How to Test a Power Transistor with a Multimeter
As an Amazon Associate we earn from qualifying purchases.
How to test a power transistor with a multimeter uses the same diode-mode idea as a small transistor, with three twists: the metal tab or case is the collector, some power parts are Darlingtons with built-in resistors, and a meter’s milliamp-level test says little about amp-level duty. This guide covers bipolar power transistors on the bench. Mains circuits can kill, so it covers only loose parts and powered-down boards; panel and service work belongs to a licensed electrician, and the meter’s CAT rating must suit any circuit it touches.
Quick answer
- Safety first: unplug, discharge capacitors, and remove the part from the board and heat sink; never probe a power transistor on a live mains circuit.
- Take the pinout from the datasheet. On the two sheets we opened the tab or case is the collector, and the middle lead is not always the base.
- In diode mode the base-emitter and base-collector junctions should conduct one way only, and the collector-emitter pair should not conduct. A Darlington reads differently.
- A pass shows the junctions are intact at about a milliamp. It does not show ratings or behavior at amps.
Electrical safety before you start
Work on mains circuits can kill. The datasheets we opened describe their power transistors as general-purpose amplifier and switching parts, and equipment of that kind often plugs into a wall. Unplug it, and leave the power supply section to a qualified technician if you have to open it. Panel or service work belongs to a licensed electrician. Fluke’s diode-testing guide, which describes the diode mode used here, tells you to make certain all power to the circuit is off and notes that charged capacitors can leave voltage in the circuit.
Heat is the second hazard. The datasheets rate junction temperatures up to 150 °C for one part and 200 °C for another, so a transistor that was recently running can burn a finger. The meter’s CAT rating, which comes from the manual or listing for your meter, says which circuits it may be connected to. If neither states a CAT rating, do not assume the meter is safe for household mains.
How a power transistor differs from a small one
Three parts show the spread. The Philips 2N3906 sheet is a small-signal PNP. The onsemi TIP120 sheet, which we read from a copy hosted by CERN, describes a Darlington in a TO-220 package. The TT Electronics 2N3055 sheet describes a part in a hermetic TO-3 metal package. The ratios in the last column are our own division.
| Item | 2N3906 | TIP120 | 2N3055 |
|---|---|---|---|
| Package | TO-92 plastic | TO-220AB | TO-3 metal |
| Collector-emitter voltage | 40 V (listed as −40 V) | 60 V | 70 V |
| Continuous collector current | 200 mA | 5 A | 15 A |
| Dissipation at 25 °C ambient | 500 mW | 2.0 W | 6 W |
| Dissipation at 25 °C case | not listed | 65 W | 117 W |
| Current rating ÷ 2N3906 rating | 1 | 25 | 75 |
The TIP120 and 2N3055 carry 25 and 75 times the 2N3906’s rated current. The ratings in the table belong to these three sheets, so read the sheet for your exact part.
Know the pinout before you probe
The two power sheets order their leads differently. The TIP120 sheet lists four pin positions for its TO-220 outline: pin 1 base, pin 2 collector, pin 3 emitter and pin 4 collector. The extra collector position, pin 4, corresponds to the metal tab, which is our reading of the outline. The TT Electronics sheet for the 2N3055 lists pin 1 as base and pin 2 as emitter, with the case as the collector.
That has two consequences. First, the middle lead of a TIP120 is the collector, so the Wright State lab’s rule that the middle terminal is usually the base does not apply here. Kuphaldt’s textbook warns that the base is often not the middle lead and that only a meter check or the data sheet settles it. Second, touching a probe to the tab or case is the same as touching the collector lead, which matters when you read the pattern below.
Get the part ready
- Remove it from the circuit. Fluke says a resistance-mode reading taken with a diode in a circuit can produce a false reading and that it may be necessary to remove one end of the diode from the circuit. A transistor on a board has the same exposure to parallel paths.
- Take it off the heat sink. The case or tab is the collector, so a part bolted to a grounded sink without insulation ties its collector to the chassis. That is our inference from the sheets.
- Let it cool. Junction ratings of 150 °C and 200 °C are high enough to matter.
- Find the datasheet. You want the pin list and, for a Darlington, the schematic.
The diode-mode readings
Set the meter to diode mode. The University of Washington lab says the meter outputs a constant current of about 1 mA and shows the junction voltage; in reverse it cannot force 1 mA through, so the display climbs to the range limit, usually about 1.5 to 2.0 volts, or shows an over-range symbol. Kuphaldt’s textbook says a bipolar transistor registers as two diodes back to back, and that the emitter-base junction shows a slightly higher drop than the collector-base junction.
For a plain NPN power part such as the 2N3055, using the sheet’s pin order:
| Red probe | Black probe | Expected in diode mode |
|---|---|---|
| Base | Emitter | Forward voltage, with Fluke’s 0.5 to 0.8 V as the silicon range |
| Base | Collector (case) | Forward voltage, slightly lower than the emitter reading |
| Emitter | Base | OL |
| Collector (case) | Base | OL |
| Collector (case) | Emitter | OL |
| Emitter | Collector (case) | OL |
Kuphaldt says a PNP gives exactly opposite readings, with the black lead on the base, so a PNP power part flips the table. Kuphaldt’s textbook also treats any count other than two conducting pairs of the six combinations as a defective or non-bipolar part.
Darlingtons are different
The TIP120 sheet lists “Monolithic Construction with Built-In Base-Emitter Shunt Resistors” among its features and draws a Darlington circuit schematic with values of approximately 8.0 kΩ and 120 Ω. So a base-emitter reading on such a part is not a plain junction reading. The sheet also lists a higher base-emitter on-voltage limit than the 2N3055 sheet: 2.5 V at 3 A against 1.4 V at 4 A. Those are working conditions, not meter conditions, but they point the same way, and we expect a Darlington’s forward base-emitter reading to sit noticeably above the 0.5 to 0.8 V of a single junction.
That expectation is our inference. The pages we opened give no diode-mode number for a Darlington and do not say how the meter’s test current routes through the built-in resistors. Compare your readings with the schematic on the sheet for your exact part, and treat a collector-emitter reading that conducts in either direction as a fault signal, as for any transistor.
What a milliamp cannot show
The datasheet limits are set at test conditions a meter cannot reach. The table compares the sheets with the meter facts we have: about 1 mA of test current in diode mode, from the Washington lab, and a maximum open-circuit voltage of 3.2 V on the ohms ranges of the mas838 meter, used here as one example. The ratios are our own arithmetic.
| Quantity | TIP120 sheet | 2N3055 sheet | Meter |
|---|---|---|---|
| Rated continuous collector current | 5 A | 15 A | 1 mA in diode mode |
| Rating ÷ meter test current | 5,000 | 15,000 | 1 |
| Cutoff current test voltage (ICEO) | 30 V, limit 0.5 mA | 60 V, limit 1.0 mA | 3.2 V open circuit |
| Test voltage ÷ meter voltage | 9.4 | 18.8 | 1 |
| Saturation voltage limit | 2.0 V at 3 A | 0.75 V at 4 A | cannot be reproduced |
A transistor can pass the diode check and still leak at tens of volts or drop too much voltage at amps. The TIP120 sheet itself says that performance may not match the electrical characteristics under different conditions. The meter confirms two junctions and an open collector-emitter path at a milliamp, and nothing about the rest.
The ratings assume a heat sink
The 65 W and 117 W figures apply at a case temperature of 25 °C, and both sheets give a derating slope above that: 0.52 W per °C for the TIP120 and 0.67 W per °C for the 2N3055. At a 100 °C case, our own arithmetic gives 65 − 0.52 × 75 = 26 W for the TIP120 and 117 − 0.67 × 75 = 66.75 W, about 67 W, for the 2N3055. A multimeter check says nothing about whether a part keeps its case that cool in service. A mounting or cooling problem, rather than a dead junction, can therefore be the real cause of a misbehaving power transistor, and a diode-mode pass would not reveal it.
The gain socket on a power part
If your meter has an hFE socket, its range and bias matter. The mas838 manual lists a 0 to 1000 range at a base current of 10 μA and a collector-emitter voltage of 3 V. The TIP120 sheet lists a typical gain of 2500 at 4 A, so a healthy Darlington can exceed that range, which is our inference from the two documents. The 2N3055 sheet shows gain minimums falling with current: 40 at 0.5 A, 20 at 4 A and 5 at 10 A. The Utah lab likewise plots beta as a function of collector current.
Under the Auburn lab’s definition of beta as collector current over base current, a 10 μA base current means a gain reading of 100 corresponds to 1 mA of collector current, our multiplication. That is far below the amps in the data sheet’s conditions, so a socket reading on a power part is a rough sort at best. Wright State’s rule of thumb for small parts, that above 100 is normal, does not obviously transfer.
Related guides
- The basic diode-mode routine is shown on a small part in the NPN transistor test guide.
- A dedicated tester is covered in the MK-168 transistor tester roundup.
- Board-level checks continue in PCB board testing.
- Another switching part is covered in the relay test guide.
- For a rating primer, see what a 600V CAT III label means.
How we chose the steps and the sources
We kept a step only when a page opened on 2026-10-21 states it, and the safety steps needed at least two pages. The pages are two manufacturer data sheets (the onsemi TIP120 sheet from a CERN-hosted copy, and the TT Electronics 2N3055 sheet), a Philips small-signal data sheet at the University of Pennsylvania, an open textbook (Kuphaldt), four university lab pages (Washington, Utah, Wright State and Auburn), a manufacturer guide (Fluke) and one meter manual (mas838). The onsemi site refused our direct requests for the original files, so those pages do not count. We kept ten pages.
Ratings, pin lists and meter facts come from named pages. The ratios, the inference about tabs and heat sinks, and the Darlington expectation are our own and each is marked.
What we could not verify (what the sources do not say)
The pages we opened do not give a diode-mode number for any power transistor, so the expected readings are the single-junction figures from Fluke and the small-part examples. They do not describe a manufacturer-recommended test for power parts, or a loaded bench test with only a multimeter. They do not cover MOSFETs or IGBTs, which are separate parts with different structures. They also do not say how to remove a part from a heat sink safely, how hot is too hot to handle, or how a particular meter’s test current is routed in a Darlington. The two datasheets are for specific parts and makers, and other variants carry different ratings.
Frequently asked questions
Is the metal tab on a power transistor live?
On the two sheets we opened, the tab or case is a collector connection. On a powered board it can sit at the collector’s voltage, so treat it as live and keep your fingers off it.
Can I test a power transistor without removing it?
Parallel paths can fool the reading, and Fluke says it may be necessary to remove one end of a diode from the circuit to test it. Take the transistor out, or lift its leads, for a pattern you can trust.
Why does a Darlington read differently?
The TIP120 sheet lists built-in base-emitter shunt resistors, so the base-emitter path is not a plain junction. Compare your readings with the schematic on the sheet.
Does a passing diode check mean the transistor is good?
No. It shows two junctions and an open collector-emitter path at about a milliamp. Leakage at working voltage and gain at amps are untested.
Can I use the hFE socket on a power transistor?
You can, but the meter drives only a small base current, and a Darlington’s typical gain can exceed the socket’s 0 to 1000 range. Treat any result as a rough sort.
Bottom line
Switch everything off, take the part off the board and heat sink, use the datasheet pin order, and treat a diode-mode pass as proof of two intact junctions and nothing more.
Sources
- https://sy-dep-epc-lpc.web.cern.ch/components/datasheets/epc-lpc%20(converters)/TIP120%20TIP121%20TIP122%20TIP125%20TIP126%20TIP127-TO220%20BJT-On%20semi.pdf (opened 2026-10-21)
- https://www.ttelectronics.com/TTElectronics/media/ProductFiles/Datasheet/2N3055.pdf (opened 2026-10-21)
- https://workforce.libretexts.org/Bookshelves/Electronics_Technology/Electric_Circuits_III_-_Semiconductors_(Kuphaldt)/04%3A_Bipolar_Junction_Transistors/4.03%3A_Meter_Check_of_a_Transistor_(BJT) (opened 2026-10-21)
- https://www.fluke.com/en-us/learn/blog/digital-multimeters/how-to-test-diodes (opened 2026-10-21)
- https://staff.washington.edu/kd1uj/BEE332/Lab1/BEE%20332%20Lab%201%20BJT%20characterization.pdf (opened 2026-10-21)
- https://akizukidenshi.com/goodsaffix/mas838.pdf (opened 2026-10-21)
- https://www.engineering.upenn.edu/~belab/equipment/electrical/2n3906.pdf (opened 2026-10-21)
- https://my.ece.utah.edu/~ece2100/Lab_BJT_Intro_nodata_S03.pdf (opened 2026-10-21)
- https://cecs.wright.edu/~dkender/bme3512/BME3512SP17Lab4.pdf (opened 2026-10-21)
- https://www.eng.auburn.edu/~niuguof/2210labdev/html/transistor.html (opened 2026-10-21)
