# Finding Shorts

While swapping SDRAM chips on one of the Mesh Potato prototypes a short developed between 3V3 and GND. This is nasty as 3V3 and GND are both power planes so the short could be anywhere on the board. I had a few goes at finding the short over the last few days but no luck.

The short was very low resistance, giving the same reading as a dead short on my multimeter. I tried the trick of high current (3A) low voltage (0.3V) in an attempt to “burn” the short out but that didn’t help either.

Then I started dreaming up ways of locating the short. The first idea I had was to pass a high frequency signal through the shorted nets. High frequency currents follow the path of least impedance so maybe I could track the signal path with some sort of small loop connected to my oscilloscope. Then I extended that idea to a DC current. I figured that current must pass through the PCB 3V3 plane, the short, then back through the ground plane. The power planes have a low but not zero resistance. As the current flows through the planes, a voltage drop will gradually develop (V=IR) along the path of the current.

So if I measured the voltage at various nodes on the 3V3 and ground nets, I should see slightly different voltages along the path of the current. If I measure two nodes that have no current flow between them, they should be at the same voltage. So by looking for small changes in the voltage I should be able to establish the path of the current through the planes. When I get to the short, the voltage between the 3V3 and GND node should be equal.

So I rigged up a high current but low voltage (1.6A, 80mV) through the short. The low voltage means no components can get damaged, despite the current. I connected one multimeter probe to the power supply negative terminal, and probed various nodes with the other multimeter probe. I measured about 30mV on most GND nodes and 50mV on most 3V3 nodes. However one 3V3 node was 45mV. It was a capacitor near where I had removed the SDRAM chip.

So I carefully pulled the capacitor off and looked underneath:

### 6 comments to Finding Shorts

• When I was a lad there was an instrument maker called Hewlett-Packard. They made these nice little logic probes in two forms. One detected pulsing voltages. The other detected pulsing currents. The current detector was great for finding shorts like this.

• david

Thanks Steve. I really like the line “When I was a lad there was an instrument maker called Hewlett-Packard”…

• Philip

If you had just posted a bit earlier… We have just bought an IR camera that should have showed tracks with slightly elevated temperatures. It would have been good to pay something back.

• An IR camera or a thermal one? Thermal cameras work pretty well for this sort of thing, but IR cameras need the object to really cook before anything shows up. You’ll notice people like the fire services never bother with cheap compact IR cameras to find people. They use those expensive, clumsy and often rather fragile thermal cameras, as that’s what it takes to do the job.

I used to use thermal cameras quite a lot for analysing both current flows and the effectiveness of passive cooling schemes. Its an interesting activity, as when you can see the thermal scene in fine detail it often looks very different from how you might have imagined it would.

• The downside of this technique (and some may say it’s actually a benefit) is that in the case of small very-thin shorts (tiny flake of copper or whatever), that current can actually burn out the short. With the short now gone, there’s no way to determine where it was, and you’re left wondering if it might come back sometime, after some vibration or whatever.

But like I said, some consider this a benefit, and more than one short has been cleared by deliberately passing 5 amps or more through it.

• Mark R Rivet

Ok, here it is. I stumbled across a 100% sure fire method to find plane to plane shorts no matter what. Since finding this method I am 100% successful at finding plane to plane shorts on PCBs. I have had th opportunity to use the method very often as I am an electronic technician for contract manufacturers for thirty years; yes, I’m old; 50. So anyway, it involves using a current source and a sensitive volt meter. The instrument I used is the one that discussed the method. It was a Polar instruments tone ohm 850. Its a little involved, so if you email me I will provide a detailed description on how to do it along with diagrams and how to construct a test pcb so you can see it work with an uncluttered pcb.