Both were at PCIM. Both had SiC MOSFET demos running. Both had oscilloscopes, probes, switching circuits, and live waveforms.
From a distance, the setups looked similar.
They weren't.
The question each company was trying to answer
This is where the difference starts — not in the hardware, but in the question.
Tektronix and Keithley were asking: → What is this device, at the physics level?
Keysight was asking: → How does this device behave inside your circuit?
Both are legitimate questions. But they lead to completely different instruments, workflows, and results.
Tektronix + Keithley: going deeper into the device
The Tektronix/Keithley demo at PCIM was organized around two stations — static DC characterization and dynamic switching power characterization.
The static station was the more unusual of the two.
A Keithley 4200A-SCS semiconductor characterization platform. A 2657A High Power SourceMeter capable of 3,000 V and up to 50 A pulsed. A shielded test fixture. Current resolution down to 0.0017 nA.
The display behind the bench read:
"Quantify Trapped Charges on SiC MOS Devices — Using the Patent Pending Forced Current Quasistatic CV Method."
Trapped charges. Gate oxide physics. Sub-nanoamp leakage measurement.
This is not the language of circuit debugging. This is the language of materials science and device reliability.
The reason this matters for SiC specifically: the SiO₂/SiC interface has fundamentally worse quality than the SiO₂/Si interface in conventional silicon devices. Charge traps at that interface shift threshold voltage over time, degrade transconductance, and create long-term reliability degradation that standard production testing doesn't catch.
New JEDEC reliability standards for SiC devices are starting to require quantification of these effects. The Forced Current Quasistatic CV method gives you a number — not a qualitative impression, but an actual interface trap density you can track across stress cycles and aging tests.
The dynamic station extended the same philosophy into switching behavior.
Tektronix 6 Series MSO with IsoVu isolated probes — 160 dB CMRR, ±2,500 V differential range, ±60 kV common mode voltage. Dynamic Rds(ON) extraction without external clamping. A patented deskew algorithm aligning voltage and current channels at picosecond precision before computing switching energy.
The measured values from the live demo: M1 = 56.30 nJ turn-on energy, M2 = 862.1 nJ for the full switching cycle metric.
→ The emphasis throughout: absolute accuracy, traceable parameters, JEDEC-aligned methodology.
This is a system built for people who need to know what the device is.
Keysight: making the switching behavior visible in your system
The Keysight approach at PCIM was built around a different engineering moment.
Not characterization. Debugging.
The oscilloscopes on display — MXR and EXR series — were positioned as tools for engineers actively developing SiC and GaN power circuits. The headline message was bandwidth: high-bandwidth oscilloscopes with fast sample rates, capable of capturing the sub-nanosecond edges that modern SiC gate transitions produce.
The probe solutions on show were differential high-voltage probes optimized for floating measurements in half-bridge and full-bridge topologies. The emphasis: clean waveform capture at the switching node, with enough common-mode range to survive the harsh electrical environment of a real power stage.
The software tools visible were inline power analysis — real-time switching loss integration, efficiency measurement, harmonic analysis. The kind of workflow an engineer runs while iterating on gate driver design, PCB layout, or snubber networks.
Where Tektronix was asking "what is the trap density in this gate oxide," Keysight was asking "does this ringing on the drain stay below the voltage rating, and where is it coming from."
Both questions belong to the same development process. They belong to different phases of it.
→ The emphasis throughout: fast iteration, visual clarity, system-level insight.
This is a system built for people who need to know what the device does in their circuit.
The measurement gap between them
There is a specific technical boundary where these two philosophies diverge — and it's worth understanding precisely.
Common-mode rejection is one place.
IsoVu's 160 dB CMRR means the probe rejects common-mode interference by a factor of 100 million to 1. In a SiC switching circuit where dV/dt at the switching node can reach 50–100 V/ns, even a small common-mode coupling into a measurement channel can corrupt the waveform and produce false switching energy calculations.
Keysight's differential probes offer meaningful common-mode rejection — but not at that level. For system-level debugging where you're looking at waveform shape and relative timing, it's sufficient. For absolute energy calculations where a 1% measurement error matters, the gap is relevant.
Current resolution is another.
Keithley's 4200A-SCS resolves currents at the femtoampere level. Gate leakage currents in degraded SiC gate oxides can be in the picoampere to nanoampere range. Seeing that requires instrumentation that a general-purpose oscilloscope with a current probe simply cannot provide.
But that resolution comes at a cost. It's slow. It requires careful fixturing. It operates in a controlled environment, not in a live power stage.
→ High resolution and real-time capture are, to a significant degree, in tension with each other. → Each approach accepts one tradeoff in order to win at the other.
What this split reveals about SiC as a technology
SiC MOSFET development is at a particular stage where both measurement philosophies are simultaneously necessary — and that's not a permanent state.
In the early years of a new semiconductor technology, characterization and reliability testing dominate. You need to understand what the device is before you can trust it in a system. Gate oxide reliability, threshold voltage stability, dynamic Rds(ON) behavior under real switching conditions — these need to be quantified, standardized, and tracked across manufacturing lots.
That's the Tektronix/Keithley world.
As the technology matures and moves into volume production, the center of gravity shifts. Engineers are no longer asking whether SiC can be trusted — they're asking how to build a 30 kW inverter that meets efficiency targets, passes EMC, and survives 10 years of automotive thermal cycling. The device is a component. The system is the problem.
That's the Keysight world.
SiC is currently in both places at once.
Established device families like Wolfspeed Gen 4, Infineon CoolSiC, or onsemi EliteSiC are mature enough that system-level design work is the primary engineering challenge. Newer entrants, next-generation gate oxide technologies, and extreme-environment applications are still deep in characterization territory.
→ The two demo setups at PCIM weren't competing. → They were addressing different points on the same technology adoption curve.
A framework for choosing
If you're developing a power electronics product — an EV inverter, a data center power module, an industrial drive — and you need to decide where to invest in measurement capability, the choice depends on your position in the development cycle.
Early-stage device evaluation or reliability qualification: → You need static characterization, CV extraction, dynamic Rds(ON) under stress conditions, JEDEC-compatible test methods. → Tektronix + Keithley is the complete solution.
Active circuit design, gate driver optimization, layout iteration: → You need fast waveform capture, floating measurements at high common-mode voltage, real-time loss calculation. → Keysight's oscilloscope and probe ecosystem is optimized for this workflow.
High-volume production verification or incoming inspection: → Different tools again — automated test platforms, parametric testers, boundary scan. Neither PCIM demo was aimed at this phase.
The mistake is assuming one approach covers all three. No single instrument vendor does everything well across the full development chain — and no serious power electronics lab should expect it to.
What PCIM showed, in a single observation
Two booths. Two philosophies. The same SiC MOSFET sitting at the center of both.
One instrument ecosystem was asking: can we trust this device? The other was asking: can this device build a better system?
Both questions need an answer. The industry is still working on both simultaneously.
That's actually a sign of a technology that's growing up — not one that's already arrived.
All photos: Thomas · @SignalByThomas