48 V and 800 V Answer to Different Documents, Not Just Different Voltages

EV & automotive · 5 min read · 56 cited facts

A programme brief that asks one rig to 'support automotive testing' across 48 V mild-hybrid modules and 800 V packs has specified nothing yet. Automotive power is at least three documents, and they do not overlap. LV 148 governs the 48 V board net. LV 123 governs the traction side, where the 800 V pack lives. ISO 16750-2 governs the 12 V and 24 V control supply that every vehicle keeps regardless of what drives the wheels. Each was written for a different electrical world, and each makes demands on the test equipment that the other two never mention.

Three regimes at one table

Two things in that table are easy to read past. First, the bands never meet: 52 V tops out the LV 148 world well before the lowest LV 123 class begins at 90 V, and no clause bridges the gap — a rig sized for one band earns no credit in the other. Second, the LV 148 source-impedance row specifies the test source itself: 10–100 mΩ (VDA 320). A 48 V rig therefore needs programmable output impedance in the milliohm range, a setting a pack-scale cycler does not carry. The document is prescribing the electrical character of the thing feeding the module, not merely the module's window.

The control supply is its own test programme

ISO 16750-2 reached its fifth edition in 2023, and a control-supply clause should name it. Its supply-voltage codes — Code A 6–16 V through D 10.5–16 V on the 12 V side (Table 3), Code E 10–32 V through H 18–32 V on the 24 V side (Table 4) — decide the whole test before any pulse gets drawn, because every profile hangs off the code's window. And no traction programme escapes this document: the ECU that manages the 800 V pack runs on 12 V or 24 V itself.

The load-dump row does the disqualifying. The pulse arrives from a defined source resistance — Ri 0.5–4 Ω on 12 V systems, 1–8 Ω on 24 V systems — at up to 202 V: a specified source-impedance event rather than a voltage setpoint, which rules out any supply that only knows how to hold a programmed voltage. Two gaps in this section stay gaps on purpose: the momentary-drop levels of 4.6.1.1 did not extract from our source and are not quoted; the 12 V starting-profile values of Table 11 likewise — only the 24 V profile of Table 12 was read cleanly. Clause text here traces to a vendor republication cross-checked against the official preview, so qualification work requires the ISO original.

What this does to the equipment list

Put the three equipment implications side by side and the reason one cycler rarely stretches across two voltage classes becomes mechanical rather than rhetorical. The traction side wants a bidirectional DC supply spanning 0–800 V with ripple superposition and millisecond profile sequencing — the territory of the B2C+ class of converter, up to 800V and 555A in a single chassis. The 48 V side wants 0–70 V with programmable milliohm output impedance. The control side wants 0–36 V with superimposed ripple from 10 Hz to 200 kHz, polarity reversal to −26 V, and a series switch that opens in ≤ 10 µs to ≥ 10 MΩ. Three sets of demands, three documents — and a quotation can only be checked against the ones the programme actually names.

Write the document, not the word

• For the 48 V bus: LV 148, the unrestricted band 36–52 V (VDA 320), the transient set — 70 V for 40 ms (E48-02), 60 V for 60 min (VDA 320) — and the 10–100 mΩ source impedance.
• For the traction bus: LV 123 and the voltage class by name — HV_2b 250–450 V and HV_3 520–750 V (circuit limit 800 V pk) live in the same document and still need different rigs — plus the 20 V/ms overvoltage slew (§10.4).
• For the control supply: ISO 16750-2:2023 with the supply-voltage code spelled out, Table 3 or Table 4, because Code A 6–16 V and Code G 22–32 V exercise different electronics.
• One named document per bus, in writing. A rig quoted against 'automotive' was quoted against nothing.