I've been commissioning and maintaining industrial power electronics for about eight years. In that time, I've personally destroyed — and documented — enough IGBT switches to fill a small parts bin. Now I keep a checklist for our team so nobody repeats my mistakes.
Last year, I replaced the 50N60 IGBTs in a 3-phase VFD converter eleven times. Eleven. Each time, the customer asked the same question: "Are you using cheap parts?"
They weren't cheap parts. They came from an authorized distributor. But I didn't have a good answer, so I just nodded, ordered more, and swapped them out. The worst part? The PLC counter in the control cabinet had been telling me the real story the whole time. I just didn't think to look at it.
That's the thing about IGBT switch failures in VFD converters. The symptom is obvious — blown switches, tripped breakers, smoke. The cause is almost never what you think it is.
If you're searching for "50N60 IGBT" right now, you're probably in the same spot I was. The drive trips. You open the panel and the IGBT has failed short, or the case is cracked, or you measure collector-to-emitter and read zero ohms.
The 50N60 is a 50-amp, 600-volt N-channel IGBT. It's a workhorse in small drives and converters, and it's also the part I blamed for months while it died over and over. First I blamed the components. Then the manufacturer. Then the environment — too hot, too dusty, utility spikes. I was wrong three times in a row.
Every failure followed the same ritual: order parts, wait for delivery, swap, test, hand the machine back. And every time, I convinced myself it would be the last time.
Here's what I finally discovered when I pulled the PLC counter data and actually read it instead of resetting it.
The drive had been restarting under load. Repeatedly.
An IGBT is a transistor, not a mechanical contactor. When you command it on before the DC bus has settled, you force it to operate in its linear region — partway between off and full saturation. The silicon die absorbs massive conduction losses and heats up. Then it heats up more. And eventually it fails short or cracks.
But my mistakes went deeper than that. There were three of them, and they added up.
I was driving the gate with 12V. Standard IGBTs specify 15V (±10%) to guarantee saturation. At 12V, the switch turns on partially. "Almost on" is where the heat comes from.
Between the high-side and low-side switches in each leg of the bridge, you need dead time. I had set 500 nanoseconds. The datasheet recommended 1 microsecond for the 50N60. I thought I knew better. I didn't.
The gate was being driven with essentially no deliberate series resistance. That makes switching edges extremely fast, which sounds great, but it creates voltage overshoot across collector-emitter. The peak exceeded the 600V rating on a regular basis. The IGBT degraded every cycle until it died.
A simple PLC was running the production sequence. Every cycle, it cut power to the drive inputs instead of sending a controlled stop command. The VFD fault log showed DC bus overvoltage trips. And my restart logic cycled the drive back on after 5 seconds — before the DC bus had discharged.
I could have seen all of this in the counter values and fault codes weeks earlier. But I was too busy ordering parts to read the logs.
Let me put real numbers on this, from our actual job costs.
Eleven IGBT replacements at roughly $6–12 each for a genuine 50N60 (distributor pricing as of early 2025) — call it $130 in parts. Small. But each swap took two technicians about two hours, including lockout, inspection, reinstall, and testing. That's 22 man-hours at a $45/hour burdened rate. About $1,000.
Then the real cost: downtime. The converter was feeding a production line with output valued at around $400/hour. Every failure meant 4–6 hours of stoppage, so each event cost $2,000+ in lost output. Eleven events took us past $20,000 — caused by a 5-second restart delay and a 12V gate supply. The parts themselves were never the expensive part.
And the trust? The customer brought in two outside vendors to audit my work. I can't blame them. I'd do the same in their position. I also had to look the same plant manager in the eye and say "it's fixed" eleven times. That gets old quickly.
In March 2023, I had three IGBT switch failures in one week. Three. That's when I finally sat down with the PLC counter, the VFD fault log, and the event timestamps side by side.
When I compared the counter data with the VFD's fault history, the pattern was unmistakable. Counter increment at 10:42:07, overvoltage fault at 10:42:09, restart command at 10:42:14. Seven seconds between fault and restart. The drive never had a chance.
The parts weren't the problem. My control logic was.
I also compared my settings to the drive's application manual. They recommended a minimum 30-second restart delay after a fault. I had set 5 seconds because I thought I was minimizing downtime. I wasn't. I was turning a $10 part into a $22,000 problem.
After we rebuilt the logic and started actually reviewing the counter data monthly, we caught 47 potential fault conditions across our other equipment in the next 18 months. None of them made it to the failure stage. That's what efficiency actually looks like — not faster restarts, but fewer failures.
If you're staring at a dead IGBT switch right now, here's what I'd check before ordering replacements:
I'm not saying every IGBT failure is caused by a design mistake. Counterfeit parts exist. Heatsinks get undersized. Dust and humidity kill more electronics than bad logic ever will. I've seen all of those.
But in my experience, repeated IGBT switch failures in 3-phase VFD converters are usually self-inflicted. We're in a hurry. We take shortcuts. We trust our gut instead of the data that's sitting right in front of us.
The PLC counter was running the whole time. It wasn't hiding anything. I just wasn't reading it.
Check the logs first. Not the parts order. It's a lot cheaper.