The Capacitors That Weren't There
At the Texas Instruments booth at PCIM Europe in Nuremberg, one demo setup pulled me in for an unexpected reason.
Not the inverter topology. Not the efficiency number. The capacitors.
Looking at the evaluation board inside the acrylic enclosure, the left side of the PCB was dominated by a row of six electrolytic capacitors — Rubycon, 63V, 2700µF each. Substantial. Visually prominent. The kind of capacitor bank that any power electronics engineer recognizes immediately as bulk energy storage.
But the banner above the demo said: "Single-stage AC/DC converter for micro inverters." And the background panel behind the booth talked about a proprietary transformer design that "reduces" something "by nearly 90%."
The question worth asking: if this is a single-stage design meant to minimize size and component count, why the large capacitor bank — and what is TI's actual claim about reducing it?
The Problem With Traditional String Inverters: The Electrolytic Bottleneck
To understand what TI's single-stage micro inverter approach is addressing, start with the standard topology it replaces.
A conventional solar inverter — string or micro — takes DC from the panels and converts it to AC for grid injection. The standard approach uses two power conversion stages:
→ Stage 1 (DC-DC): A boost converter takes the variable solar panel voltage (which changes with irradiance, temperature, and MPPT tracking) and raises it to a stable intermediate DC bus, typically 350–400V
→ Stage 2 (DC-AC): An inverter stage converts the stable DC bus to grid-frequency AC
The intermediate DC bus requires bulk energy storage — typically large electrolytic capacitors — to buffer the power difference between the steady DC input and the pulsating AC output. The instantaneous power delivered to the grid varies at twice the grid frequency (100 Hz in a 50 Hz system), while the solar panels deliver roughly constant power. The electrolytic capacitor absorbs the difference.
These electrolytic capacitors are the lifespan bottleneck of the inverter. Aluminum electrolytic capacitors have a limited service life, particularly at elevated temperatures. They are also the largest and heaviest components in a micro inverter, limiting miniaturization. Reducing or eliminating them is one of the primary design challenges in micro inverter development.
The single-stage approach eliminates the intermediate DC bus entirely — connecting the solar panel directly to the AC conversion stage without a separate DC-DC boost stage. The panel voltage variation is handled within the single conversion stage itself, trading topological complexity for fewer components and potentially smaller passive energy storage requirements.
The TI Evaluation Board: TIDM-01035ME2
The evaluation board — labeled TIDM-01035ME2 — is a reference design for a single-stage AC/DC micro inverter operating at 230VAC output, 4A maximum. Several component choices visible in the photos tell the design story:
Rubycon 63V / 2700µF electrolytic capacitors (×6)
Six large electrolytics in a row on the left side of the board. At 63V operating voltage, these are on the DC input side — the solar panel interface. Their purpose is to filter the switching current ripple from the converter's input and to provide the energy storage that allows the MPPT algorithm to operate without being disturbed by the 100 Hz power pulsation of the AC output side.
The fact that these capacitors are present, prominent, and rated at 63V (a reasonable solar panel string voltage for a micro inverter) suggests the single-stage design hasn't eliminated the input capacitance requirement — but has potentially reduced it compared to the multi-stage equivalent, where capacitance would be required both at the intermediate high-voltage DC bus and at the panel input.
BOURNS transformer (yellow, center)
The transformer is the heart of the single-stage isolation topology. In a conventional flyback or LLC resonant converter, the transformer provides galvanic isolation between the DC side (solar panels) and the AC side (grid). The physical size of this transformer — and TI's claim that their proprietary design reduces transformer size "by nearly 90%" compared to conventional approaches — is directly related to the switching frequency. Higher switching frequency allows smaller magnetic components. GaN-based switching at hundreds of kilohertz rather than tens of kilohertz is what enables this reduction.
Wolfspeed power MOSFETs (right side)
The power switching devices visible on the right side of the board are Wolfspeed components — consistent with TI's claim of an "industry's first GaN intelligent power module" achieving "more than 99% efficiency." GaN devices enable the high switching frequency that reduces transformer size, and at the same time deliver lower switching losses than silicon MOSFETs at comparable current ratings.
The ITECH Solar Array Simulator: Creating a Controlled Sun
The DC input to the TI inverter demo was provided by an ITECH IT-N2121 Solar Array Simulator — not a real solar panel, but an instrument that replicates the I-V characteristic of one with programmable precision.
The screen during the demo was showing:
→ Output: 41.954V / 7.142A / 299.64W — the actual operating point
→ Curve: EN50530 — the European standard for PV inverter efficiency testing, which specifies the I-V curve shape used for MPPT testing
→ Vmp = 42.00V / Pmp = 300.00W — the maximum power point that the MPPT algorithm should be tracking
→ MPPT efficiency: 99.8% — the inverter's MPPT algorithm is tracking within 0.2% of the theoretical maximum power point
The colored curves on the ITECH display show both the I-V curve (current as a function of voltage, red) and the P-V curve (power as a function of voltage, green), with the operating point marker sitting precisely at the maximum power point intersection.
This is the measurement foundation of the demo: by using a calibrated solar simulator rather than actual panels, the test conditions are repeatable and comparable. The ITECH instrument's EN50530 curve mode ensures the inverter's MPPT performance is evaluated against the same standardized panel characteristic used for official efficiency certification testing.
The product card behind the ITECH unit also reveals another instrument in the setup: the IT7900P Regenerative Grid Simulator — a 3-in-1 instrument functioning as AC power supply, grid simulator, and RLC AC load. This would be the AC output side instrument, absorbing the grid-injected power from the inverter and providing the controlled 230V / 50Hz grid reference that the inverter synchronizes to.
The TI GUI: Live Monitoring From the Web Browser
The wall-mounted screen above the demo bench was running the TI PCIM GUI — a web-based monitoring interface hosted at dev.ti.com/gallery/view/5167043/PCIMCyclo/ver/1.0.0/ — displaying the inverter's operating data in real time during the demo.
The GUI layout reveals the complete measurement picture of the running system:
AC side (top left): VGrid(V) and IGrid(A)
The grid voltage and current waveforms are shown together. The voltage trace (larger amplitude) shows a clean sinusoidal grid voltage. The current trace shows the injected current — and the visible harmonic distortion in the current waveform is the engineering reality of a converter running in a live demo: the current waveform is not perfectly sinusoidal, showing some distortion that an engineer would quantify with THD measurement and compare against grid injection standards (EN 61000-3-2, IEC 61727).
DC side (bottom left): VDC(V) and IDC(A)
The DC input voltage from the ITECH solar simulator — approximately 42V, stable — and the DC current tracking the MPPT algorithm's operating point.
Grid measurement box (top right):
→ Grid Voltage (RMS): 234.3V — the actual grid voltage at the measurement point
→ Grid Current (RMS): 1.2A — the injected AC current magnitude
→ Grid Frequency (Hz): 49.5 Hz — slightly below nominal, consistent with a real grid measurement
DC measurement box (bottom right):
→ DC Voltage: 42.1V / DC Current: 7.2A — cross-checking with the ITECH display confirms consistent readings
Temperature monitoring: Multiple temperature channels visible in the Temperatures graph — semiconductor junction temperatures and heatsink temperatures, the thermal performance parameters that determine whether the efficiency claims are sustainable at rated power over long operating periods.
The web-based GUI interface is a TI-specific design choice: the evaluation firmware exposes the converter's internal measurements through a GUI Composer-based web application, accessible from any browser on the same network. This lowers the barrier to evaluating the reference design — engineers can interact with the running hardware without needing proprietary software or specialized hardware tools.
What "Single-Stage" Actually Means for Micro Inverter Design
The architectural distinction between single-stage and two-stage conversion is more consequential than it might appear from the component count alone.
In a two-stage design, each stage can be individually optimized: the DC-DC boost converter optimizes MPPT tracking and voltage regulation; the DC-AC inverter stage optimizes grid synchronization and current quality. The intermediate DC bus provides a buffer that decouples the two optimization problems.
In a single-stage design, both problems must be solved simultaneously in one conversion stage. The same switching network handles MPPT tracking, galvanic isolation, voltage transformation, and grid-synchronous current injection simultaneously. The control algorithm is correspondingly more complex — which is precisely what the PI controllers visible in the GUI control panel are managing: Enable PI Control / Enable Duty Cycle Comp. / Enable PI Controllers.
The trade-off TI is proposing: more complex control in exchange for:
→ One fewer power conversion stage → fewer switching losses → higher system efficiency
→ No high-voltage intermediate DC bus → smaller, lower-voltage capacitors → longer service life
→ Smaller transformer (via GaN high-frequency switching) → smaller overall footprint
→ Fewer total components → lower BOM cost and higher reliability
For a micro inverter — a unit that attaches to a single solar panel, operates outdoors for 20+ years with minimal maintenance access, and must minimize both cost and volume — these trade-offs point strongly toward single-stage architecture as the direction of development.
The Capacitors, Revisited
Coming back to the opening observation: the six large Rubycon electrolytics on the TI evaluation board are not evidence against the single-stage thesis. They are the input-side energy buffer, necessary to allow the MPPT algorithm to operate cleanly.
The reduction TI is claiming — "nearly 90% reduction" — refers to the transformer size, not the input capacitance. By switching at GaN frequencies rather than silicon MOSFET frequencies, the magnetic core volume required for the transformer decreases dramatically. The transformer visible in the demo — yellow BOURNS core, modest physical size — is the physical expression of that claim.
What the single-stage architecture does eliminate is the intermediate bus capacitance: the large electrolytic bank that would traditionally sit between the DC-DC and DC-AC stages at 350–400V. Removing that capacitor — the highest-voltage, highest-stress, shortest-lifetime component in a conventional two-stage inverter — is the design decision that most directly extends the micro inverter's service life.
The six Rubycon units at 63V remain. The 400V intermediate bus bank is gone. For a device designed to last 25 years on a rooftop, that distinction matters.
Demo observed: Texas Instruments TIDM-01035ME2 single-stage AC/DC micro inverter evaluation board · GaN power devices · BOURNS transformer · 230VAC / 4A · ITECH IT-N2121 Solar Array Simulator (EN50530, 42V / 300W) · ITECH IT7900P Regenerative Grid Simulator · TI PCIM GUI real-time monitoring interface
All photos: Thomas · @SignalByThomas