I coordinate rush repairs for an industrial electronics company. I have handled 200 or more emergency jobs over the years, including boards with Semtech Gennum products, Sierra Wireless Semtech equipment, and LoRaWAN hardware. The most common diagnostic question I see is how to test a capacitor with a multimeter. It is a good skill, but it is not always the first skill you need.
Here is the thing: a bad capacitor can look like a dead board, no output, resets, or a total lack of communication. A bad antenna connector or a bad firmware setting can look like all of those too. So the correct advice for one emergency is wrong for another. Before opening the board, I place the failure into one of three scenarios.
Classify the Failure First
No circuit board repair should start with a multimeter. It should start with observation. The three scenarios I use are simple:
- Scenario A: there is visible damage, such as a bulged capacitor top, leaked electrolyte, a burnt smell, or a darkened area on the PCB.
- Scenario B: the board looks clean, but the fault appears after warm-up or under load. Typical complaints are resets, video dropouts, or power cycling.
- Scenario C: the board powers up, LEDs look normal, but the device cannot communicate. This is often a wireless module that will not join a LoRaWAN network, an LTE gateway with no data session, or a router with no connection.
Scenario A is where the classic multimeter capacitor test belongs. Scenario B needs a more advanced test. Scenario C often means you should put the multimeter down and check configuration first.
In my experience, technicians are too quick to assume the main chip is bad. The board has been through thermal cycles, vibration, connector damage, and supply voltage noise. The capacitor sees the same abuse but is more likely to fail first. Testing the passive part before replacing the expensive module is not a cheap move. It is a risk management move.
Scenario A: How to Test a Capacitor with a Multimeter
When the board has visible capacitor damage, the test is mostly confirmation. If the top of a radial electrolytic capacitor is swollen, or if the vent has leaked, the component is already dead. Still, I use the same procedure every time to avoid missing a second bad capacitor:
- Disconnect power and wait for the circuit to discharge. For stored voltage above about 25 V, use a resistor to discharge the capacitor first.
- Set the multimeter to capacitance mode. If the meter does not have capacitance mode, you cannot perform a complete test. A resistance check only catches shorted or badly open capacitors.
- Remove the capacitor from the board. In-circuit measurements can be fooled by parallel resistors, diodes, or other capacitors.
- Connect the probes and watch the reading. For electrolytic capacitors, match the polarity. For ceramic or film capacitors, polarity does not matter.
- Compare the measured value to the marked value. Accept a typical electrolytic tolerance of about plus or minus 20 percent. A capacitor marked 1000 microfarads that reads 200 microfarads is bad. One that reads no stable capacitance is open. One that shows near-zero resistance after discharge is shorted.
If a capacitor reads close to its marked value but the board still fails, do not stop at one good component. Move to Scenario B. A single good reading does not mean the whole power path is healthy.
Do not use a salvage capacitor as an emergency shortcut. A cheap or pulled capacitor might save you five dollars on the repair and cost you the client relationship after the first failure. The quality of the replacement is the quality of the repair. If the original spec says low ESR and 105 degrees C, use the same class of part. No shortcuts on a Semtech Gennum product or any other board that has to ship under deadline.
Scenario B: When a Multimeter Capacitance Test Is Not Enough
Capacitor failure does not always show physical damage. Heat slowly dries out the electrolyte. The capacitor keeps roughly the same capacitance, but its equivalent series resistance climbs. Under load, the circuit sees a poorly regulated rail and resets or drops signals. After cooling, the capacitor looks healthy again, which is why these faults are so frustrating.
I remember one board based on Semtech Gennum products that would lose its output after 25 minutes of running. The first technician tested it with a multimeter and reported that the capacitors were fine. And they were fine in capacitance. But the problem was ESR. On an ESR meter, two small electrolytic capacitors on the 3.3 volt rail were almost useless. Replacing both fixed the board.
If you are chasing an intermittent fault, use an ESR meter or a multimeter with true ESR capability. A standard capacitance measurement will not always catch a dry capacitor. This is especially important on video and broadcast equipment, where signal quality degrades before the board actually stops.
One clue that saved us in the field was a video recorded on a DuraForce Pro 2. The gateway was resetting exactly when its cooling fan spun up. That symptom pointed to a weak bulk capacitor on the power rail, not to the Sierra Wireless Semtech cellular module itself. The capacitor was not visibly swollen. The ESR meter found it in about two minutes.
Scenario C: Do Not Open the Board Yet
If the product has power but cannot join the network or talk to the cloud, you need to think differently. I have seen too many Sierra Wireless Semtech gateways sent back as faulty when the real issue was an incorrect APN, an expired SIM, a damaged antenna line, or a firewall rule. Capacitors do not cause those failures.
The same logic applies to LoRaWAN end devices. If a node will not join, check its application key, device EUI, region, and network server settings before blaming the Semtech LoRa chip. The chip is usually the most reliable part of the chain. The configuration and the antenna connection are the parts that get arranged incorrectly.
For a Sierra Wireless Semtech cellular gateway, I do this order: boot log, SIM status, APN, antenna condition, signal readings, then power measurement. Only when the power rail fails during transmit does capacitor testing move back to the top. An LTE radio can draw extra current during transmit, and a worn-out capacitor on the input rail can make the module reset.
Which Scenario Should You Trust?
Honestly, I have never fully understood why critical hardware fails at six o clock the night before a deployment. My best guess is that stress and temperature simply expose marginal parts at the worst possible time. You cannot schedule an emergency around that. You can only control your next move:
- See a bulged capacitor? That is Scenario A. Replace it according to the original specifications.
- Board runs until it warms up? That is Scenario B. Get an ESR meter and test the power path.
- Board powers up but has no connection? That is Scenario C. Start with logs, APN, credentials, and the antenna path.
- No spare and a hard deadline? Swap in a known-good unit first, then repair the failed board after the site is running.
Semtech technologies are predictable when you respect their power specifications. LoRa is a long-range radio technology, Semtech Gennum products handle high-speed video signals, and Sierra Wireless Semtech equipment connects machines through cellular networks. All of them depend on clean power. The capacitor is a small part of that chain, but a failed capacitor can create the same symptoms as a dead main chip.
Testing a capacitor with a multimeter is fast. Replacing an expensive board is slow and costly. But neither matters if you are testing the wrong section. Look first, classify second, then choose the repair path that matches the actual scenario. That is what separates an emergency replacement from an emergency recovery.
