Test method for breakdown voltage of discrete components

Discrete Device Breakdown Voltage Testing: Methods That Actually Work

Breakdown voltage is the number that separates a working device from a destroyed one. Push a diode past its reverse breakdown and it either clamps cleanly or it burns. Push a MOSFET past its drain-source breakdown and the gate oxide shatters. Push a BJT past its collector-emitter breakdown and the junction punches through permanently.

This is not a parameter you can guess. You have to measure it, and you have to measure it right. The problem is that most engineers treat breakdown voltage testing like a simple voltage ramp. They crank up the supply, watch the current, and call it done. The number they get back is either wrong or dangerously optimistic.

Getting breakdown voltage right requires understanding what the test is actually doing to the device, how fast you ramp the voltage, how much current you allow, and what temperature the junction is sitting at when you measure it. Skip any of those variables and your data is just a guess with a unit attached.

Why Breakdown Voltage Is Not a Single Number

Everyone looks at the datasheet and sees a number. BVdss equals 400 volts. BVces equals 600 volts. BR equals 100 volts. Simple, right? Not even close.

That number on the datasheet was measured under specific conditions: a defined current limit, a defined voltage ramp rate, a defined junction temperature. Change any one of those conditions and the number moves. Sometimes by a few volts, sometimes by tens of volts. For devices with soft breakdown characteristics like Zener diodes, the shift can be enormous.

A Zener diode rated at 5.1 volts might read 5.3 volts at 1 milliamp and 4.8 volts at 100 milliamps. Both readings are correct. Neither one is wrong. They are just measured at different current levels, and the datasheet has to pick one reference point. If your test uses a different current, your number will not match the spec, and you will either reject good parts or accept bad ones.

This is why breakdown voltage testing is not about hitting a number. It is about replicating the exact conditions under which the spec was defined, then reading the result.

Core Breakdown Voltage Test Methods

DC Ramp Test: The Most Common and Most Misunderstood

The DC ramp test is what most labs use. You connect the device, set a current compliance limit, and slowly increase the reverse voltage until the current hits the compliance threshold. The voltage at that point is your breakdown voltage.

Sounds straightforward. It is not. The ramp rate changes everything.

If you ramp too fast, the junction heats up before the breakdown point is reached. The measured voltage will be higher than the true cold-junction breakdown because the heat reduces the critical electric field. You get an optimistic number that does not reflect real operating conditions.

If you ramp too slow, leakage current has time to accumulate and create localized heating. The breakdown point shifts downward, and you get a pessimistic number that rejects good devices.

The standard ramp rate for most discrete devices is 10 to 100 volts per second. For power MOSFETs and IGBTs, JEDEC JESD24-2 specifies a ramp rate of 10 volts per microsecond for pulsed tests, but for DC tests the rate should be slow enough to avoid self-heating, typically under 1 volt per second for high-voltage devices.

The current compliance limit matters just as much. For a diode, the standard test current might be 1 milliamp or 100 microamps. For a MOSFET, it is usually 250 microamps or 1 milliamp, depending on the voltage rating. If you set your compliance at 10 milliamps instead of 1 milliamp, you are measuring breakdown at a much higher current, and the voltage will be lower than the spec. Your device looks like it failed when it actually passed.

Pulse Breakdown Test: The Only Honest Method for Power Devices

DC testing gives you a number that is contaminated by self-heating. The only way to get a true cold-junction breakdown voltage is to use a pulse test.

The pulse breakdown test applies a fast voltage pulse, typically 100 microseconds to 1 millisecond wide, with a duty cycle under 1 percent. The pulse is fast enough that the junction does not heat up, but long enough to reach the breakdown point and give you a stable reading.

The standard setup uses a pulse generator with a series resistor to limit current. The resistor value is chosen so that if the device breaks down, the current stays below the damage threshold. For a 400-volt MOSFET, the series resistor might be 10 kilohms, which limits the breakdown current to 40 milliamps even if the device shorts completely.

You measure the voltage at the moment the current hits the compliance threshold. Because the pulse is short, the junction temperature has barely moved, so the number you read is the true avalanche breakdown voltage, not a thermally shifted value.

This method is mandated by JEDEC JESD24-2 for power MOSFETs and IGBTs, and by MIL-STD-750C for military-grade devices. If your test report does not specify pulse conditions, the data is not comparable to any datasheet that was measured with pulses.

Current-Ramped Breakdown: The Soft Breakdown Problem

Some devices do not have a sharp breakdown knee. Zener diodes below 5 volts, for example, have a gradual turn-on that looks more like a slope than a cliff. For these devices, a voltage ramp test gives you an ambiguous result. Where exactly did breakdown happen? At 4.8 volts? At 5.0 volts? The answer depends on where you draw the line.

The current-ramped method solves this. Instead of ramping voltage, you ramp current. You force a slowly increasing current through the device and measure the voltage across it. The voltage at a defined current level is your breakdown voltage. For a 5.1-volt Zener, the spec might define breakdown at 20 milliamps. You ramp the current to 20 milliamps and read the voltage. No ambiguity, no knee-hunting, no argument about where breakdown starts.

This method is standard for low-voltage Zener diodes and is codified in JEDEC JESD28-B. It gives you a repeatable number that does not depend on how sharp the breakdown knee is.

Test Setup Details That Make or Break Your Data

Voltage Ramp Rate Control: Slower Than You Think

The ramp rate is the single biggest source of error in breakdown testing, and most engineers do not control it properly.

A bench power supply with a manual knob does not give you a controlled ramp. The rate depends on how fast you turn the knob, which changes every time. Even a programmable supply can have step transitions that look like ramps but are actually staircases.

Use a supply with a programmable ramp function. Set the rate explicitly, typically 1 volt per second for devices under 100 volts and 0.1 volts per second for devices over 400 volts. Verify the actual ramp rate with an oscilloscope before you start testing. The supply display is not trustworthy.

For pulsed tests, the rise time of the pulse matters. A pulse with a 10 nanosecond rise time on a 400-volt device creates a dV/dt of 40 volts per nanosecond. That fast edge can trigger parasitic turn-on in MOSFETs through the Miller capacitance, which gives you a false breakdown reading. The pulse rise time should be controlled to 100 nanoseconds or slower for devices over 200 volts.

Current Limiting: Protect the Device and Your Data

The current compliance on your supply is not just a safety feature. It defines the test condition.

If you set the compliance too high, the device will burn before you get a reading. If you set it too low, you will never reach breakdown and the test will timeout. The compliance must match the datasheet test condition exactly.

For a MOSFET with BVdss of 400 volts tested at 250 microamps, set your compliance to 250 microamps plus a small margin, maybe 300 microamps. The margin protects you from noise spikes that could trigger a false fail, but it must be small enough that you are still measuring at the correct current.

Use a series resistor in addition to the supply compliance. The resistor provides a hard current limit that does not depend on the supply's response time. If the device breaks down hard, the resistor limits the current instantly, protecting the device from destructive failure. A 100 kilohm resistor in series with a 400-volt supply limits the current to 4 milliamps, which is safe for almost any discrete device.

Temperature Control: The Variable Everyone Forgets

Breakdown voltage has a temperature coefficient. For most silicon devices, BV increases with temperature at a rate of roughly 0.05 to 0.1 percent per degree Celsius. A device with a 400-volt breakdown at 25 degrees will read 420 volts at 125 degrees. That is a 20-volt shift, which is enough to move a device from pass to fail or vice versa.

If you test at room temperature but the device will operate at 100 degrees in the field, your room-temperature breakdown number is almost useless. You need to test at the temperature that matches the application, or at least correct the number using the temperature coefficient.

For qualification testing, run the breakdown test at three temperatures: minimum operating temperature, room temperature, and maximum operating temperature. Record the breakdown voltage at each point. The spread tells you how much the parameter drifts, and the worst-case number is what you use for design margins.

Common Mistakes That Ruin Breakdown Data

Testing with the device unplugged from the heatsink. The junction heats up during the test, and without a heatsink, the temperature rises uncontrolled. Your breakdown voltage drifts upward as the junction heats, and you get a number that is 10 to 20 volts higher than the true cold-junction value.

Forgetting to wait for leakage stabilization. When you first apply reverse voltage, the leakage current is high and drops over seconds as the junction charges deplete. If you read the voltage before the current stabilizes, you are measuring a transient, not a steady-state breakdown. Wait at least 5 seconds after applying voltage before recording the reading. For high-voltage devices, wait 10 to 30 seconds.

Using a two-wire measurement for high-voltage devices. The resistance of the test leads adds a voltage drop that you attribute to the device. For a 400-volt test with 100 milliohms of lead resistance and 1 milliamp of leakage current, the lead drop is 0.1 millivolts, which is negligible. But for a 10-volt Zener test with the same leads and 10 milliamps of current, the drop is 1 millivolt, which is 0.01 percent of the reading. Still small, but at 100 milliamps it becomes 10 millivolts, which is 0.1 percent. Use four-wire Kelvin connections for any breakdown test where accuracy matters.

How to Set Up a Proper Breakdown Test Sequence

Start with a leakage check at 80 percent of the rated breakdown voltage. Apply 80 percent of BV, hold for 10 seconds, and measure the leakage current. If the leakage is above the datasheet maximum at that voltage, the device has a defect and you should not proceed to full breakdown. This step catches weak devices before you destroy them.

Then ramp to breakdown at the specified rate. Use the programmed ramp function on your supply, not a manual adjustment. Monitor the current continuously. When the current hits the compliance threshold, record the voltage immediately. Do not keep ramping. The moment you hit the compliance, stop.

Hold the breakdown voltage for 1 second, then ramp back down to zero. This hold time lets you verify that the current is stable and not a transient spike. If the current drifts upward during the hold, the device is degrading and you should reject it.

Repeat the test three times on the same device. The breakdown voltage should be within 1 to 2 percent across all three runs. If the spread is larger than that, something is wrong with your test setup, not the device. Check your ramp rate, your current compliance, and your probe connections.

Run the same sequence at the high and low ends of the operating temperature range. The difference between the cold and hot breakdown voltages is your temperature coefficient. Use that number to derate the device in your design. A 400-volt MOSFET that gains 20 volts at high temperature needs to be derated to 380 volts in a 100-degree environment, not 400.