring too.
• A good Digital Volt Meter (DVM): Ideally it should have microVolt resolution.
• Needle Test Probes: Regular test probes are too crude. Make you own needle probes from old pens using sewing needles soldered on to test wires and then glue/ hotmelt/ epoxy the needles and wires into the empty pen shells. They will look like regular test probes except the tip will be a sewing needle and very sharp. It will be able to penetrate the solder resist coating on the PCB traces. Do not blame me when they fall off the workbench and get stuck in your leg. Be careful.
• Power Source: Use an adjustable power supply or other power source to have a continuous current into the PCB but not so large that something will be damaged.
Adjust the power source to have a continuous current into the PCB and the current of course will eventually flow into the short. Pick a suitable current that will not destroy something.
From the input power point there will be PCB traces going to different sections of the PCB. Basically the runners will branch out to individual circuits. Any current flowing trough a conductor will cause a voltage drop since the conductor will have some resistance.
The goal is to find the voltage drops since that is where the current is flowing. Measure voltage across a section of a trace. If needed, stick the needles through the soldermask into a copper trace a few millimeters apart, the further apart the better since the measured voltage will be larger. If you see a voltage, it means that there is current flowing through that runner on the way to the short. Keep following the runners with the current until you reach the short.
Once you are familiar with the procedure it is extremely fast. Note: You will always see a few microVolt caused by thermal emf effects (thermocouples formed by the different metals in the test system but you will learn to recognize them and ignore them.)
The technique is more difficult with multi-layer boards with internal power planes but it still works. Last week I had to do a failure analysis of a large complex PCB with about 35 ICs and it was a 10 layer PCB board with internal power layers. A customer had applied 12V instead of 5V and multiple ICs were destroyed. The ICs failed drawing excessive current of about 50mA so it was a real bad situation since it was not even a real short.
I found the 5 bad ICs. I even calculated the current into each one by measuring the voltage drop across a PCB trace that was only 3mm long. By knowing the mechanical dimensions of the copper trace and the copper resistivity and the voltage drop, I could calculate the current. It was correct to about 15%.
It works! This procedure is extremely helpful for advanced trouble shooting.