Just Add Alcohol

[German_EE]

Once in a while I come across something very simple but amazingly useful, the latest came along yesterday. So, imagine the scene, you have a PCB full of ICs and one of them has a short between the 5v line and ground, how do you find it?

The action starts at 43:28 if you don't want to watch the whole repair.



Just a current limited 5v supply and some IPA on a cotton bud and he's saved hours of work lifting pins. 

[ebastler]

Nice -- poor man's thermal camera. 

I tend to just use my fingers directly to feel the temperature (or my lips for better sensitivity). But I see how this might be able to localize hot spots quicker and more clearly. And it looks more professional than the lip method. 

[RoGeorge]

Sorry, but no.

Would have been interesting to detect a short circuit of a component that does not heat, like a shorted MLCC on the power lines, but finding which IC is getting hotter is not so interesting.  Some parts are running hotter than others anyway.  Could simply touch them to feel the temperature instead of putting a drop of alcohol on each one.  Seems like a useless trick.

Did I missed what is actually happening there (didn't watched the full video)?

Later edit:
Maybe it'll work on shorted MLCC caps, too, don't know, next time will try it.  Thanks for sharing.

[kjr18]

Louis Rossmann uses alcohol to detect shorts all the time, and in his opinion it works better than any thermal camera.

[wraper]

Louis Rossmann uses alcohol to detect shorts all the time, and in his opinion it works better than any thermal camera.

For it to work, component should be frying. More often than not faulty part heats only by a fraction of degree a few a degree Celsius over ambient.

[ebastler]

For it to work, component should be frying. More often than not faulty part heats only by a fraction of degree a few a degree Celsius over ambient.

I think that's overstating the limitations of the evaporation method. Ethanol and IPA have a boiling point around 80°C, hence temperatures above 50°C or so should already become quite noticeable by accelerated evaporation. And on the other hand, many healthy components will warm up to "a few degree C over ambient" in most working circuits, right?

But keep the arguments coming; I could do with a justification to buy a thermal camera! 

[chris_leyson]

I tend to just use my fingers directly to feel the temperature

Did that once on a board with an ECL chip inserted the wrong way around, it's got to be one of these chips...ouch, took a chunck of skin of my finger tip. Years later I used the back of a middle finger to find a warm inductor in an antenna tuner, oh look, sparks... that took three months to heal. I would go with the alcohol method 

[wraper]

hence temperatures above 50°C or so should already become quite noticeable by accelerated evaporation. And on the other hand, many healthy components will warm up to "a few degree C over ambient" in most working circuits, right?

50^oC is quite frying in my opinion. Most components don't heat up above PCB temperature. Also you can compare it with working board. For example I know that Raspberry pi CPU has a hot spot on it's corner of about 32-33^oC at room temperature (without booting). If it is a few degrees higher or has another hot spots, CPU is faulty.  Shorted MLCC usually have very low resistance, so you need to push quite a bit of current into the power rail for it to become visible even in thermal camera. Sometimes it's exactly opposite, faulty component is cooler than it should be. Thermal camera saves hell a lot of time when doing repairs. Very often it takes a few seconds to diagnose what would take hours otherwise.

[wraper]

Years later I used the back of a middle finger to find a warm inductor in an antenna tuner, oh look, sparks... that took three months to heal. I would go with the alcohol method 

Instead you would get.. ignition, liftoff 

[RoGeorge]

alcohol to detect shorts

Does this kind of trick works to detect shorted MLCC capacitors, too, or is it just for ICs?

[wraper]

alcohol to detect shorts

Does this kind of trick works to detect shorted MLCC capacitors, too, or is it just for ICs?

It could sometimes, but in most cases no. Usually too low resistance, therefore not much power dissipated to rise temperature significantly, even if you push a few amps through the cap. Alcohol certainly sucks to detect real shorts, only useful for increased power consumption.

[kjr18]

alcohol to detect shorts

Does this kind of trick works to detect shorted MLCC capacitors, too, or is it just for ICs?

I don't remember right now if it also worked for caps, he's using it for detecting what components are shorted to ground on macbook motherboards. As it has many power rails, with many things that might fail short, he uses lab power suppy to feed low voltage, high current into power rail, and pours alcohol to see where it evaporates first, so he does it on not really working board.
For example in this video (around 19:00):

[Berni]

Yeah the trick does work, PCB cleaner can also be used for it since its mostly a mix of alcohol.

But for hard enough shorts it might indeed not get warm, especially if a PSU quickly gives up and starts power cycling. If i can't locate it in a reasonable amount of time then i go for the brute force method. Solder some wires onto the problematic supply rail and hook it up to a powersupply set to the voltage that rail is supposed to be. Then i just increase the current until something does start getting warm. If nothing else, at some point the PCB trace going to the faulty component will get hot.

Actually i fixed things with my brute force method before. If the short is something small like a tin whisker or a stray solder ball from a bad reflow process then enough current sometimes burns it open and it starts working again.

[RoGeorge]

For example in this video (around 19:00):

The cap in the video has about 900 ohms (measured in the video at 26:40), not sure if the evaporation trick still works with close to zero shorted caps.

Will give it a try next time, thank you for searching and linking the video.

[ebastler]

The cap in the video has about 900 ohms (measured in the video at 26:40)

No, I understand that the 900 Ohm resistance is what he measured in the circuit after removing the cap (to confirm that the short was gone).

[TheNewLab]

Excellent idea! I always have IPA at my lab desk!

But keep the arguments coming; I could do with a justification to buy a thermal camera! 

Justification needed? thermal camera is in COLOR!

[james_s]

Yeah the trick does work, PCB cleaner can also be used for it since its mostly a mix of alcohol.

But for hard enough shorts it might indeed not get warm, especially if a PSU quickly gives up and starts power cycling. If i can't locate it in a reasonable amount of time then i go for the brute force method. Solder some wires onto the problematic supply rail and hook it up to a powersupply set to the voltage that rail is supposed to be. Then i just increase the current until something does start getting warm. If nothing else, at some point the PCB trace going to the faulty component will get hot.

Actually i fixed things with my brute force method before. If the short is something small like a tin whisker or a stray solder ball from a bad reflow process then enough current sometimes burns it open and it starts working again.

I once fixed a Space Duel arcade board a bit like that. I got the machine non-working and brought it home, powered it up and was very surprised to see it come up to the attract mode but the vectors looked all blurry. I checked and was horrified to see that the 5V rail was up over 8V so I shut it off. Repaired the power supply and it has been working now for almost 20 years since then. I never did figure out what the original problem was, the guy I got it from said he had checked the 5V and it was fine but the board wouldn't boot at all, so it seems one fault fixed another.