How to Test a Network Switch Power Supply Like an Admin Buyer: A Practical Guide Using a Multimeter

When You Might Need This (And I Really Did)

This guide is for people like me — administrators or junior IT buyers who suddenly find themselves needing to verify why a piece of network gear is acting flaky. Not engineers. Not bench techs. People who order the equipment and sometimes have to figure out why it's not working before calling in the cavalry.

Specifically: testing a capacitor in a power supply using nothing but a multimeter. I had to do this last year when a stack of Juniper EX2300 switches started behaving unpredictably. The symptom? Random reboots. The vendor shrugged. Our in-house engineer was on leave. So I grabbed a multimeter, and after 45 minutes of YouTube tutorials, I wrote this down so I'd never forget.

This is a 4-step process. It's not comprehensive electrical engineering. It's what an admin buyer can do to get a yes/no answer in 10 minutes.

Step 1: Set Your Multimeter Correctly

First: capacitors store energy even when the power is off. I learned this the hard way when I grabbed one that had been sitting for 20 minutes and got a jolt (not a big one, but a memorable one). So step zero is discharge the capacitor by shorting the leads with a screwdriver (metal shaft, plastic handle). Or wait longer. I now wait at least an hour for anything bigger than a standard ATX power supply.

Multimeter settings (I use a cheap $20 digital model from Amazon, works fine):

• Turn the dial to capacitance mode — usually marked with a '–|(–' symbol. If your meter doesn't have this (mine didn't), read the resistor and ESR method below.
• Plug the probes: black into COM, red into the port labeled for capacitance (often the same as the diode/ohm port).

Real talk: I assumed any multimeter would measure capacitance. Nope. Check before you start. (Note to self: buy a proper multimeter with capacitor ranges next time.)

Step 2: Read the Capacitor Value and Set Range

The capacitor will have a number printed on it, usually in microfarads (uF). Example: 1000uF, 16V. You need to see if the actual value matches (within a tolerance range). For most aluminum electrolytics (common in network switch power supplies), the tolerance is ±20%. So a 1000uF capacitor can read anywhere from 800 to 1200uF and still be considered fine.

Set the multimeter to a range above that value if it's a manual-ranging meter. If auto-ranging, just make sure the capacitor value is within its measurement capacity (usually up to 100mF which is 100,000uF, so fine).

Quick reference based on what I see in Juniper EX and QFX series power supplies:

• Small filter caps: 10–100uF
• Bulk storage caps: 470–4700uF
• If it reads less than 70% of rated value, it's degraded.

Step 3: Test the Capacitor (Two Methods)

Method 1: Capacitance mode (direct)

1. Discharge the capacitor fully.
2. Touch the probes to the leads (+ is usually longer or marked with a stripe). Polarity matters — red to +, black to -.
3. Wait 2–3 seconds for the reading to stabilize. It will start high and drop to the actual value.
4. Compare to rated value. If it's below 70% of rated or showing OL (overload), the capacitor is likely bad.

Method 2: No capacitance mode (resistance test)

If your meter can't measure capacitance directly: set it to resistance (ohm mode), at 2k or 20k range. Touch probes to capacitor leads (observe polarity). The reading should start low and climb steadily as the capacitor charges. If it stays at 0 or jumps to high immediately, the capacitor is shorted or open. Not ideal for value, but works as a quick sanity check.

I prefer capacitance mode when available. The surprise for me: the bad capacitor in my EX2300 read 37uF instead of 470uF. That was the culprit. Never expected a 93% drop in capacitance would still let the switch almost work. It would boot, run for 15 minutes, then crash.

Step 4: Interpret Results and Decide Next Steps

After testing, here's what the numbers mean in plain terms:

• Reading within 20% of rated value: Capacitor is fine. Look elsewhere (connectors, solder joints, voltage regulator IC).
• Reading lower by 30% or more: Degraded. It might still function but could fail under load. If you're already inside the power supply, replace it.
• Reading zero or OL in capacitance mode: Short or open. Definite failure. Replace immediately.
• Reading increases slowly in resistance mode: Normal. If it stays at 0 or jumps instantly, bad.

My rule: if the capacitor is more than 5 years old and the equipment is critical, I replace all electrolytics in the power supply section. It's cheap ($2–5 per capacitor) compared to a failed switch during business hours.

Common Mistakes I've Made (So You Don't Have To)

• Not discharging first. I got a spark. Not dangerous, but startling. Wait an hour for big caps or use a resistor discharge tool.
• Assuming all multimeters measure capacitance. Mine didn't. Had to improvise with resistance mode and a stopwatch. Not great.
• Mixing polarity on the probe. In capacitance mode, if you reverse polarity, the reading may be wildly off or show a negative value. If you see -473uF, flip the probes.
• Testing in-circuit without removing the power supply. The circuit's other components can affect the reading. For accurate results, desolder at least one lead of the capacitor. If you can't, the reading may be lower than actual.
• Ignoring the ESR value. Capacitors can have correct uF value but high Equivalent Series Resistance (ESR), which still causes failure under load. That's a separate test — a cheap ESR meter costs $20–30. I'm not a specialist, so when in doubt, I just replace them.

If you're managing Juniper switches (EX, QFX, or older MX) and see intermittent reboots, power supply capacitors are a frequent failure point. I've personally had this issue with 3 of 12 EX2300-C-12P units in a single rack in early 2024. After replacing the caps on two of them (the third was still under warranty), the problem disappeared. A lesson learned: never assume brand-new gear is always perfect.

The most frustrating part: these failures happen during retail peaks — exactly when you can't afford downtime. You'd think power supplies would have better QC, but they're often the cheapest component in the build. I now verify power supply testing as part of my pre-deployment checklist for any network gear above $1,000.