The question was on the wall.
Large font. Yellow background. Hard to miss.
"Is your test bench complete?"
That's an unusual opening for a trade show booth. Most booths tell you what they have. This one asked what you're missing.
Standing at the OMICRON Lab and Picotest joint booth at PCIM, that question started to make sense — not as marketing, but as a genuine engineering problem.
Three stations. Three different views of the same system.
The booth was organized around three live measurement setups, each addressing a distinct aspect of power integrity. They looked different. They were solving different problems. But they were all asking the same underlying question: does this power system actually behave the way you think it does?
Station 1 → Loop-Gain Stability Analysis
The Bode 100 — frequency range 1 Hz to 50 MHz — was connected to a live switching power supply via an injection transformer.
The screen showed the classic two-panel Bode plot: gain response on top, phase response below. Cursors marked at two frequencies, with values readable on screen — gain margin and phase margin extracted automatically by the Bode Analyzer Suite software.
This is the standard method for verifying closed-loop stability in a power supply. You inject a small perturbation signal into the feedback loop. You sweep across frequency. You measure how the system responds at each point.
The phase margin reading — the angular distance from the crossover frequency to the −180° line — tells you how much cushion the system has before it becomes unstable.
The physical setup used PCBite probes on the live board, positioned precisely at the injection and measurement points. The injection transformer handled the coupling, maintaining DC isolation between the analyzer and the switching circuit.
→ What you're measuring: the linearized small-signal behavior of the feedback loop → What you're inferring: whether the system will remain stable under normal operating conditions
Station 2 → PDN / Output Impedance
The Bode 500 — frequency range mHz to 450 MHz — was set up for power distribution network impedance measurement.
This is a fundamentally different measurement from loop gain. You're no longer looking at the control loop. You're looking at the passive power delivery network: the output capacitors, PCB traces, ferrite beads, and planes that carry current from the supply to the load.
The impedance curve on the monitor showed the characteristic shape of a real PDN: low impedance at DC and low frequencies, then a resonance peak, then high-frequency rolloff shaped by parasitic inductance. Each inflection point in that curve is a physical feature of the board — a capacitor's self-resonance, a decoupling network's cutoff, a trace inductance becoming visible.
The Bode 500's μΩ to MΩ impedance range means it can resolve the kind of ultra-low impedance that modern digital loads demand. A DDR5 memory bus or a high-current FPGA core rail requires sub-milliohm impedance across the relevant frequency band. Conventional instruments can't measure there accurately. The combination of Bode 500 hardware and Picotest's low-noise PDN probes (P2130A/P2131A series) is specifically optimized for this regime.
Station 3 → PSRR and Step Load Response
The third station combined two measurements that are often treated separately but are deeply related.
PSRR — Power Supply Rejection Ratio — measures how much noise on the supply rail gets through to the output of a downstream circuit. This matters enormously for analog front ends, ADCs, RF stages, and any circuit where the supply noise floor affects signal quality.
The setup used a Picotest J2120A signal injector to introduce a known disturbance onto the supply rail, with a Keysight InfiniiVision oscilloscope capturing the output response. The screen showed a characteristic PSRR curve — attenuation in dB versus frequency, typically good at low frequencies and degrading above the control loop bandwidth.
Next to it: step load response. A fast electronic load switching on and off, with the oscilloscope capturing the supply's voltage deviation and recovery time. The two measurements together give you a complete picture of how the supply behaves dynamically — both how it responds to disturbances from above (PSRR) and how it handles demands from below (load transients).
The Picotest ICONIC: what that enclosure was doing
One piece of hardware at this booth was easy to walk past without understanding.
The aluminum-chassis unit labeled ICONIC — built by Picotest — is a multi-channel analog simulation platform. It generates controllable, programmable power rail environments with realistic noise profiles, crosstalk, and transient behavior.
Connected to the Tektronix 6 Series MSO on the same bench, the live waveform told the story clearly: yellow channel showing the stimulated rail with structured switching noise, red channel showing the coupled response — a distinct ringing pattern, then damped oscillation, then settling.
This is not a signal generator pretending to be a power supply. It's a purpose-built environment for stress-testing how a circuit responds to realistic power rail conditions — the kind of conditions that exist in a real PCB with multiple switching regulators sharing a common ground plane, before you've actually built that PCB.
→ The measurement is not "does this supply work." → The measurement is "how does this load circuit behave when the supply is realistic."
Why this matters more than it looks
Power integrity has historically been treated as an afterthought — something you address when something goes wrong.
That approach worked when clock speeds were in the MHz range and supply rails had comfortable margins.
It stopped working reliably somewhere around DDR4, PCIe 4.0, and 48 V bus architectures for data centers.
Modern high-performance systems impose several conditions simultaneously that make power integrity a primary design constraint:
→ Load currents switch faster — di/dt from a modern FPGA or AI accelerator can exceed hundreds of amps per microsecond → Supply voltages are lower — a 0.8 V core rail has almost no headroom for transient deviation → Multiple rails interact — shared ground impedance means a GPU power event can affect an ADC rail on the same board → EMC requirements tighten — PDN resonances that were once irrelevant now appear as emission peaks in conducted noise tests
The Bode 100 and Bode 500, together with Picotest's probe and injector ecosystem, address this shift directly. They provide the frequency-domain visibility into power system behavior that oscilloscopes alone cannot give you.
An oscilloscope shows you what happened. A network analyzer shows you why it was going to happen.
The measurement philosophy behind the bench
There's a subtle but important distinction in how these instruments approach their measurements.
A Bode plot is a small-signal measurement. You inject a perturbation small enough that the system responds approximately linearly. The result — gain and phase versus frequency — is a model of the system in that regime.
It tells you: under these conditions, with this signal level, the system behaves this way.
It does not tell you: the system will never oscillate, under any condition, at any load.
This is not a limitation of the instrument. It's a property of measurement.
Every stability measurement is a conditional statement. The condition is: the assumptions of the measurement hold.
For most practical power supply designs, those assumptions hold well enough. Phase margin above 45°, gain margin above 10 dB — these are engineering criteria that work, in practice, for the vast majority of designs.
But understanding why they work, and where the limits of that assurance are, is what separates an engineer who uses the Bode plot as a checklist from one who understands what it's actually measuring.
The OMICRON Lab and Picotest booth was aimed at the second type of engineer.
What "Is your test bench complete?" actually means
By the time you've walked the full length of the booth — loop gain, output impedance, PSRR, step load response, ICONIC simulation — the question at the entrance starts to resolve.
Most engineers have an oscilloscope. Some have a spectrum analyzer. Fewer have a network analyzer configured for power supply measurements. Fewer still have a PDN probe capable of sub-milliohm resolution. Almost none have a structured method for testing PSRR alongside transient response alongside loop gain — on the same device, in the same session.
Each measurement answers a different question about the same system. None of them is sufficient alone.
→ Loop gain tells you about stability margins → Output impedance tells you about the PDN's frequency behavior → PSRR tells you about noise rejection → Step load tells you about dynamic response → ICONIC simulation tells you how your downstream circuit survives all of the above simultaneously
The booth wasn't selling instruments. It was demonstrating that a complete view of power integrity requires multiple measurement dimensions — and that most benches, even in serious development labs, are only looking at one or two of them.
That's what the question was really asking.
All photos: Thomas · @SignalByThomas