Short Finding on a PCB

[MultiMike]

After my adventure with the industrial controller that has now blissfully come to a close, I find myself wanting to arm myself for the next time this kind of thing happens.

Firstly, I have every intention of building the Shorty device designed by Kripton2035 on this group.
Secondly, I have every intention of getting an el-cheapo thermal imaging camera to keep at home, because that was near miraculous in finding the short.

What this topic is about is the question of how many methods are required?   My understanding (such as it is!) is that I was able to use the thermal camera only because of the kind of short that I was chasing.   Are there different kinds of shorts that require different techniques?  What are the items that you feel are 'must-haves' for short finding?

Mike

[masterx81]

Usually i use a 4w sensitive ohmmeter (34401a), an adjustable psu and a thermocamera. If the basic tests with the ohmmeter doesn't find the short (but you can identify the area of interest), i start to put 50/100/150...ma up to 1a to the short while monitoring with a thermal camera. A small capacitor can heat up and become visible with as low as 50ma (if dead short, not totally shorted can be 100/200ma), a semiconductor with good thermal dissipation (smt devices with thermal pad) can need up to 1a to show a rise of few °c. A good psu is a must. It must not have any poweron spike, good current regulation, and an on/off button. And avoid to "hot plug" the pcb under test while the psu is powered on, as all psu have some output capacitance and is sufficient to put some serious currents if hot plugged. Use always the on-off button. I use a korad 3305, cheap and do it's job.
I'm also interested in the shorty device, for sure it can save a lot of time in the first test of finding the short.

[PeteH]

I have also applied a current, similar to the thermal camera method, and hunted with a volt meter. Closer you get to ~0VDC, the closer you are to the short. (Layout knowledge required)

[DavidAlfa]

I always do that way. Set the correct voltage, limit the current to 0 (So the power supply actually cuts the output), connect, and slowly rise the current.
Otherwise, it's impossible to find the short when there're thousands of parts connected to the same power rail.
Oh, and nothing  of thermal cameras. Cheap finger temperature sensor.

[Specmaster]

You can also the limited current approach in conjunction with something like IPA or liquid flux applied to the suspect area and watch where it evaporates the fastest.

[masterx81]

I always do that way. Set the correct voltage, limit the current to 0 (So the power supply actually cuts the output), connect, and slowly rise the current.
Otherwise, it's impossible to find the short when there're thousands of parts connected to the same power rail.
Oh, and nothing  of thermal cameras. Cheap finger temperature sensor.

The advantage of thermal camera is that it can find a temperature increment with way more precision. Can detect a few °C increase, with the finger i start to "feel" only when the temperature increase is more that 7~10°c, with thermal camera 2°c is enough to spot it, and for that differenzce you may need to double the current. I've used for years the "finger" approach, buh with a thermal camera it's all more easy. And i use a cheap camera. Some power ic's are so well cooled by the pcb that can request a lot of current only for get few °c of increase. More current may put stress also on other things (like traces on the pcb).
In any case before putting current to the circuit i always prefer a check with milliohm meter.

[nctnico]

A thermal camera and a current limited PSU are the best tools. I have a Flir C200 (about the cheapest Flir with a real thermal imager) which works really well for electronics purposes.

But still a milli-Volt meter can be necessary to figure out where a short is if it is in a big component. On various occasions I had a short inside a big FPGA and that didn't show using the thermal camera.

[Smith]

I had a nasty short on a relatively big PCB (>A4 sized TV main board). I couldn't find the shorted device, and the short was hundreds of milliohms. The short was on the 3V3, and most of the hardware was running on 3V3. I soldered the 4wire wires from the multimeter on the PCB to keep it as stable as can be. Then I just sprayed freeze spray on every device until I found out it was the HDMI switch. This freezing lowered the resistance a few milliohm. Any other cooling component had no measurable effect parallel on the short. Of course heating the component would have about the same effect.

Was a nice and fast cost effective way, way before thermal camera's became somewhat affordable.

Just to be clear, the PCB had at least 4-8 layers, and the tracks where very thin. I did not want to blow a track by trying to heat up the component by applying current to the PCB.

[Shock]

Minimum is a multimeter with ohms. Aside from various methods of detecting and measuring temp change, the most advanced would be a current probe or use of signal injection and measurement. As mentioned you can use the voltage drop method (microvolt range) or low ohms method (milliohm or ideally microohm resolution).

External voltage source can be something as simple as a battery, but a bench supply with current limiting and milliamp display adds a lot of flexibility.

[nigelwright7557]

I have seen  it done using a decent DMM on low ohms.
The pcb had a shorted cap so he just moved around the pcb and checked resistance across the caps until he found lowest.

I guess easiest way is thermal camera and look for hot spots.

I just built a new pcb myself today and it didnt work.
-12 volts was 0 volts and 3.3 volts was 2 volts !

Turned out I messed up CAD software clearance check and didnt spot a zero volts to -12 volts short where 2 track crossed..
Quick cut and the pcb burst into life.
Will take more care next time.

[rfclown]

I don't have a thermal camera at home. At work a thermal camera often finds the problem. When other methods fail I'll use a mondo power supply. I have an HP supply that will do 15 amps. Look for smoke. This is assuming the issue is on a supply which has beefy metal to all things powered on the supply.

[David Hess]

If I had one, my preferred method would be a multimeter which has in circuit current measurement capability.  These multimeters apply a current between the probes to measure the resistance of the trace, and then use that resistance and a voltage measurement to calculate the current through it.  The Keithley 2001 and 2002 multimeters support this.

[Shock]

If I had one, my preferred method would be a multimeter which has in circuit current measurement capability.  These multimeters apply a current between the probes to measure the resistance of the trace, and then use that resistance and a voltage measurement to calculate the current through it.  The Keithley 2001 and 2002 multimeters support this.

Sounds like you are referring to the "IN-CIRCUIT" DCI measurement mode. Is this anymore advantageous than a 4 wire DCR measurement? Seems like it's basically the same thing.

[kripton2035]

in my experiments, you need around 1-4mΩ resolution to locate a shorted component on a pcb.
if you have a 2-wires probe, you can at best go as low as 10-20mΩ
a 4-wires probe is mandatory for this job.

[David Hess]

If I had one, my preferred method would be a multimeter which has in circuit current measurement capability.  These multimeters apply a current between the probes to measure the resistance of the trace, and then use that resistance and a voltage measurement to calculate the current through it.  The Keithley 2001 and 2002 multimeters support this.

Sounds like you are referring to the "IN-CIRCUIT" DCI measurement mode. Is this anymore advantageous than a 4 wire DCR measurement? Seems like it's basically the same thing.

It initially works similarly to a 4 wire resistance measurement, but the excitation is AC to measure the resistance in the presence of a voltage offset, and then the measurement returns current=voltage/resistance.

[1audio]

I have a gadget from many years ago that is ideal for finding PCB shorts. Its called a short squeek. It was made by Global Specialties. It must have been expensive new or there would be hundreds out there.
Essentially it makes a high pitched squeek that gets higher in pitch as the resistance between the probes goes down. It senses down to small lengths of copper trace. Its really fast to use. Fortunately I don't need to deal with this often but much faster and easier than a milliohmeter (I have 2) for this task. If you see one get it.

[kripton2035]

the shorty device does the same, a tone that goes higher as the resistance goes down.
my shorty-with-display does it too, with an added display of the resistance that is better at finding very low ohms IMHO.

[shakalnokturn]

Yet another example is the HY-TRAK 100D (photo), it's a 4 wire Ohmmeter with a sound that has a frequency inversely proportional to measured resistance.

As you already know, my preferred method is the thermal cam and lab PS combination.

[Whales]

Cheap cheat alternative to thermal cameras that's worth mentioning: pour alcohol over the board and watch where it evaporates the quickest.

[David Hess]

Cheap cheat alternative to thermal cameras that's worth mentioning: pour alcohol over the board and watch where it evaporates the quickest.

I have done that using freeze spray.

[MultiMike]

if you have a 2-wires probe, you can at best go as low as 10-20mΩ
a 4-wires probe is mandatory for this job.

My gripe about the 4-wire 'kelvin clip' things is the sheer size of them.  When I was dealing with the board that started this whole thing I was using a sharp-point probe for searching and a meat-hook for the ground.   This big clips... I haven't used them, but they don't look like it would be feasible to do the searching with them.
I'm willing to be found wrong, though.

Mike

[kripton2035]

you can make yourself a probe for the ground, and the other one with some sort of steel brush to find roughtly where the short is, then use a sharper probe.

[Shock]

There are also various milliohm meter projects that use the multimeter to display measurements.

 

[AndrewBCN]

There are also various milliohm meter projects that use the multimeter to display measurements.

These are probably adequate for measuring milliohms but not very practical for finding shorts on a PCB, because what you want is the audio feedback telling you if you are near/at the short without taking your eyes off the PCB on which you are working, which is the main advantage of the Shorty/Shorty-with-display. The way I see it, the display in the Shorty-with-display is for confirmation.

[Greybeard]

Some methods are described here:
https://www.artisantg.com/info/Polar_Toneohm_850A_Manual.pdf

[paul@yahrprobert.com]

For years I've been using the following method:
  1) Connect a signal generator to your short and pump a hundred ma or so into it with a sinewave about 500 Hz.
  2) Make a small pickup coil with fine wire and a hundred turns or so.  When I get back home in October I can take a picture of my coil, it has a ferrite core about 5 mm OD with a slot cut in it, and glued onto the end of a plastic rod.  Another source of good coils are old relays.  Just crack them apart and take out the coil and core.
  3) Make an audio amplifier with a lot of gain, maybe 5000 or so.  When I get home I'll dig up the circuit diagram.  3 transistors and misc parts.
  4) Then connect some headphones and start tracing.  Its amazing how you can narrow down where the current is flowing.  Also good for finding ground loops.

[Shock]

These are probably adequate for measuring milliohms but not very practical for finding shorts on a PCB, because what you want is the audio feedback telling you if you are near/at the short without taking your eyes off the PCB on which you are working, which is the main advantage of the Shorty/Shorty-with-display. The way I see it, the display in the Shorty-with-display is for confirmation.

I use a camera with 14x zoom that outputs to an video display so can freely look at instruments even if under magnification.

The tone function is almost entirely superfluous in my opinion. Products like the leak seeker use range windows to allow tone resolution, but it's not as simple to operate. The shorty w/display itself is unnecessary if you have other equipment, but the advantage is it's easier setup than the voltage drop, current and signal injection methods.

I'll take resolution over audio queues and led lights (other designs) any day.

[Shock]

For years I've been using the following method:

Yes the inductive method has it's advantages, can also be used with a scope like near field probes to find dead integrated circuits.

[edigi]

The method to find a short on a PCB can vary a bit depending on the actual case. If using that much power that the generated heat shows up in thermal camera is not an option then the other widely used method is a current limiting PSU + a good resolution voltmeter (accuracy is not a concern here). If a 6.5 digit benchtop DMM is not available another alternative is a handheld multimeter with high resolution mode (like Brymen BM869s).
By checking the voltages at various places of the PCB the short can be usually figured out within reasonable time (still slower than with thermal camera).

It's a kind of 4W resistance measurement but when measuring very low resistance it's always better to use a controllable current and voltage measurement (so DMM is in voltage measurement mode).

Finding shorts in RF stuff follows an entirely different principle, that is based on the time delay it takes to sense that there is a short (TDR). If a fast scope and fast edge signal generator is available that can be used directly but if not then via a signal sweep and inverse FFT (the VNA/AA way) to get back to time domain the distance to short can be also determined (FDR). These methods require knowing the propagation velocity though.

So there are really a plethora of options depending on the case...

[Shock]

High Q testers (aka ring/lopt/fbt testers) are useful for finding inductor/transformer leakage and shorts. Inductors typically look like a short to DC, so the tester applies a quick pulse and determines how good the decaying oscillations are, essentially how high the Q (quality) is.

The model I have is an old Dick Smith kit shown in the middle.

[nigelwright7557]

I designed a pcb recently and got it made in China.
When I powered it up no -12volts but 12 volts was ok.
Changed regulator but that didnt fix it.
So started looking for shorts on pcb.
I then spotted what looked like an odd track configuartion.
So looked it up in my cad software and 2 tracks were crossing which I rarely do.
Then noticed they had 2 different net names so shouldnt cross.
I did a quick clearance check and it failed.
I must have forgot to do a final clearance check before getting pcb made.
Luckily it was an easy single cut to fix it.

I built up a different pcb and couldnt get my pickit3 to run the code.
SO went back to basics and just had PIC and critical components installed.
Still didnt work.
Replaced PIC but still the same.
So built up another pcb using components from previous pcb and it worked.
So must have been a pcb fault. I can only guess a missed via.

[paul@yahrprobert.com]

n a previous post I described my setup for finding shorts or ground loops. I promised some pictures.
Here's a picture of my pickup coil:

That is an Amidon T-23-J core, J is the high permeability material.  There's about 80 turns of #37 wire on the core.
My audio amplifier is as shown:

Here's a link to a video showing the action of this setup on a 20 ma p-p 500 Hz square wave:
https://drive.google.com/file/d/1IqwqOlvJwNok4HL6XdDo0dnJLdaOAQS_/view?usp=sharing

[iMo]

How did you cut the slot?

[paul@yahrprobert.com]

I used a thin diamond dremel saw.

[dazz1]

n a previous post I described my setup for finding shorts or ground loops. I promised some pictures.
Here's a picture of my pickup coil:
(Attachment Link)
That is an Amidon T-23-J core, J is the high permeability material.  There's about 80 turns of #37 wire on the core.
My audio amplifier is as shown:
(Attachment Link)
Here's a link to a video showing the action of this setup on a 20 ma p-p 500 Hz square wave:
https://drive.google.com/file/d/1IqwqOlvJwNok4HL6XdDo0dnJLdaOAQS_/view?usp=sharing

Hi
Have you tried a video tape recorder play head?

[kripton2035]

...without taking your eyes off the PCB on which you are working, which is the main advantage of the Shorty/Shorty-with-display. The way I see it, the display in the Shorty-with-display is for confirmation.

when you search for short on a pcb, like a shorted capacitor, only hearing the sound will not let you find the exact faulty capacitor
with the display, you can make the difference that will isolate the capacitor.
see one of my video about it the difference is less than a milliohm.

[Shock]

The inductive method should work reasonably well. I'm not sure it's as useful, but there's probably situations where it's more useful. If you look at some of the short finders they also include an inductive or hall effect sensor.

The audible tone method as an indicator of proximity is ok as long as you have appropriate sensitivity. As kripton2035 says though, if it's virtually a dead short across a component having resolution on a display can help make it easily distinguishable from a neighboring component.

[DavidAlfa]

I usually inject 50mA and slowly rise the current until something starts heating up.

[MarginallyStable]

It it is a component shorted.... I have often had success with using a precision ohm meter across the short, and selectively applying freeze spray to locate the component that affects the resistance the most. Remove the component to verify/repeat.

[paul@yahrprobert.com]

Where the inductive method really shines is when you've got a larger system, or a whole lab (I grew up in a fusion research lab) when you've got a ground loop  somewhere in a cable tray with hundreds of wires. When you've got a short on a pcb the small probe lets you find the trace where the current is flowing, but if you have a multilayer board with a very dense layout you can get lost for sure.

[scopeman]

I have made PCB's for this project but have not had the time to build it yet.

https://www.edn.com/milliohm-squawker-great-at-finding-shorts-and-reverse-engineering-pcbs/

Sam
W3OHM

[Zucca]

Esoteric...