PaperlayerView exact revision on arXiv

Source-linked AI summary

Modularity is Not Enough: Demonstration of a Solderless 400 V DC, 2.5 kW Three-Phase Inverter

Luc Imperiali, Aaron Griesser, Jonas Huber

arXiv:2609.28244v1eess.SYcs.SY

TL;DR

Soldered joints hinder complete disassembly, repair, and component reuse, while solderless power converters at several hundred volts and kilowatt power remained unexplored. The authors mount converter components using pressure contacts formed by screw-clamped resin molds and rubber compression pads instead of solder. The kilowatt-class demonstrator maintained stable effective on-state resistances over 120 thermal cycles; vibration sweeps from 5 Hz to 2 kHz exceeding 10 g in all three axes left the powered PCB intact without contact degradation.

  • Problem

    Soldered joints hinder complete disassembly, repair, and component reuse, while solderless power converters at several hundred volts and kilowatt power remained unexplored.

    Source: §I · p. 1

  • Method

    The authors mount converter components using pressure contacts formed by screw-clamped resin molds and rubber compression pads instead of solder.

    Source: §I · p. 1, A. Solderless Interconnection Concept · p. 2

  • Results

    The kilowatt-class demonstrator maintained stable effective on-state resistances over 120 thermal cycles; vibration sweeps from 5 Hz to 2 kHz exceeding 10 g in all three axes left the powered PCB intact without contact degradation.

    Source: §IV · p. 7

  • Takeaways & Limitations

    The removable solderless stack-up supports non-destructive component exchange and end-of-life disassembly, and the authors report feasibility within the tested operating range.

    Source: §I · p. 2, §IV · p. 8

    The life-cycle comparison covers manufacturing-phase embodied carbon only, not other impact categories or use-phase repair, reuse, and recovery scenarios.

    Source: §IV · p. 7

Abstract

from arXiv · show

This paper presents the design and experimental evaluation of a fully solderless realization of a 400 V, 2.5 kW GaN-based variable speed drive (VSD), using screw-clamped resin molds and rubber compression pads instead of soldered interconnections. The power stage uses 650 V GaN power transistors and is operated at a switching frequency of 200 kHz. The solderless demonstrator is compared to a soldered reference realization using an identical printed circuit board (PCB). Over 120 thermal cycles with heatsink temperatures up to 90 {\deg}C, the solderless contacts show no degradation in effective on-state resistances (including contact resistances). Separately, open-loop vibration sweeps from 5 Hz to 2 kHz with acceleration amplitudes above 10 g were performed on the solderless assembly and left the continuously powered demonstrator electrically intact; subsequent resistance and nominal-power checks likewise indicate no contact degradation. An initial life-cycle assessment (LCA) indicated a higher embodied carbon footprint for the solderless realization due to 3D-printed resin molds, whereas a prospectively evaluated injection-molding scenario reduces the carbon footprint to near that of the soldered reference. The solderless assembly furthermore enables non-destructive component replacement, as demonstrated after a power transistor failure, as well as component re-use. These results support the feasibility of repair-oriented, industrially relevant kilowatt-class solderless power converters.

I. Introduction

Circular-economy goals motivate repairable and reusable power converters, but soldered interconnections hinder disassembly and component recovery. The paper demonstrates a solderless converter at 400 V and 2.5 kW, extending prior solderless concepts beyond low-voltage signal electronics.

I. Introduction

  • Soldered interconnections hinder complete disassembly, repair, and component reuse, making circular-economy-compatible hardware more difficult.Solder joints generally require destructive processes during recycling.Source: §I · p. 1
  • Earlier solderless recovery approaches had been demonstrated only for low-voltage signal electronics and non-production environments.Conductive adhesives remain permanent, dissolvable PCBs are incompatible with repair, and liquid-metal mounting appears best suited to fewer high-value components –.Source: §I · p. 1
  • The paper demonstrates a fully solderless 400 V, 2.5 kW GaN-based VSD using screw-clamped resin molds and rubber compression pads.The authors describe it as the first experimental demonstration of a fully solderless power converter at relevant power and voltage levels.Source: §I · p. 1
  • The solderless stack-up enables non-destructive component exchange and straightforward end-of-life disassembly.These are described as prerequisites for repairable and recyclable converter hardware.Source: §I · p. 2

II. Design of a Solderless VSD Demonstrator

The section introduces the solderless interconnection idea, outlines the design workflow, and presents the demonstrator and its specifications.

II. Design of a Solderless VSD Demonstrator

  • The section proceeds from the solderless interconnection idea to the design workflow and realized demonstrator, with specifications summarized in Table I.The passage provides an organizational overview rather than design or evaluation results.Source: §II · p. 2

A. Solderless Interconnection Concept

Components are pressed onto PCB pads without solder by screw-clamping a PCB over components housed in resin-mold cavities with rubber compression pads. The compliant pads maintain contact force while separate thermal coupling connects the power transistors to a heatsink.

A. Solderless Interconnection Concept

  • Screw-clamped resin molds and rubber pads press component terminals onto PCB pads to establish solderless electrical connections.Components sit in dedicated mold cavities, with the PCB placed on top and fastened down.Source: A. Solderless Interconnection Concept · p. 2, Figure 2 · p. 2
  • Approximately 0.2 mm pad compression accommodates component-height and terminal-geometry tolerances while maintaining contact pressure.The rubber pads are nominally about 0.5 mm thick and provide compliance and a spring effect.Source: A. Solderless Interconnection Concept · p. 2
  • For top-cooled GaN transistors, the mold presses on the leads while terminal pads transfer clamping force to the PCB pads.A TIM and a separately fastened heatsink provide thermal coupling without loading the electrical contacts through the mold.Source: A. Solderless Interconnection Concept · pp. 2–3
  • Reverse-side test pads enable checking component seating and clamping force before energizing the converter.The checks are performed adjacent to each component pad after assembly.Source: A. Solderless Interconnection Concept · p. 3
  • Very flat passives may rest on soldermask instead of contacting PCB pads, so their footprints must remove soldermask beneath the component bodies.This issue was observed in early prototype PCBs.Source: A. Solderless Interconnection Concept · p. 3

B. Design Workflow

The workflow combines electrical and mechanical design, iterating PCB and mold geometry before assembly and verification. The demonstrator board has three phase legs, but only Phase C was assembled for the prototype.

B. Design Workflow

  • The first workflow stage defines requirements and topology, then selects components from an optimization Pareto front according to efficiency and power-density targets.The authors note that other design methods could also be used.Source: B. Design Workflow · p. 3, Figure 3 · p. 3
  • The PCB-design stage develops schematics, selects power-stage and auxiliary components, defines 3D component bodies, adjusts some soldermask footprints, and routes the board.The 3D bodies are exported to support the mechanical design stage.Source: B. Design Workflow · p. 3, Figure 3 · p. 3
  • PCB layout and mold geometry are iterated until pad dimensions, mold envelopes, screw locations, and clearances meet electrical and mechanical constraints.Component bodies from PCB design are used to generate mold cavities, and rubber-pad dimensions are derived from the resulting geometry.Source: B. Design Workflow · pp. 3–4, Figure 3 · p. 3
  • The workflow fabricates the bare PCB, 3D-prints molds, assembles the components, and verifies contacts, output waveforms, thermal cycling, and vibration response.The stages follow the electrical and mechanical design iterations.Source: Figure 3 · p. 3
  • Only Phase C was assembled on the first prototype, while Phases A and B show the bare PCB pad layout.The board has three phase legs and shared control and auxiliary circuitry along its upper edge.Source: Figure 4 · p. 4

C. Realization

The demonstrator integrates three identical GaN phase legs, with separate resin molds for functional units; the molds and compression pads clamp components against the PCB.

C. Realization

  • The demonstrator integrates three identical phase legs based on Infineon IGLT65R045D2 GaN transistors, although only one phase leg was assembled for demonstration.Each phase leg includes a half-bridge, gate drivers, output capacitors, current measurement, heatsink mounting holes, and resin molds.Source: C. Realization · p. 4
  • The prototype uses separate molds for each functional unit, simplifying debugging and mechanical iteration but increasing initial assembly effort.Consolidating molds in future designs could reduce manufacturing complexity while retaining selective repairability.Source: C. Realization · p. 4
  • The assembly uses 3D-printed resin molds and laser-cut rubber pads, with components placed in mold cavities before the PCB is screwed down to terminal pressure.Contact integrity was checked on reverse-side test pads before installing the thermal interface material and heatsink.Source: C. Realization · p. 4
  • Fig. 5 compares measured soldered-reference and solderless-demonstrator waveforms at nominal single-phase power of approximately 830 W.It shows filtered load voltage and inductor current over four output cycles, plus switching-interval waveforms at 5 µs/div.Source: Figure 5 · p. 4

III. Results

Characterization compared the solderless demonstrator with a conventional soldered reference board of identical layout, using one phase leg at 400 V dc-link voltage and 200 kHz.

III. Results

  • The solderless demonstrator was benchmarked against a conventional soldered reference board with an identical layout, apart from minor soldermask adjustments.The characterization covered electrical performance, thermal cycling, vibration, and embodied carbon footprint.Source: §III · p. 4
  • Electrical characterization used a single phase leg at 400 V dc-link voltage and a 200 kHz switching frequency.Nominal-power tests used an external 100 µH output inductor and adjustable load resistor, yielding approximately 830 W per phase and 7.3 A RMS phase current.Source: §III · p. 4
  • The reported evaluation spans output-waveform quality, on-state resistance, thermal cycling, vibration robustness, and embodied carbon footprint.Source: §III · p. 4

A. Electrical Performance

The solderless assembly retained switching and loaded output behavior comparable to the soldered reference, including at nominal single-phase power.

A. Electrical Performance

  • Both assemblies showed comparable switch-node voltage slopes exceeding 100 V/ns in zero-current-switching tests up to 400 V.The reported tests indicate that solderless contacts do not impair the switching transition.Source: A. Electrical Performance · p. 4
  • At nominal single-phase power, both realizations produced the same sinusoidal load voltage and inductor current, with effectively indistinguishable switching-interval waveforms.These results indicate correct inverter operation without solderless interconnections preventing nominal-power operation.Source: A. Electrical Performance · p. 4

B. Thermal Cycling Tests

Effective on-state resistance, including PCB contact resistance, showed no relevant systematic change during thermal cycling of the solderless assembly through 120 cycles, including elevated temperature swings.

B. Thermal Cycling Tests

  • Fig. 6 records high-side and low-side heatsink temperatures alongside ambient temperature during the baseline and elevated-swing cycling campaigns.Cycles 1–100 used approximately 20 °C swings; cycles 101–110 and 111–120 used approximately 45 °C and 50 °C swings, respectively.Source: Figure 6 · p. 5
  • Both converters underwent 100 loaded power cycles with approximately 20 °C heatsink temperature swings, followed by two ten-cycle, higher-temperature campaigns on the solderless board.Resistance was measured after every 20 cycles and again at cycles 110 and 120; the measured value included intrinsic device and PCB contact resistance.Source: B. Thermal Cycling Tests · p. 5
  • Fig. 7 plots temperature-compensated effective on-state resistance, including PCB contact resistance, against cycle count for both assemblies.The solderless board alone extends through cycle 120, following extra cycles with approximately 45 °C and 50 °C swings.Source: Figure 7 · p. 5, B. Thermal Cycling Tests · p. 5
  • The solderless switches began at approximately 36 mΩ effective on-state resistance versus roughly 35.5 mΩ for the soldered reference, with no systematic penalty inferred.At 7.3 A RMS phase current, the 0.5 mΩ difference corresponds to about 26 mW additional conduction losses per switch relative to 830 W output power.Source: B. Thermal Cycling Tests · p. 5
  • No relevant systematic change in solderless effective on-state resistance was observed over 120 thermal cycles, including additional cycles with temperature swings up to approximately 50 °C.The elevated-stress campaigns heated the heatsinks to 80 °C and 90 °C; fluctuations were similar to the soldered baseline and mostly within or close to instrument accuracy.Source: B. Thermal Cycling Tests · pp. 5–6

C. Vibration Tests

The powered solderless demonstrator remained electrically intact during open-loop vibration sweeps, and post-test resistance and nominal-power checks showed no contact degradation. The test covered 5 Hz to 2 kHz and targeted accelerations above 10 g along three axes.

C. Vibration Tests

  • After screws were reinstalled with thread-locking adhesive and the heatsink was removed, the reassembled converter operated correctly on its first attempt.The campaign assessed the solderless assembly after complete disassembly and remounting of the molds and components.Source: C. Vibration Tests · p. 6
  • The continuously powered board operated without switch-node interruption during open-loop sweeps from 5 Hz to 2 kHz targeting accelerations above 10 g along three axes.For reduced risk, the vibration setup used a 12 V dc-link rather than the nominal 400 V.Source: C. Vibration Tests · p. 6
  • With the board mounted horizontally, Z-axis vibration showed resonant peaks well above 10 g, while X- and Y-axis responses stayed comparatively flat.The measured Z-axis peaks occurred near 170 Hz, 800–1000 Hz, and around 1.45 kHz.Source: C. Vibration Tests · p. 6, Figure 9 · p. 6
  • Post-vibration resistance measurements remained within the observed ranges, and nominal-power operation and a further resistance check showed no contact degradation.The subsequent loaded check used 400 V dc and approximately 830 W single-phase output power.Source: C. Vibration Tests · p. 6

D. Embodied Carbon Footprint

The life-cycle comparison estimates manufacturing embodied carbon for a soldered reference, the resin-printed solderless demonstrator, and an injection-molded solderless scenario. The prototype’s mold fabrication raises its footprint, while the modeled injection-molding scenario brings it close to the reference.

D. Embodied Carbon Footprint

  • The functional unit is the complete three-phase converter PCB, including the power stage and auxiliary circuitry.The inventory uses component-level estimates and ecoinvent data for items such as resin 3D printing.Source: D. Embodied Carbon Footprint · pp. 6–7
  • The solderless inventory uses a 95.4 cm2 PCB footprint, about twice the 43.1 cm2 volume-optimized soldered reference footprint.The prototype footprint includes resin-mold areas and reflects separate molds per component group for debugging and mechanical iteration.Source: D. Embodied Carbon Footprint · p. 7
  • The solderless demonstrator inventory includes resin molds, screws, and rubber pads, while the future scenario models the same mold mass made by injection molding from ABS.The resin molds are modeled from measured mass and a resin-printing inventory; the rubber pads include laser cutting.Source: D. Embodied Carbon Footprint · p. 7
  • Approximately 9.2 kg CO2eq for the resin-printed solderless realization compares with approximately 6.3 kg CO2eq for the soldered reference and approximately 6.5 kg CO2eq for the injection-molded scenario.The resin-printed molds add approximately 2.9 kg CO2eq; electronic components dominate all three scenarios.Source: D. Embodied Carbon Footprint · p. 7, Figure 10 · p. 7

IV. Conclusion and Outlook

The work demonstrates a solderless 400 V, 2.5 kW GaN drive with stable electrical contacts under the tested thermal and vibration conditions. It also reports repair and disassembly advantages, while identifying further reliability, manufacturing, and life-cycle evaluation needs.

IV. Conclusion and Outlook

  • The solderless assembly showed comparable switch-node waveforms and stable effective on-state resistance over 120 thermal cycles, while vibration testing left the powered board intact with no subsequent contact degradation.Vibration sweeps covered 5 Hz to 2 kHz with accelerations exceeding 10 g in all three axes.Source: §IV · p. 7
  • A semiconductor that overheated above 100 °C was replaced quickly, illustrating non-destructive repair; injection-molded housings also reduce the prototype’s carbon-footprint penalty toward the soldered reference.The resin 3D-printed prototype has a clearly higher embodied footprint than the soldered reference, while the injection-molding scenario diminishes that drawback.Source: §IV · p. 7
  • Mechanical design remains open, including mold-size optimization, co-optimization of screw locations with electrical layout and contact pressure, and alternatives to screws.The prototype still requires substantial manual assembly, whose effort and automation opportunities remain to be quantified.Source: §IV · p. 7
  • Reliability qualification should extend beyond the initial thermal-cycling and open-loop vibration tests to humidity, shock, drop, and pressure-contact corrosion studies.Selective terminal plating is proposed as a possible stability improvement, but its additional environmental impacts also require assessment.Source: §IV · p. 7
  • The life-cycle comparison covers manufacturing-phase embodied CO2eq, not other impact categories or use-phase scenarios such as repair loops, reuse, and end-of-life recovery.Further assessment is proposed for resource depletion, toxicity, water use, mold and component reuse, and material recovery.Source: §IV · p. 7
  • The conclusion frames the 400 V, 2.5 kW GaN demonstrator as extending reversible interconnection without compromising tested electrical, thermal, or vibration performance.The stated route toward repairable and recyclable converters depends on adapting mold fabrication beyond prototyping resin printing.Source: §IV · pp. 7–8
Loading 2609.28244v1…