DO-160 vs MIL-STD-704: Where Civil and Military Aircraft Power Standards Split

Aviation & 400 Hz · 6 min read · 67 cited facts

A programme that has to satisfy both MIL-STD-704 and DO-160 Section 16 is not running one test campaign twice. The two documents describe the same physics from opposite ends: MIL-STD-704F defines what the aircraft's electrical system is allowed to deliver at the equipment terminals, while DO-160 defines what the equipment must tolerate, sorted by category. The numbers overlap enough to tempt you into a single test plan and diverge enough to wreck one written by analogy.

The division of labour matters before any limit does. 704F is a power-quality definition — the test methods live separately in MIL-HDBK-704-1 through -8, numbered by power type as the SAC, TAC, SVF, TVF and SXF series. DO-160 §16 is itself the test method: conditions, durations and tolerances in one document. A statement of work that says 'tested to 704F' without naming the handbook methods has not specified a test. One that says 'DO-160 §16, category A(CF)' nearly has.

The overlap: same 115 V bus, different numbers

Read the surge rows against each other. 704F's normal transient reaches 180 V but is over in 10 ms. DO-160's normal surge stops at 160 V yet holds for 30 ms, and its abnormal surge sits at 180 V for a full 100 ms before stepping down to 148 V for 1 s. The amplitudes are close; the durations are not. A source with the voltage headroom for one envelope has not demonstrated the energy delivery or the sequencing for the other.

Where DO-160 asks for things 704F frames differently

DO-160's §16.5.1.4 defines momentary interruptions as a matrix: 19 profiles from 2 ms to 200 ms at 0 V, plus partial sags of 15/35/50/65%. It also demands voltage modulation — an envelope to 5.0 V at 115 V across 1–200 Hz (§16.5.1.2). 704F's limits are organised around bus states instead — normal, abnormal, emergency, starting — so a bench built to step through DO-160's numbered profiles still needs envelope sequencing added before it can walk 704F's state transitions, and the reverse is equally true.

The traffic runs the other way on abnormal and low-voltage states. 704F defines its abnormal overvoltage as a curve — V = 124.6 + 2.77/t for 0.05 ≤ t ≤ 6.925 s (Figure 4) — where DO-160 holds discrete points: 97 V and 134 V for 30 min, undervoltage at 60 V and 10 V for 7 s (§16.5.2). And 704F carries whole operating regimes on the DC side: emergency operation at 16–29 V (§5.3.2.3) and electric starting at 12–29 V (§5.3.2.4) on the 28 V bus. DO-160's nearest counterpart is the category B/Z engine-start profile — a drop to 10 V, then a 0.3 V/s ramp for 35 s up to 20.5 V (§16.6.1) — a slow ramp rather than a held state.

Frequency: one standard bounds it, the other categorises it

704F holds the 400 Hz bus at 393–407 Hz with 4 Hz modulation (Table I), spans 360–800 Hz for variable-frequency platforms, and lets frequency move at 250 Hz/s in normal operation and 500 Hz/s in abnormal (Table II). DO-160 splits the same territory into categories: A(CF) at constant 400 Hz, A(NF) across 360–650 Hz, A(WF) across 360–800 Hz. The procurement consequence is real — a DO-160 programme can legitimately buy for its category and nothing more, while a 704F variable-frequency programme needs the full sweep and programmable df/dt from day one.

Where both documents stand right now

MIL-STD-704F is current and set to stay that way. Revision F dates from 12 March 2004 with Change 1 of 5 December 2016, and the DLA ASSIST record carries a Revision F Notice 4 validation dated 24 June 2026, with the next review due in June 2036. Revision G does not exist to wait for. Design against F and cite the handbook methods by number.

DO-160 is the document in motion. RTCA SC-135 closed its Review and Comment on Revision H during 2025 with more than 2,000 member comments across all 26 sections, and Section 16 was rewritten specifically to be easier to apply across aviation equipment. The committee is dispositioning comments through the first half of 2026, with publication expected in the autumn. None of that changes the levels a source has to produce — it changes which document your test plan cites by the time the report is written.

Can one bench cover both?

Mostly, yes — and this is where the overlap pays. Both standards resolve to a programmable AC source reaching 180 V RMS line-to-neutral across 360–800 Hz with millisecond envelope sequencing, and both then place a second, separate demand on the DC side, where the 28 V and 270 V buses carry their own transient envelopes. What does not transfer is everything around the source: the sequences, the hold durations, the tolerances and the report structure belong to their own documents. Budget the second campaign as a campaign, not as a re-run with different headers.