How do you find shorts?

[yanir]

Hi does anyone have any recommendations for an inexpensive tool or technique for finding shorts on a pcb. I know about the the resistance method, I just don't have a fancy high resolution DMM like Dave (jealous). I found something online called the leak seaker which will do the job, but looks a bit home made.

[alm]

Inject a fairly high current. Voltage drops are proportional to current, so if the current is a few amps, measuring sub-ohm resistances can be done with almost any DMM. This is essentially a four terminal measurement.

Bob Pease described a very sensitive voltmeter with audio feedback especially for this purpose in his 'Troubleshooting Analog Circuits' book, and possibly in an EDN article.

Another crude trick is to feed it even more current, and hope that the short burns up before the rest of the PCB does. Not recommended with mains voltages, especially if there are components on the board.

[Bloch]

Not jused it myself and not inexpensive ! but look really nice . Iprober 520 it a current probe for use on the PCB. www.farnell.com/datasheets/1387648.pdf and http://uk.farnell.com/aim-tti-instruments/iprober-520/probe-current-positional-on-pcb/dp/1908148

[Bloch]

Another crude trick is to feed it even more current, and hope that the short burns up before the rest of the PCB does.

Yep that can work and at the same time use a finger as a temperatur sensor will help find something 

[yanir]

The short I need to find is on the tx line of a uart connection between a micro and a blue tooth module. The board is populated. The high current method will destroy my components and my 5mil traces. The short is probably under one of those components (or inside) and I would like to avoid removing both to find out.

[alm]

Not jused it myself and not inexpensive ! but look really nice . Iprober 520 it a current probe for use on the PCB. www.farnell.com/datasheets/1387648.pdf and http://uk.farnell.com/aim-tti-instruments/iprober-520/probe-current-positional-on-pcb/dp/1908148

Or just get a used HP 547A and some way to inject a square wave (eg. function gen). You should be able to find one for under $50, possibly including the current injector to generate the square wave.

[olsenn]

Your multimeter doesn't need to be accurate to detect shorts... even the cheapest Radioshack meter will suffice; if the ohms reading between the two points in question is less than (let's say) 1 ohm, then you have a short. You can make yourself a continuity buzzer if you feel like it

[alm]

If your DMM has insufficient resolution, the resistance will read a short anywhere on the nets, which if you're lucky are power and ground, with dozens of components across them. Now what?

[ejeffrey]

Your multimeter doesn't need to be accurate to detect shorts... even the cheapest Radioshack meter will suffice; if the ohms reading between the two points in question is less than (let's say) 1 ohm, then you have a short. You can make yourself a continuity buzzer if you feel like it

You don't understand.  The point is not to detect whether there is a short, but to find the exact location.  You run a lot of current through the short.  Then you use a sensitive voltmeter to track down the point of the short by measuring the voltage drops across the trace resistances.

[yanir]

Your multimeter doesn't need to be accurate to detect shorts... even the cheapest Radioshack meter will suffice; if the ohms reading between the two points in question is less than (let's say) 1 ohm, then you have a short. You can make yourself a continuity buzzer if you feel like it

I know I have a short, like you described I used the multimeter to determine that. I want to know where along the trace the short is.

[Short Circuit]

5mil traces have fairly high resistance; about 30mOhms per cm (using 35u copper), and you can safely send 1 amp through them.
Do that and you get 30mV voltagedrop for each cm to the short.

 Of course you should set the power source to a safe opencircuit voltage to prevent disaster when the short is removed!

[yanir]

Not jused it myself and not inexpensive ! but look really nice . Iprober 520 it a current probe for use on the PCB. www.farnell.com/datasheets/1387648.pdf and http://uk.farnell.com/aim-tti-instruments/iprober-520/probe-current-positional-on-pcb/dp/1908148

Or just get a used HP 547A and some way to inject a square wave (eg. function gen). You should be able to find one for under $50, possibly including the current injector to generate the square wave.

The micro can produce the square wave on the tx pin, and it shouldn't harm the bt module. I'll look into that. Thanks.

[alm]

I'm not suggesting that you should spend money on a current tracer to find a single short, but it's at least more than an order of magnitude cheaper than the TTi current probe, which is completely overkill for this application. They are great tools for this job, especially on boards with solder mask.

I agree with Short Circuit (who should be an expert on this subject, given the nick name) that supplying something like 3.3 V/1 A from a lab supply should work fine in this case. 1A should be no problem for the trace, and there's no danger for the components since the voltage will be close to zero as long as it's shorted. The 3.3 V limit is to limit the voltage in case the short is suddenly cleared.

[pickle9000]

You can always start cutting traces. Check the short with any ohmmeter (solder a couple wire on place to make life easy). Then make a guess and start cutting, splitting the difference is fastest. Typically done on the positive side (it's hard to isolate sections with the ground). The trick is to make sure they can be soldered back together again when your done.

If you are good at lifting pins lift the supply pins on the chips. This is a little dangerous but with practice not too bad. Lifting pins on dip packages is virtually impossible, for those a big old-fashioned "soldapullt" is the answer. Best solder sucker ever made.

Wire wrap wire and a sharp knife to put them back together. If you want them to look pretty matching fingernail polish works well.

If you are going to use the high current method, take a picture (both sides) or place the board on a scanner. Most scanners will keep in focus even if 10mm from the board, no macro lens required.  If you do blow a trace or worse then you can get part numbers and such off the picture.

...mike

[yanir]

The board is my design, and I can build another or replace the damaged part. I am trying do the least amount of rework to fix the issue.

the micro pin is rated at 25mA. If I put 3.3v @ 1A on this trace and the short blows out,  will that damage the pin?

Also the trace isn't very long, not much longer then acentimeter. Does that complicate things or is that enough?

[vk6zgo]

If you only have about a centimetre of trace,get a very bright light & a magnifying glass.
You will probably find it by eye,if it is a trace you can see,not inside a multilayer board.
Are you completely sure the short isn't at the micro pin itself?Or inside the device?

[Uncle Vernon]

How do you find shorts?

I find then much more comfortable than long trousers during warmer weather.  I generally find them in the drawer above the sock drawer.

If you only have about a centimetre of trace,get a very bright light & a magnifying glass.
You will probably find it by eye,if it is a trace you can see,not inside a multilayer board.
Are you completely sure the short isn't at the micro pin itself?Or inside the device?

Got it in one. No amount of tomfoolery is going to be as accurate as a visual or assisted visual inspection. Even with all but the most complex multi layer boards most faults can be visually identified. 

[JuKu]

If you can, cut the trace from both ends and rewire.

In general, I'd like to find a method or hints for this, too. My pet problem is bringing up a new board: The power supply hits the current limit so one of the parts is wired wrong and trying to self-destruct. But which one? This is especially hard problem if there is a power plane. (I guess I really need to buy the iprober...)

[yanir]

If you can, cut the trace from both ends and rewire.

In general, I'd like to find a method or hints for this, too. My pet problem is bringing up a new board: The power supply hits the current limit so one of the parts is wired wrong and trying to self-destruct. But which one? This is especially hard problem if there is a power plane. (I guess I really need to buy the iprober...)

That's probably the fastest way. It's right next to the Rx trace. So I will probably have to cut and rewire both.

[Fraser]

I own and use Polar Toneohm short tracers. They come in several versions. The simplest being an audio only signal injector and current pick-up probe. The more expensive model that I have uses audio short tracing and milliohm measurement. Amazing bits of kit and you may be able to recreate the functionality using nothing more than a current sourcing signal generator, simple inductive pick-up (try a cassette recorder head) and an oscilloscope to see the signal amplitude. The amplitude will decrease as you approach the exact location of the short.

On ebay Toneohm's cost between $50 and $1500 depending upon model. I have the 500, 700 and 850 models and recommend them all. They find failed Tantalum capacitors very easily so are an essential tool for owners of old Racal test equipment !

You can find the service manuals, including schematics, for the Polar Toneohm series here:

http://128.238.9.201/~kurt/manuals/manuals/Other/

They are not that complicated and the inductive pick-up is the only real challenge. I bought spare pick-up probes from Polar (~$60 each) as I couldn't decide how to best make a clone of my probes. They are likely to contain just a small pick-up coil on a ferrite core as they are VERY directional. Ther is no electronics within the probe itself.

This could be a fun project for someone with the available time to design a probe and PCB :-) 

Aurora

[robrenz]

This looks like it may be similar to what Aurora mentioned.  I watched the video and it looks pretty impressive.

http://www.eds-inc.com/leak.html

[yanir]

This looks like it may be similar to what Aurora mentioned.  I watched the video and it looks pretty impressive.

http://www.eds-inc.com/leak.html

Yeah, that's leak seeker I mentioned in one of my posts. It looks pretty easy to use. But I couldn't find any reputable dealer that carried it. It seems that it can be bought directly but, they seem pretty small, so I wasn't sure about it.

[Fraser]

Hmmm that Leekseeker seems similar to the Toneohms but my units don't do the auto window calibration. I will be honest and say I have never seen a need for such a feature. Be aware that the ToneOhms often have the superb signal injection and detection mode. In that mode you literally trace the conductor to the short and the signal disappears as soon as you pass the point of short circuit. No direct contact with the PCB track is needed for the probe and it can even trace a PCB track that is of unknown location inside the PCB  The probe sensitivity control allows trace detection from around 1mm up to as far away as 4cm ! The direction of trace travel may also be mapped as the probe head is very directional and has only one peak sensitivity orientation

Many PCBs that I have worked on use internal power rails and I often do not have a schematic. The inductive signal trace mode of the ToneOhm is simply amazing for finding the intrenal traces and the point of short circuit.

All of mine came from eBay with the 500 costing me only GBP30, the 700 was GBP40 and the 850 cost me GBP35 .... an amazing bargain in as new condition     Keep a look out on ebay for current tracer's, fault tracer's, PCB short finder's and ToneOhm's 

Aurora

[yanir]

If you only have about a centimetre of trace,get a very bright light & a magnifying glass.
You will probably find it by eye,if it is a trace you can see,not inside a multilayer board.
Are you completely sure the short isn't at the micro pin itself?Or inside the device?

I'm not sure. How can you test for that?

It's a simple trace that runs from the micro pin to the module pin, 1 to 1.5 centimeters. There are ground planes underneath each device, so I was thinking that the short is between pad to the gnd plane hidden from view (5mil space). But it could be a short in one of the devices. is there a way to rule that out before lifting pads or removing parts?

[vk6zgo]

If you only have about a centimetre of trace,get a very bright light & a magnifying glass.
You will probably find it by eye,if it is a trace you can see,not inside a multilayer board.
Are you completely sure the short isn't at the micro pin itself?Or inside the device?

I'm not sure. How can you test for that?

It's a simple trace that runs from the micro pin to the module pin, 1 to 1.5 centimeters. There are ground planes underneath each device, so I was thinking that the short is between pad to the gnd plane hidden from view (5mil space). But it could be a short in one of the devices. is there a way to rule that out before lifting pads or removing parts?

The visual check can pick up a lot of things--just because something seems the most likely shouldn't rule out checking for other things.

You did say "pin" so I assume the devices are normal SMDs with visible pins.
Again ,using the lamp & glass,you should be able to desolder the suspect pins,lift them ever so slightly ,& put a bit of thin paper (or even better,Mylar),under the pin,then check again for shorts.
If this is not practical,the suggestion, made in one of the other posts,of cutting the track seems the best option.

[yanir]

Fixed it! It was a short to the RF Shield Can. I was a little over zealous with the solder.

First, I lifted the micro pin, and the trace was still shorted to ground, but the pin was clear.
Next I wanted to lift the modules pin but it's a lead-less package (rf module on a pcb with pads on the bottom. So I couldn't do that (or remove the module easily).

So I then cut the trace, the short was still on the module side. I wasn't too happy, I didn't want to  have to remove the whole thing. But then I thought lets just try wicking away solder from the modules shorted pin, maybe I would get lucky.

And the short cleared! The solder must have crept up the side wall (it's like a via cut in half) to the can causing the short.

After putting on a bodge wire from the micro to the module my UART connection works like a charm.

Now I know to be a little more careful when hand soldering rf modules with cans. And I have a new test to run (check for ground shorts on the RF module pins).