Jump to content

Finding Short Circuits

From RetroTechCollection
Failed dipped tantalum capacitors
Dipped tantalum capacitors from a Heathkit oscilloscope that failed short circuit and showed little outward damage (photo: WizardTim, CC BY 4.0). A shorted part on a supply rail often looks like this: nothing to see.

Finding short circuits on a circuit board means locating the part, joint or piece of copper that joins a supply rail, or a signal, to ground or to another net. A shorted rail stops a machine dead: the power supply shuts down, a fuse blows, a regulator overheats or the rail sits at zero.[1][2] A multimeter shows that a rail is shorted but not which part on it is to blame.[1] The methods that narrow it down are isolating sections of the rail, injecting a limited current and finding the part that warms up, and following the voltage drop along the copper.

An open circuit, where a signal does not arrive, is found with Continuity Testing and Signal Tracing. A failed part found here is replaced with the methods in Through-Hole Desoldering and Surface-Mount Rework.

When to use it

[edit | edit source]
  • A supply rail measures close to zero ohms to ground with the power off, or much lower than the same rail on a known-good board.[3][2]
  • The power supply shuts down, or keeps trying to start and shutting down again, when the board is connected. Goldwasser describes a supply that pulses in this way as shutting down on overcurrent or overvoltage.[4]
  • A fuse blows again as soon as it is replaced; see Fuse Identification and Replacement. The Sega Dreamcast Controller Port Fuse Replacement tells the reader to find the short before fitting a new fuse.[5]
  • A part runs much hotter than it should. Ray Carlsen notes that a shorted C64 RAM chip often gets very hot to the touch.[6]
  • A signal line is stuck at one level. Hewlett-Packard describes a node stuck by a shorted input, a solder bridge to another node or a dead driver; see also Logic Chip Testing.[7]

The parts that fail short on a rail include integrated circuits, transistors, diodes and multilayer ceramic capacitors.[1] Solid tantalum capacitors also fail short, and a shorted tantalum can take a regulator or a fuse with it.[8] Shorts in the copper itself include solder bridges and unetched copper filaments between a track and a ground plane.[7][9]

Safety

[edit | edit source]
  • Inject current only into the low-voltage DC rails of a board. Never connect a bench supply to the mains side of a power supply, and follow the CRT Discharge Procedure on any machine with a picture tube.
  • Keep the injected voltage between zero and the rail's normal voltage, and keep its polarity the same as the rail's.[1] Reverse voltage is among the causes of electrolytic capacitor failure that Nichicon lists, so a rail full of electrolytics must never see the supply backwards.[10]
  • The shorted part gets hot by design. Zhu and Morris keep case temperatures below about 50 °C, and recommend not exceeding 70 °C, to avoid damaging parts during the search.[1]
  • Freeze spray charges up as it leaves the can and can put a static charge on the board; Techspray sells an anti-static version for static-sensitive work.[11]

Tools and materials

[edit | edit source]
  • A multimeter with a resistance range and a 200 mV range. A 3½-digit meter resolves 100 µV on its 200 mV range.[12]
  • A bench power supply with an adjustable current limit.[13][1]
  • Something to see heat: a thermal camera, an infrared thermometer or a thermocouple, or a fingertip.[12][13]
  • Electronics freeze spray, for the frost method.[11][14]
  • Isopropyl alcohol, for the evaporation method.[12]
  • The circuit diagram and component layout, to know which parts sit on the shorted rail.
  • A known-good board of the same type for comparison, if one is available.[1]
  • Optionally, a current tracer with a logic pulser for shorts between low-voltage nets, as described by Hewlett-Packard.[7]

Procedure

[edit | edit source]

Confirm the short and find the rail

[edit | edit source]
  1. Switch off, unplug the supply and remove every cable, cartridge and expansion card.
  2. Measure the resistance from each supply rail to ground at the board's power input. A brief beep that stops is the rail's capacitors charging, not a short.[2]
  3. Compare each reading with the same rail on a known-good board.[3] A short is not always close to zero ohms: in Zhu and Morris's case study the 12 V rail measured 16 Ω, which was not a dead short but was abnormal, and the cause was one leaky ceramic capacitor.[1]
  4. Note which rail is affected and list the parts the circuit diagram shows on it.

Isolate sections of the rail

[edit | edit source]
  1. Take out socketed chips on the affected rail and measure again. If the reading returns to normal, refit the chips one at a time, measuring after each, until the short comes back. Lifting chips until the faulty one is found is the method Hewlett-Packard describes; a socket makes it harmless.[7]
  2. Where the rail feeds a sub-board, a port or a connector, unplug what is attached and measure each side.[5]
  3. If the short is still not found, the remaining method of isolation is to lift chips or cut tracks. Hewlett-Packard warns that doing this blind is slow, tedious and often damaging to the board and the chips, so use the methods below to point at a part first.[7] Commodore's own diagnostic manual uses a targeted version: to find which RAM chip is loading a line, it cuts the +5 V pin of each chip in turn.[15]

Inject current and find the hot part

[edit | edit source]

Zhu and Morris (Rockwell Automation) published this method in Electronic Device Failure Analysis in 2021. The figures assume a room at about 25 °C.[1]

  1. Connect the bench supply across the shorted rail and ground, at a convenient point on the rail such as the leads of a capacitor on it. In their case study the wires were soldered to a capacitor on the rail.[1]
  2. Set the current limit to 0.2 A and the voltage to zero.[1]
  3. Raise the voltage slowly towards the rail's normal voltage, watching the board, both sides, with the thermal camera.[1]
  4. Look for any part that rises above 40 °C. Let it settle, but do not let it pass 50 °C; if it does, turn the voltage down.[1]
  5. If nothing reaches 40 °C at the 0.2 A limit, raise the limit in steps (0.5 A, then 1 A, then 2 A) until a part does.[1]
  6. Confirm the suspect. With a known-good board, the same part running more than 10 °C hotter confirms it. Without one, take the part to 60 to 70 °C to be sure.[1]
  7. Read the supply as well as the camera. A very low voltage at a high current means a very low-resistance short, and then a rise of only 5 to 10 °C on a part may be the fault. Capacitors, transient suppressors and varistors heat very little even when shorted.[1]
  8. Remove the suspect part and measure the rail again.[1]

Without a thermal camera, the same injected current can be found by other means:

  • Feel for the part that is hotter than its neighbours, or use an infrared thermometer or a thermocouple.[12]
  • Spray the powered board with freeze spray. Within a few seconds moisture from the air frosts the board; the frost melts first over the part that is heating.[11]
  • Wet the board with isopropyl alcohol and power it; the alcohol evaporates first from the hot part.[12]

Follow the voltage drop

[edit | edit source]
Finding a short by voltage drop (diagram)
Following the voltage drop. With a small current injected into the shorted rail, the reading from the supply's connection point grows along the copper that carries the current (A to D) and stops changing on a branch that carries none (E) or past the short (F). Drawn for this wiki.

A short with almost no resistance, such as a solder bridge or a copper filament, may not get warm at all. The copper between the supply and the short still drops a small voltage, and that can be followed.[9][12]

  1. Connect the bench supply between the shorted nets with a low voltage and a modest current. LarryD on the Arduino forum uses about 0.3 V and 300 mA, which he chose so that neither the parts and their protection diodes nor the tracks would be damaged; contributors on the EEVblog forum use 0.5 V or less with a current limit of 100 mA to 1 A.[9][12]
  2. Set the meter to its 200 mV range. Leave one lead on a fixed point, for example where the supply connects, and touch the other along the shorted track.[9]
  3. Watch how the readings change from point to point; their size does not matter. The reading changes along the copper that carries the current to the short, and stays the same along any branch that carries none.[9]
  4. Follow the change towards the short. On a ground plane, one EEVblog contributor references the meter near where the supply's negative lead joins the plane and looks for the highest reading, which he finds at the ground end of the shorted part.[12]
  5. Once in the right area, measure the voltage across each part there that is connected to the rail.[12]
  6. Inspect that spot under a microscope or magnifier. LarryD's example ended at an unetched copper filament joining a +5 V track to the ground plane.[9]

At 1 A, a meter with 100 µV resolution resolves 100 µΩ of copper, which is enough for most tracks.[12]

Shorts between signals: the current tracer

[edit | edit source]

Hewlett-Packard's 547A current tracer senses the magnetic field of current steps at its tip, so it can follow current along tracks and wires without touching them. For a supply-to-ground short, HP disconnects the power supply, pulses the supply terminal with a logic pulser, and follows the path that carries the most current. For a stuck signal, a tracer shows whether the driver is pushing a large current into the node (something is clamping it) or almost none (the driver is dead).[7]

Checking the repair

[edit | edit source]
  1. With the suspect part removed or the short cleared, measure the rail to ground again and compare it with a known-good board.[1][3]
  2. Power the board from the current-limited bench supply, with the limit set just above its normal draw, so that any remaining fault cannot do further damage.[3]
  3. Measure every rail against the machine's own specification. Zhu and Morris confirmed their repair by checking that every rail came back within specification.[1]
  4. Find out why the part failed before connecting the original power supply. Carlsen traces shorted C64 RAM chips to failing power supplies and checks the supply's output before the repaired machine is powered from it again.[6]

Common mistakes

[edit | edit source]
  • Injecting a voltage above the rail's normal voltage, or with the polarity reversed.[1][10]
  • Turning the current up quickly. Zhu and Morris increase the power slowly because a small part can overheat before the camera shows it.[1]
  • Watching only the top of the board. Parts on the underside short too.[1]
  • Ruling out a capacitor because it stays cool. Capacitors, suppressors and varistors may warm only slightly.[1]
  • Removing parts one at a time on a guess, which takes time and risks damaging the board.[1][7]
  • Reading the rail's capacitors charging as a short.[2]
  • Replacing the shorted part and powering up from the same faulty supply that killed it.[6]

Worked examples on this wiki

[edit | edit source]
[edit | edit source]

References

[edit | edit source]
  1. ↑ 1.00 1.01 1.02 1.03 1.04 1.05 1.06 1.07 1.08 1.09 1.10 1.11 1.12 1.13 1.14 1.15 1.16 1.17 1.18 1.19 1.20 1.21 1.22 1.23 Zhifeng Zhu and Garron K. Morris (Rockwell Automation), "Finding Shorted Components on Printed Circuit Boards by Infrared-Based Direct Current Injection Method", Electronic Device Failure Analysis, vol. 23, no. 1, February 2021, pp. 22-26, online edition: components that fail short on power rails (ICs, transistors, diodes, MLCCs); a multimeter finds the shorted rail but not the part; inject from 0 V up to the rated rail voltage with a 0.2 A limit, then 0.5, 1, 2 A; look for parts over 40 °C, keep under 50 °C, case temperature recommended not to exceed 70 °C; compare with a golden sample (over 10 °C difference) or take to 60-70 °C; monitor both sides; TVS, MOV and capacitors heat little; a 5-10 °C rise can matter at low voltage and high current; increase power slowly to avoid overshoot; case study of a 16 Ω 12 V rail, a hot C137 at about 37 °C, all rails within specification after removal. Retrieved 25 October 2026.
  2. ↑ 2.0 2.1 2.2 2.3 SparkFun Electronics, "How to Use a Multimeter", section "Continuity" (CC BY-SA 4.0): check continuity between VCC and GND with the power off; a beep means a short; the short beep of capacitors charging on a supply rail is normal; a regulator running very hot indicates the system is drawing too much current. Retrieved 25 October 2026.
  3. ↑ 3.0 3.1 3.2 3.3 Battery Explosion, Capacitor or Corrosion Damage, on this wiki, sections "Assessing the damage" and "Checking the repair": measure each rail to ground and compare with a known-good board; power from a current-limited bench supply with the limit just above normal draw.
  4. ↑ Samuel M. Goldwasser, "Notes on the Troubleshooting and Repair of Small Switchmode Power Supplies", version 2.86, 28 January 2014, section "The series light bulb trick": a pulsating load means the supply is trying to come up but shutting down on an overcurrent or overvoltage condition. Retrieved 25 October 2026.
  5. ↑ 5.0 5.1 Sega Dreamcast Controller Port Fuse Replacement, on this wiki: with nothing plugged into the ports, a reading close to zero ohms from the port side of F1 to ground means a short remains and must be found before a new fuse is fitted.
  6. ↑ 6.0 6.1 6.2 Ray Carlsen, "RAM (Random Access Memory) problems in the C64/128 computer", updated 9 June 2011: shorted RAM produces a blank screen and often gets very hot; RAM normally runs barely warm; shorted RAM usually points to a failing power supply, which must be checked before the computer is turned on again. Retrieved 25 October 2026.
  7. ↑ 7.0 7.1 7.2 7.3 7.4 7.5 7.6 John F. Beckwith, "Current Tracer: A New Way to Find Low-Impedance Logic-Circuit Faults", Hewlett-Packard Journal, December 1976, pp. 2-8, PDF at Trinity College Dublin: solder bridges, shorted inputs and dead drivers on stuck nodes; cutting traces or lifting ICs is time-consuming, tedious and often damaging; locating Vcc-to-ground shorts by pulsing the disconnected supply terminal and following the greatest current. Retrieved 25 October 2026.
  8. ↑ Capacitor Failure Symptoms, on this wiki, section "Tantalum", citing Vishay: a solid tantalum fails short unless it has a fuse, and a shorted tantalum can take a regulator, a fuse or series resistors with it.
  9. ↑ 9.0 9.1 9.2 9.3 9.4 9.5 LarryD, "Locating short circuits on PCBs", Arduino Forum, Tutorials, 3 March 2019. Community source: about 0.3 V DC and 300 mA injected between the shorted traces, low enough not to damage components including input protection diodes or board traces; a DVM on the 200 mV range with one lead fixed; relative readings matter; branches with equal readings carry no current; the short found was an unetched copper filament between +5 V and the ground plane. Retrieved 25 October 2026.
  10. ↑ 10.0 10.1 Nichicon, Application Guidelines for Aluminum Electrolytic Capacitors, CAT.8101H; reverse voltage listed among causes of failure. Hosted as File:Nichicon Application Guidelines for Aluminium Electrolytic Capacitors.pdf; see Nichicon Application Guidelines for Aluminium Electrolytic Capacitors.
  11. ↑ 11.0 11.1 11.2 Techspray, "Using Freeze Spray to Diagnose Faulty Electronics": spray a powered board, moisture frosts it within seconds, and an area that melts faster than the rest marks the fault; freeze sprays are generally non-flammable; spraying generates static charge, and an anti-static version is sold for static-sensitive devices. Retrieved 25 October 2026.
  12. ↑ 12.00 12.01 12.02 12.03 12.04 12.05 12.06 12.07 12.08 12.09 EEVblog Electronics Community Forum, "How to find short?", Beginners board. Community source: look or feel for parts hotter than they should be with a thermal camera, IR thermometer, thermocouple or fingers; spray the board with isopropyl alcohol, power it and see where it evaporates fastest; follow the voltage gradient with 0.5 V and a current limit of 100 mA to 1 A; a 3½-digit DMM resolves 100 µV on its 200 mV range, giving 100 µΩ at 1 A; reference the meter near the supply's negative connection on the ground plane and look for the highest reading at the shorted part; measure across every powered component in the area. Retrieved 25 October 2026.
  13. ↑ 13.0 13.1 Recommended Tools, on this wiki: a bench power supply with current limit, and a thermal camera or IR thermometer for finding a shorted part on a dead board.
  14. ↑ Chemtronics, Freez-It technical data sheet, products ES1050 and ES1550, rev. G (8/18): tracking intermittent failures and shorts; non-flammable; cools to −60 °F (−51 °C). PDF copy. Retrieved 25 October 2026.
  15. ↑ Commodore 64 RAM Replacement, on this wiki, section "Commodore's signal checks", citing Commodore's C64 Diagnostic Instruction and Troubleshooting Manual: cut pin 8 (+5 V) of each RAM chip in turn to find a chip loading a multiplexer output.
Retrieved from "https://retrotechwiki.com/index.php?title=Finding_Short_Circuits&oldid=30987"