Conclusion first: "mechanical load" in GB/T 28046.3 is not one test item but a bundle of three stresses — sinusoidal vibration, random vibration and mechanical shock — plus free fall and stone impact. On the same board, sinusoidal vibration exposes local fatigue at resonance points, random vibration exposes cumulative damage across the whole frequency band, and shock exposes transient inertial force. Their failure targets do not overlap, so skipping one means dropping a whole class of failure mode.
1. Three stresses, three different failure targets
Sinusoidal vibration is single-frequency excitation. Sweeping the frequency finds the resonances — a bracket, a panel or the connector on a long cantilever is amplified several times over at a particular frequency. Standard practice is to sweep at a low level first to locate the resonance, then dwell at that frequency so that fatigue concentrates at the weakest point.
Random vibration is broadband excitation. Energy is applied across the band at once, with the distribution described by PSD (power spectral density, g²/Hz). It is closer to the broadband character of real road and engine excitation, and it targets cumulative damage in small-scale connections such as solder joints, pins and BGAs across the whole band.
Mechanical shock is transient excitation. A half-sine pulse delivers high acceleration over a short time, testing the inertia of heavy devices and the retention of fasteners. Relays, electrolytic capacitors and heat sinks — parts that are heavy and cantilevered — move first under shock.
| Test type | Main failure targeted | How conditions are described |
|---|---|---|
| Sinusoidal vibration | Local fatigue at resonance, cracking of brackets and cantilevers | Frequency range, acceleration amplitude, sweep rate, dwell time |
| Random vibration | Cumulative broadband damage in solder joints and pins | PSD profile (g²/Hz), overall RMS acceleration, test duration |
| Mechanical shock | Inertial failure of heavy devices, fastener loosening, housing cracking | Peak acceleration (g), pulse duration (ms), waveform |
| Free fall | Drop damage during assembly, transport and service | Drop height, drop surface, number of drops, attitude |
A note on sourcing: the condition names follow the published clause framework of GB/T 28046.3 and ISO 16750-3; the actual values are set by mounting location and are subject to the official clauses and the OEM technical protocol.
2. The class follows mounting location, not part weight
Two controllers of the same size, one in the cabin and one in the engine bay, are not in the same league mechanically. The engine bay adds engine ignition excitation and higher ambient temperature; unsprung underbody locations take road shock directly; parts near the driveline must also account for torque fluctuation.
12 V and 24 V systems differ little here — mechanical load depends on where the part is mounted, not on the system voltage.
A common mistake is choosing the class by part weight. Weight affects fixture design and shock response, but it does not change the class the standard assigns. The class depends only on mounting location; get it wrong and the subsequent conditions, acceptance items and report conclusions all have to be redone. Temperature classes work the same way in the companion article: fix the location first, then the class, and only then discuss conditions (what changed in the ISO 16750-4:2023 temperature classes).
For the full clause framework, compare the GB/T 28046 environmental testing and ISO 16750 standard pages: line up the clause numbers and sub-items first, then check them against the customer protocol.
3. Fixtures and axes fail more often than test conditions
When a test produces no result, it is usually not the conditions that were written wrong but the fixture that gave up first.
The natural frequency of the fixture must be above the upper limit of the test band. If the fixture resonates on its own it amplifies energy at some frequencies and absorbs it at others, so the acceleration measured on the sample no longer matches the setpoint and the report loses traceability. The control accelerometer goes on the interface between fixture and sample, not on the shaker table — the two positions frequently read quite differently.
The same applies to axes. All three orthogonal axes normally have to be run, preferably oriented to the in-vehicle attitude. Reporting on a single axis leaves the failure risk in the other two directions to the vehicle integration stage, and by then the cost is no longer that of one test.
4. Four things to settle before submitting samples
1. Mounting location and class. Put it on the order form with the protocol clause it is based on; do not just write "to the national standard".
2. Sample state. Powered or unpowered, loaded or unloaded, monitored online or not. A loaded test can catch intermittent opens and contact resistance jumps during vibration, failures that static re-measurement usually misses.
3. Acceptance items. How functional status is judged (how functional status A to E is decided), which parameters are monitored, at what sampling rate, and whether an intermittent open counts as a failure. Without clear acceptance items the report can only be read as each party prefers.
4. Mounting scheme. Whether there is a fixture that reproduces the in-vehicle attitude, what tightening torque is used, and where the control point goes. When TMC LAB takes on a mechanical load commission, the mounting location, the fixture scheme and the control point position go into the test plan and are confirmed with the customer before the shaker runs, so nobody finds afterwards that the control point was wrong and the data unusable.
5. A few straight answers
The step most often cut from mechanical load testing, and the one that should never be cut, is the resonance search. Without it, sinusoidal vibration becomes hours spent at the wrong frequency, looking like work but hitting nothing. Random vibration has its own shortcut: match the Grms, change the shape of the PSD profile, and the number on the report is the same while the damage distribution is entirely different.
Vibration and shock are also often discussed together with thermal shock, yet the mechanisms are far apart — one tests mechanical fatigue, the other thermal stress (how to set the temperature range and cycle count for thermal shock), and only running both counts as coverage. Shaker time and fixture fabrication consume equipment hours; for the practical arrangements see the vibration testing service page.
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.
