Trying to repair my 80's vintage function generator, an IEC F72. Found the 5V rail was low. Found it was actually shorted when I pulled the three terminal regulator and connected my lab supply and saw it current limit!! I have a schematic, and the board is a beautiful 80's through hole lab quality piece. The voltage rails are supplied through a few jumpers here and there, so I desoldered the jumpers to isolate the portion of the board with the short. Unfortunately, I isolated it to a still large section of the circuit, and nothing is getting hot!! Obviously, I checked the electrolytics, they are still fine. I tried using air in a can to put frost on the board and look for something warm, but I'm stumped. One difficulty is that the circuit branches to two subsections from the jumper I have isolated the problem to. I was thinking I could wind some wire around a piece of ferrite and use it with an oscilloscope to see an increase in potential and crudely determine which trace is carrying the several amps of current, and maybe follow it. Anyone have any experience or tips? Would that even work for detecting DC with sensitivity sufficient to track the current?
Thanks for any advice!
KI4TJ
I find shorts on laptop main boards by applying some current to the shorted power rail, then measuring voltage drops on the ground plane. This avoids having to know how the power rail track is routed.
hp used to make a current probe, the 428x, that used a ferrite core as a flux gate magnetometer. Applied Physics makes a newer version. Here's a link to one person's experience with the ilk:
http://www.prc68.com/I/HP428.html The article discusses how it works and alternate solutions.
Low voltage ceramic and tantalum caps can also short circuit. Since this is a 5 V circuit, could there be any transorbs or ICs that have failed?
A good 4 1/2 digit DMM can easily show the mV voltage drops through typical traces with currents on the order of 100 mA. It may take a little while to puzzle it out though and a map with the various voltage readings can make it easier.
Cheers,
A sensitive measurement of DC current by the magnetic field is difficult. It needs a reasonable higher current to make it work.
If enough current is used, one might be able to use a good hall effect of GMR based sensor. A flux gate probe is tricky to get very small and might be influenced by other effects (like ferrites at larger distance, mybe eddy currents). Hall effect sensors tend to have quite some drift.
The alternative way to trace current flow is looking at the low voltage drop, so using a high resolution voltmeter and the copper resistance.
In my humble experience an infrared camera never disappointed me.
Zucca's advice is right .......I have had great luck with a FLIR as well........the shorted area glows like a bulb on an IR image.
Alternatively , solder a small loop of wire on the stripped and exposed end of a rigid coax between the center and outer conductor and voila........you have a small flux probe that you can scope.
But the loop will not work for direct current only for A.C.
cin cin.
In my humble experience an infrared camera never disappointed me.
That's a bit expensive option tho
If you don't have FLIR, apply external power and use canned air held upside down as a freeze mist and find a component that thaws out fastest. Do current limit or you can really toast a board! Pray it isn't in the PCB itself.
Wow, thank you so much for all the really useful replies!
Old don: i tried the freeze spray and i tried putting my hand on the board. I put 5amps at 5volts through it and i cant find it! I’m confused as heck at that result... Fortunately its 70s or 80s vintage through hole 2 layered board so unlikely a short in a hidden trace.
Zucca: yeah.. i may spring for a flir but given the above im sceptical of finding a hotspot. Im keeping that as my last option if i exhaust cheaper options
helius: those links are solid gold. The webpage gives a rundown of several techniques, including what duak, kleinstein, and xavier60 mentioned following voltage drops.
Duak: none of the ics get hot and no transorbs. There are some tantalums and ceramics. I didnt know they were likely to short. None of them get hot but i will look a bit more closely.
Ill try measuring the voltage drops first, not sure if my dmm is sensative enough but we will see.
5amps at 5volts, that's 25 watts!. Maybe the voltage was dropping to something much lower while loaded.
I recently started using an AN8008 multimeter with 1uv resolution, cheap also. Just need to be aware of a possible dead spot in the mv range. Mine is +/-5uv, not a real problem though.
I use 200ma on laptop main boards to find shorted surface mount capacitors in minutes. I don't follow voltage drops on power rail tracks. I do all of the measuring on the ground plane.
it is best to connect the current source negative and DMM negative separately to the ground plane.
If you don't have FLIR, apply external power and use canned air held upside down as a freeze mist and find a component that thaws out fastest. Do current limit or you can really toast a board! Pray it isn't in the PCB itself.
Just be careful that the board doesn't crack due to the thermal stress.
My method of choice: Infrared thermometer, or moving my hand over the circuit and touching stuff. Most of the time you can feel that there's abnormal heat before you burn your finger.
Managed to repair 3 Xboxes this way - It's almost always a capacitor in never devices.
In older equipment I'd look at transistors first.
Here is my (SHORT FINDER) It works great.
If you have a good short it will make NO heat so looking for heat will not work.
If the short is between V+ and ground plane you don't have to worry about current going into decoupling caps, it won't, no voltage build up no current.
This uses any analog out hall sensor that has an output at 1/2 V+ with no magnetic field. The one I used is out of production.
This passes a current of 250mA through the short turning it on & off at about 650HZ.
The sensor is on a cable that is passed over the pcb searching for the short.
The hall signal is ac amplified about 900 x.
U3B and U3D pass the signal non-inverted or inverted synchronous to the 650hz current.
So you can run the sensor over a (week) magnet and not interfere with detection.
Instead of the output going to a VOM it should go to a voltage controlled oscillator / speaker.
If there is dead short, it will definitely dissipate heat and will glow on thermal camera or burning fingers. But if you have unusualy low resistance on some components, camera will show it as normally heated component, may be slightly brighter.
Cutting traces and lifting legs or whole components is ol’ school method.
Some time ago I found another interesting approach for finding shorts. Louis Rossman describes it in one of his videos: - finding shorts by evaporation of IPA
Also I suggest to check the electrolytic caps, especially on the power rail. I had a few ones shorted in old equipment.