Initial 400 A Regen Test — Gen7 Size 2

This report extends the Gen7 size 2 regenerative staircase of OV-TEST-HW-REGEN-200A-GEN7-SIZE2 from 200 A to a 400 A q-axis command. The limiting facility for high-current regen is the DC bus: the Sorensen bench supply can sink only about 50 A back into the grid, so above roughly 150 A of regen the bus voltage runs away. For this run a Chroma DC electronic load bank was connected across the DC link in constant-voltage mode at 150 V to absorb everything the supply cannot sink. The inverter was commanded in steps from 25 A to 400 A and held at 400 A twice, returning up to 8.8 kW to the DC link.

During the 400 A holds, narrow repetitive spikes were observed on the oscilloscope phase-current traces that were not present at lower currents. Telemetry shows clean control, no faults, and a solid 150 V clamp, so the anomaly is visible only in the analog captures and its cause is undetermined. The test was stopped for this reason. Hardware passed every target; the spike observation is resolved below — it was the oscilloscope current clamps saturating, not the inverter.

Test setup

• DUT: OpenVVVF Gen7 size 2 inverter hardware, firmware foc_demo graph (hash 855dbcee90d611cb, includes slew-limited IqVar, burst current sampling, and a modulator voltage-limit node).
• Machine: Induction machine coupled to a Sierra CP Engineering dynamometer; approximately −1,470 RPM mechanical (−7,360 RPM electrical) during the holds.
• DC supply: Sorensen bench supply (regenerative sink limited to ~50 A) in parallel with a Chroma DC electronic load bank in constant-voltage mode at 150 V across the DC link, clamping the bus.
• Instrumentation: RTE Studio telemetry, Tektronix 5 Series MSO, Chroma load front panel.
• Temperature channels: baseplate NTCs TS1 (temp_inv1_c) and TS2 (temp_inv2_c); TS3 (temp_inv3_c) was not enabled for this session due to a wiring defect unrelated to the test.

Test conditions

Procedure

1. Enabled the inverter and allowed the dynamometer to bring the coupled machine to operating speed.
2. Jogged IqVar at 10 A and 26 A to confirm orientation, then stepped 25 → 200 A in 25 A steps, then 250/300/350/375/400 A.
3. Observed the bus climb from ~74 V as regen power exceeded the supply's sink capability, with the Chroma load bank absorbing the excess and clamping the bus at 150 V from the 375–400 A region onward.
4. Held 400 A for 14.8 s (first hold). Narrow spikes were observed on the oscilloscope phase-current traces that had not been present at lower currents. Stopped the run (IqVar 0) to investigate; because IqVar is slew-limited, current decayed to under 20 A in ~7.7 s.
5. Resumed with a second 400 A hold (39.0 s total including the ~8 s slew-limited ramp). The spikes were still present. Stopped again and let the inverter cool while logging continued to t = 340 s.

Electrical results

Per-command statistics from the full-rate telemetry log (Pdc = DC-link power, Idc = DC-link current):

The bus voltage told the clamp story directly: it stayed at 74.3–74.8 V through the 150 A step, began climbing at 175 A (74.3–82.3 V), and rose stepwise to the Chroma clamp (142.5–150.4 V at the 375 A step), then held 149.6–151.8 V through both 400 A holds. Above ~150 A of regen the logged DC-link current pinned at ~54–58 A — the Sorensen at its sink limit — while the power increase showed up entirely as bus voltage until the Chroma clamp took over. The modulator voltage limiter never engaged (cg_vlimit_scale = 1.000 for the entire session); all clamping was external. Current control tracked cleanly: raw and filtered Iq agree, and hold 2 averaged 357.5 A only because its window includes the ~8 s, ~50 A/s slew-limited ramp from the IqVar 400 re-application to full current.

• Machine-terminal power (calculated): from the logged dq quantities (peak-amplitude convention, P = (3/2)(vd·id + vq·iq), Q = (3/2)(vq·id − vd·iq)). Hold 1: the machine delivered 8.39 kW while drawing ≈5.0 kvar (9.74 kVA, PF ≈ 0.86); hold 2: 7.52 kW and ≈4.3 kvar (8.67 kVA, PF ≈ 0.88). Machine-terminal and DC-link power agree within ~0.3% in both holds — calculated inverter efficiency ≈ 100%, telemetry-resolution limited at this power (the same calculation shows ≈95% at 13 kW in the 450 A run, where the loss is larger than the measurement offsets).

Thermal results

Temperatures were benign for the full session:

• Baseplate TS1 (temp_inv1_c): 28.1 °C start, 53.0 °C at the end of the first hold, 63.9 °C peak at t ≈ 203 s (just after the second stop — the thermal peak lags the power).
• Baseplate TS2 (temp_inv2_c): 31.2 °C start, 59.5 °C peak at t ≈ 202 s.
• Both channels cooled to ~41–43 °C by the end of logging. TS3 was not enabled for this session (wiring defect, unrelated to the test) and temp_motor_c is unpopulated in this recording.

Current-spike observation

Resolved — measurement artifact, not an inverter anomaly. During the first 400 A hold, the oscilloscope showed narrow, repetitive spikes superimposed on the phase-current waveforms, concentrated near the current peaks (photos above); they were not observed at lower currents. The cause, established during the follow-up 450 A run, is the measurement channel: the Tektronix current clamps are rated at 150 A and clipped/saturated at the 400 A phase currents, producing exactly this kind of spiked, distorted trace. See OV-TEST-HW-REGEN-450A-GEN7-SIZE2 for the follow-up run and resolution note.

The telemetry evidence from this session is consistent with that conclusion:

• The ~56 Hz telemetry publish rate cannot resolve waveform-scale spikes: the electrical frequency is ~123 Hz, so per-sample phase-current values are aliased by design.
• Everything is clean: all 19,012 current samples valid (cg_sample_valid), gate_fault = 0 for the entire session, no desaturation events, cg_vlimit_scale = 1.000 throughout, the bus sat solidly on the 150 V clamp, and Iq tracking was tight in both holds — none of which would be expected if the spikes were a power-stage or control phenomenon.

No further investigation of the inverter is required on this observation; future sessions should use current transducers rated for the measured current.

Telemetry overview

Open full session: View the run in the Telemetry Viewer

Observations

• The Chroma load bank solved the facility limit: with the Sorensen sinking ~50 A, the 150 V CV clamp absorbed the remainder and made the 300–400 A steps possible at all.
• Above ~150 A of regen, DC-link current pinned at the supply's sink limit and further power increase appeared as rising bus voltage until the clamp.
• The stop command does not cut current quickly: IqVar 0 slews down at the configured rate, reaching under 20 A in ~7.7 s. Emergency expectations should account for this.
• After each stop the bus collapsed from 150 V back to ~74 V in about 5 s as the Chroma stopped drawing.
• A UartError fault flag (0x200) was active from before the first command and throughout the session. Telemetry quality was unaffected (45 bad frames of ~767 k); it is treated as a latched host-link flag, not a run-affecting fault.

Conclusion

Pass. Gen7 size 2 hardware sustained a 400 A q-axis regenerative command — 415 A peak phase current, 8.78 kW peak DC-link power, ≈0.117 kWh returned — with the bus clamped at 150 V by the added Chroma load bank, no gate or control faults, and baseplate temperatures below 64 °C. The repetitive phase-current spikes observed on the oscilloscope at 400 A were subsequently determined to be the 150 A-rated Tektronix current clamps saturating at the measured current — a measurement artifact, not an inverter anomaly (see the 450 A follow-up report).

Artifacts

• Decimated telemetry log (JSONL, 1 sample/s) — open in Telemetry Viewer
• Telemetry overview plot
• Full source telemetry log, retained for traceability
• Photos: phase currents on scope, zoomed spike detail, bench with Chroma wiring, Chroma load bank front panel
• Full-resolution originals of the photos are retained alongside as *-Original.jpg.