Electrical load testing rarely fails because the equipment is inadequate; it fails because the condition table was copied from the wrong class. Run the same ECU point by point across the upper and lower limits of its operating range, or run it under sustained overvoltage, and the conclusions can be opposite — the first asks whether function drops out at undervoltage, the second whether the power supply chip has margin. GB/T 28046.2 covers exactly this: it does not ask whether your unit interferes with anything else, only whether it survives, and in what state, when the supply is abused.
1. First, draw the line: electrical loads are not EMC
Domestic projects routinely lump the two together under "electrical performance", and getting the report cover wrong is common.
- Electrical loads: DC and the low-frequency domain. Supply voltage high and low, superimposed ripple, voltage drop and slow rise, load dump, reverse connection, pin interruption, ground offset. What is assessed is withstand and retention of function.
- EMC: the high-frequency domain. Conducted and radiated emission, radiated immunity, BCI, electrostatic discharge. What is assessed is interference going in and out.
The difference is not the price of the equipment, it is where the criterion lands. Electrical loads ultimately land on one specific class from functional status A to E (the A–E grading of GB/T 28046.1); EMC lands on limit lines and immunity levels.
In one line: electrical loads ask what the unit does when the supply is broken; EMC asks how the unit and the outside world interfere with each other. If the customer agreement only says "electrical performance test", go back and establish which of the two is meant, otherwise what you deliver will not be the report they wanted.
2. The three core items: what the conditions look like
The three items named in the glossary — DC supply voltage, overvoltage, superimposed AC voltage — are the ones most often run on their own, and the ones most often left vague in the condition table.
| Test item | Common published values for 12 V systems | What it is mainly looking for |
|---|---|---|
| DC supply voltage | Hold point by point near the lower and upper limits of the nominal operating range (commonly 9 V to 16 V for 12 V systems), dwelling long enough at each point to observe function | Undervoltage reset, low-voltage derating, display flicker, communication dropout |
| Overvoltage | One common level is 18 V for 60 min (12 V systems), applied at the supply input | Voltage margin of the power devices, clamping, thermal and protection behaviour of LDO/DC-DC stages |
| Superimposed AC voltage | A sine ripple superimposed on the DC, amplitude banded by class (commonly 1 V to 4 V peak-to-peak), swept from low frequency up through the audio and switching bands (covering roughly 50 Hz to 25 kHz) | Filtering and decoupling failing in a specific band, output jitter, occasional resets |
A note on the figures: the table above summarises commonly published values for 12 V systems, for scheduling and for discussing the plan. The specific levels, durations, application points and tolerances are governed by the standard text and by your technical agreement. A 24 V system uses a completely different set of numbers, so this table cannot simply be shifted across.
The third item is the most under-estimated of the three. The first two are about holding one voltage and seeing whether the unit takes it; ripple keeps the voltage moving — and it is the dynamic process that exposes loop problems. Plenty of units hold 18 V overvoltage and reset as soon as the ripple sweep reaches a few kHz.
3. Setting the conditions: three deciding factors
The condition table is not a matter of copying a line out of the standard. Three upstream factors change the outcome.
First, the system voltage. The baseline levels for 12 V, 24 V and 48 V systems are entirely different. Get this wrong and everything after it is wrong.
Second, the supply topology and the application point. Is the unit connected straight to the battery positive terminal, or does it come through a fuse, a relay and a long harness? Whether the application point is at the supply inlet or at the unit, harness voltage drop means the voltage actually reaching the unit differs. For units on long harness runs, the application point has to be stated in the conditions.
Third, the functional status class. The same test judged as class A and judged as class C is two different projects. Class A requires full function throughout; class C allows degradation during it with self-recovery afterwards. Without the class fixed, the report has no conclusion to state.
German-brand projects add another layer: LV 124 and GB/T 28046.2 overlap in test items, and their conditions for ripple and transients do not match exactly (LV 124). Where the customer's corporate standard cites LV 124 directly, follow the corporate standard — do not "align" it to the national standard conditions.
4. Four things to settle before starting
If these four are not settled, the work will most likely have to be redone.
- Sample operating mode. Sleep, standby, full load, communication wake-up — the same test can conclude differently in each. Sleep current is only a few milliamps, so a supply sag can trigger wake-up; at full load the supply has the least margin, so overvoltage most easily trips protection. The mode table belongs in the test outline.
- Monitoring points. Between the voltage measured at the supply inlet and the voltage at a critical chip pin inside the board there is a fuse, a diode and trace resistance. The criterion says "function normal", but abnormal function usually shows up first on some chip's supply pin. Measure both, inlet and inside the board.
- Reset criterion. Reset is the most common failure form under ripple and supply drop. The criterion has to state which signal counts as a reset, how many are allowed, the upper limit on recovery time, and whether state must be written back after a reset. Writing only "function normal after the test" gives up the evidence from during the test.
- Edition year. Part 2 of the national standard is the 2011 edition; ISO 16750-2 has since been revised. Where the agreement cites ISO, take the levels from the ISO edition; where the two disagree, take the stricter side and note it in the report. The edition mismatch trap is covered separately (the relationship between ISO 16750 and GB/T 28046); one reminder here: without asking the edition, the conditions are a guess.
If the customer has only given the functional status verbally, get the class confirmed in writing before the chamber is opened. The class is the root of the criterion, and with an unstable root the data is worthless however good it looks.
5. A few honest words
Almost all of the cost in electrical load work sits in whether the condition table is right and whether the monitoring is thorough. The equipment is rarely the bottleneck — supplies, oscilloscopes and ripple sources exist in most laboratories.
The real difference is in three places: mode definition, monitoring point placement, and how the criterion is written. Write these vaguely and one question at review will stop the report; write them clearly and the same equipment produces evidence of a completely different quality.
When TMC LAB takes on electrical load work, the operating mode table, monitoring point locations and functional status class go into the test outline and are confirmed before power is applied, so it does not turn out afterwards that the criterion does not match the agreement. For a clause breakdown, start with the GB/T 28046 and ISO 16750-2 standard pages; for how electrical and mechanical loads divide the work see the condition differences for mechanical loads, and the test service entry point is automotive EMC and electrical loads.
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.
