The same line, "contact discharge", can appear on two reports whose severity differs by an order of magnitude. ISO 10605 defines only the discharge network, the discharge method and the test set-up; the list of discharge points, whether the sample is powered, what is monitored during the test and how a failure is counted are all left to the test plan. The voltage on the report cover carries the least information of anything on it.
1. ISO 10605 covers the vehicle scenario, not device-level ESD
Start by separating three systems that are often conflated:
- IEC 61000-4-2: ESD immunity for general electronic equipment, aimed at consumer and industrial scenarios;
- HBM / CDM in AEC-Q100: device-level damage models for chips during manufacture, handling and assembly, applied to bare die;
- ISO 10605: the discharge when a person touches a component on the vehicle, applied to the component assembly.
The three are not interchangeable. A chip that passes the device-level model is not thereby able to pass ISO 10605 once built into a module — the paths the discharge energy takes are fundamentally different: the former goes into the pins, the latter into the housing, gaps, connectors and harness, and from there couples into the circuit.
The corresponding national standard is GB/T 19951, whose clause framework follows ISO 10605. For clause editions and differences, defer to the official standard text rather than second-hand interpretation.
2. Two RC networks for two human body scenarios
ISO 10605 defines two discharge networks, and this is the watershed between it and general ESD standards:
| Discharge network | Scenario | What to focus on |
|---|---|---|
| 150 pF / 330 Ω | A person's finger touching the component directly (bare hand, fingertip) | Single-event energy and peak current; look at interfaces, keys, connector shells first |
| 330 pF / 2 kΩ | A person charged inside the vehicle discharging into metal parts (seat friction, charged fabric) | Higher energy with a slower rise; look at internal coupling and functional soft failure |
A note on sourcing: the RC values come from published ISO 10605 clauses; for the specific voltage levels, number of discharges and interval between them, defer to the official clauses and the OEM technical protocol.
Pick the wrong network and the data cannot be compared across projects. The first is like a fingertip tap; the second is like a person sliding off a seat and then touching the door handle. Peak current and duration are not in the same league, so the failures they expose are naturally different.
3. Contact versus air discharge: not either-or, but decided by the surface
Contact discharge: the gun tip is pressed directly onto a conductive surface. Contact is reliable and repeatability is good, so use it wherever contact is possible.
Air discharge: when the surface is an insulating coating, a plastic housing, a glass cover, a discontinuous plating or a gap, the gun tip cannot make reliable contact, and air discharge applies. The tip approaches at a defined speed and the discharge occurs by breaking down the air gap.
Air discharge is inherently more scattered than contact discharge. On the same sample, approach speed, angle and ambient humidity all change the actual breakdown voltage, so the approach method and the environmental conditions must be recorded, otherwise results cannot be reproduced between batches.
Watch out for painted metal parts: the paint layer counts as insulation and only allows air discharge, but there is metal directly underneath, so once breakdown occurs almost all the energy is dumped into the housing. These parts tend to produce the recurring pattern of "passes air discharge in the lab, still gets killed on the vehicle".
4. Powered or not: two different tests
The test plan must state the sample state, normally one of three:
- Unpowered: catches hard damage — device breakdown, blown fuses, eroded connectors;
- Powered and operating: catches functional anomalies — resets, communication dropouts, display flicker, actuator misbehaviour;
- Powered and monitored: no functional stimulus, only key signals and operating current monitored.
Quite a few projects run only the first. A part that passes unpowered may lose a function outright from a single reset while powered and operating, and that class of soft failure is precisely what generates field complaints after the vehicle ships. Whether and how the function recovers, and whether recovery needs manual intervention, should be judged against the A–E functional status classes in GB/T 28046.1, not waved through with "normal after restart".
5. Where to strike: the discharge point list deserves more discussion than the voltage
Discharge points are set by the mounting location and by where users actually touch, not by picking a few metal surfaces at random:
- Guaranteed user contact: keys, knobs, screens, the inside of door handles, USB and charging ports, steering-wheel paddles;
- Assembly and service contact: connector shells, brackets, cooling fins, nameplate screws;
- Indirect discharge: horizontal and vertical coupling planes, which test the path where a discharge to a nearby metal part couples energy into the device.
Display parts need their own point list: the cover is glass or PC and allows only air discharge, but the bezel, decorative ring and metal backlight bracket all allow contact discharge. Miss the bezel and you have not covered the real usage scenario.
6. Judging: record soft and hard failures separately
| Symptom | Classification | Common cause |
|---|---|---|
| Function interrupted, automatic reset, communication lost then self-recovers | Soft failure | Reset circuit, watchdog, disturbed MCU, ground bounce |
| Display flicker, artefacts, touch drift | Soft failure (may not recover) | Coupling into the panel flex, disturbed touch IC |
| Device breakdown, blown fuse, eroded connector | Hard failure | Energy entering the low-voltage circuit directly |
| Permanent parameter drift, lost calibration | Latent failure | Disturbed memory write, damaged reference source |
The point of that table is this: define "pass" before you judge it. Whether a soft failure that self-recovers counts as acceptable depends on the functional status class the project requires, and that has to be written into the test plan. A report that says only "function normal after test" without saying what happened during the test carries little weight at OEM review.
One coupling is worth noting: ESD and radiated immunity often strike the same weak point, the coupling path at interfaces and harness (the set-up differences in CISPR 25 radiated emission). TVS diodes, common-mode chokes and shielding added for ESD usually improve immunity performance as well; filtering added only for radiated immunity, conversely, may not survive ESD peak current.
7. A few straight answers
ESD is easily reduced to a box-ticking exercise: sample unpowered, strike a few metal surfaces, it boots after reset, write "pass". The event that actually causes trouble usually happens with the unit powered and the user's finger on the screen bezel.
Align four things before sending samples: which RC network, the discharge point list, the sample state and monitored quantities, and the criteria for soft failure. When TMC LAB takes on an ESD commission, these four go into the test plan before a slot is booked, so that nobody discovers afterwards that a critical point was never covered.
Export projects add one more layer: component EMC type approval runs through ECE R10, ESD falls under immunity, and the report has to be submitted as a set with immunity data from the ISO 11452 series — a standalone ESD report cannot carry a certification. For clause frameworks start with the CISPR 25 and GB/T 28046 standard pages; the service scope and test list are under EMC testing.
Turnaround time is arranged according to the actual project. You are welcome to state your target date when submitting samples; we will check it against the current schedule, confirm a feasible plan with you, and update you at each milestone.
