Vehicle Voltage Transients and Load Dump: What ISO 7637-2 and ISO 16750-2 Require

A 12 V vehicle supply is not 12 V, and the gap between the label and the wire is what destroys LED lamps. Under ISO 7637-2 a 12 V circuit is tested against pulses from −100 V to +150 V; under ISO 16750-2 an unsuppressed load dump can reach 101 V behind a source impedance as low as 0.5 Ω; a 24 V system is tested to −600 V and 202 V on the same two families of test. A lamp that survives one of those conditions has not automatically survived the other, and a datasheet line that names neither the severity level nor the source impedance cannot tell you which one was tested.

Quick answer

The number that matters for a vehicle lamp is not “12 V”. It is which pulse, at which severity level, with which source impedance, and with which functional status the lamp is allowed to show while the pulse is applied. A claim of “ISO 7637-2 protected” that names none of those four items cannot be checked by anyone, including the supplier who printed it, and a claim of “load dump protected” that gives no voltage and no impedance cannot be sized against a real alternator.

The division of labour between the two standards is simpler than the datasheets make it look. ISO 7637-2 is the transient-conduction standard: its part 2 is titled “Electrical transient conduction along supply lines only”, and it defines the short, fast, switching-related pulses. ISO 16750-2 is the electrical-loads standard: it describes supply-voltage quality events, the slower conditions that mostly do not destroy hardware but do change what it does, and since the 2011 edition of ISO 7637-2 it also carries the cranking pulse and the dump pulse that used to live in part 2. UN Regulation No. 10 sits above both as a type-approval requirement and reaches only part of the way into the subject: it makes a sub-assembly prove transient immunity at one fixed level, and it does not test the dump at all.

For a buyer the useful conclusion is narrow and testable. Ask for a report that states the standard, its edition, the pulse number, the severity level, the source impedance, and the functional status achieved. Anything less is a marketing sentence, not an engineering claim.

What 12 V means at the terminals

Every test in this article is defined against a reference voltage higher than the nominal label, and that single detail explains a large share of the confusion in supplier documentation. AMETEK’s ISO 7637-2 seminar material gives the supply conditions: a continuous source with an internal resistance below 0.01 Ω, and a test voltage UA of 13.5 V ± 0.5 V for a 12 V system or 27 V ± 1 V for a 24 V system. ST’s automotive driver datasheet says the same thing from the other direction, stating that its ISO 7637-2 test levels “must be considered referred to VCC = 13.5 V except for pulse 5b”.

So a “12 V” lamp is not tested at 12 V. It is tested at 13.5 V and then hit with pulses whose amplitudes are referenced to that same 13.5 V. A driver specified to survive “16 V” has about 2.5 V of margin over the charging system at steady state, which is nothing at all during a dump.

Quantity	12 V system	24 V system	Source
Nominal label	12 V	24 V	Vehicle convention
Test reference UA	13.5 V ± 0.5 V	27 V ± 1 V	AMETEK ISO 7637-2 seminar
Supply source impedance	below 0.01 Ω	below 0.01 Ω	AMETEK ISO 7637-2 seminar
OEM alternator clamp, typical	30 V to 40 V	not stated in the same source	TI TIDUC41A
Suppressed dump level US*	35 V	65 V	TI TIDUC42, ISO 16750-2 Test B

The last two rows decide whether a lamp lives, and they are the two rows most often missing from a product page.

Two standards, and the division of labour between them

TI’s 12 V and 24 V protection reference design states the split in one paragraph, and it is the clearest short version available: ISO 7637 is “Road vehicles – Electrical disturbances from conduction and coupling” and part 2 is specifically “Electrical transient conduction along supply lines only”, while ISO 16750 is “Road vehicles – Environmental conditions and testing for electrical and electronic equipment” and part 2 is “Electrical loads”. TI characterises the second as a series of supply-voltage-quality events which “for the most part, are not harmful to the electrical subsystem, but can affect its state of operation”.

That is the useful mental model: the first asks whether a device is destroyed by a fast conducted transient, the second asks how it behaves while the supply wanders or is disconnected and reconnected.

Question the test answers	ISO 7637-2	ISO 16750-2	UN R10
Fast switching transients on the supply line	Yes, pulses 1, 2a, 2b, 3a, 3b	No	Immunity at level III by the ISO 7637-2 method
Slow supply variation and cranking	Handed to ISO 16750-2 in 2011	Yes	Cranking pulse in Table 2
Load dump	Handed to ISO 16750-2 in 2011	Yes, clause 4.6.4	Not covered
Reverse polarity	No	Yes, clause 4.7	Not covered
Radio-frequency emissions	No	No	Yes, with limits

Read the table from the bottom up and the hierarchy appears: UN R10 is a legal gate, ISO 7637-2 and ISO 16750-2 are the technical content that a gate or a purchase contract borrows, and an OEM standard is what actually binds a supplier.

The 2011 edition change that confuses almost every datasheet

ISO 7637-2 has had three editions, and the third moved the two most quoted pulses out of the document. AMETEK’s seminar lists them: the 1990 first edition, the 2004 second edition with a correction dated 2004-09-15 and amendment 1 dated 2008-02-01, and the 2011-03 third edition, whose changes are summarised as “pulse 4 & 5 moved to ISO 16750-2” and “modified: levels pulses 1, 2a, 3a, 3b”.

The same deck then gives the sentence every engineer quoting pulse 5 from ISO 7637-2 should read: “The Pulses 4, 5a, 5b are no longer taken into account, so ISO 7637-2:2004 is still be referred to in case of doubt.” Littelfuse dates the handover slightly differently, writing that ISO 16750-2 replaced ISO 7637-2 for the 5a and 5b portion in 2010, and TI states plainly that the dump pulse “was actually moved from ISO 7637 to ISO 16750”. Three documents, two dates, one substance: the pulse now belongs to ISO 16750-2 and the older numbering survives because component datasheets were written against the 2004 edition and never reissued. The practical consequence is that a “pulse 5b” value today may come from the 2004 edition, from ISO 16750-2, or from a customer specification, and those amplitudes differ.

Pulse 1: the inductive kick from a load in parallel with your lamp

Pulse 1 simulates the supply being disconnected from an inductive load while the device under test remains connected in parallel with that load. TI describes it as a high-voltage, negative-going transient that “occurs when switch is open”, and gives the shape: −75 V to −100 V through 10 Ω for 2 ms on a 12 V system, and −450 V to −600 V through 50 Ω for 1 ms on a 24 V system.

Two features of that row matter. The polarity comes first: pulse 1 is negative, so a reverse-blocking element that is fine at −1 V of accidental reverse battery is not necessarily fine at −100 V, and a TVS that clamps positive excursions does nothing unless it is bidirectional or paired with a series element that blocks.

The second is the arithmetic hidden in the impedance. On a 12 V system, 100 V into 10 Ω is 10 A; on a 24 V system, 600 V into 50 Ω is 12 A. Voltage tells you what has to stand off, and impedance tells you what has to be absorbed.

ST’s datasheet quantifies the repetition at 5000 pulses for levels III and IV.

Pulses 2a and 2b: harness inductance and motor rundown

Pulse 2a is the positive counterpart of pulse 1, and it catches people out because it arrives during normal operation. TI’s description is that it “simulates transients due to sudden interruption of currents in a device connected in parallel with DUT due to inductance of the wiring harness”; AMETEK adds the switching condition, noting that pulse 2a “occurs when the load switch is opened while the ignition switch is closed”, so the supply is not removed while the pulse is applied and the device stays powered. Pulse 2a is a functional test, not only a survival test.

The amplitudes are modest by this article’s standards: 37 V to 50 V through 2 Ω, 50 μs duration, rise time under 1 μs. ST and TI agree on those numbers for levels III and IV on a 12 V system. What makes 2a consequential is the repetition — 5000 pulses at intervals as short as 0.2 s, applied while the lamp is lit.

Pulse 2b is different in kind. AMETEK is blunt: “Pulse 2b simulates transients from DC motors acting as generators after the ignition is switched off”, and Littelfuse uses the same words. Littelfuse’s level table gives +10 V for a 12 V system and +20 V for a 24 V system, with 10 pulses. A 10 V excursion sounds trivial next to a 100 V one, which is exactly why it is worth naming: 2b is the pulse most likely to be left out of a marketing claim, because a supplier can truthfully say the lamp “passed the high-voltage pulses” while never testing the one a cooling fan or window motor generates on every shutdown.

Pulses 3a and 3b: the nanosecond burst

Pulses 3a and 3b are bursts of very fast transients produced by switching, and they are the reason a lamp driver needs more than a large capacitor. TI’s tables give a rise time of 5 ns ± 1.5 ns, a source impedance of 50 Ω, and a burst structure built from t1 = 100 μs, t4 = 10 ms and t5 = 90 ms; ST gives the same burst timing and a 0.1 μs pulse width, with a total test time of one hour. The polarity separates them: 3a is the negative burst, 3b the positive one.

A 5 ns edge is a radio-frequency event wearing a voltage label. It couples through stray capacitance, it will not be stopped by a component with a few millimetres of lead length, and it stresses an input stage in a way a 400 ms dump never does. AMETEK makes the spectral point explicitly: each transient has an edge in the nanosecond range and therefore generates a very wide interference spectrum. That is also why this pulse family appears in both the immunity standard and the emission standard.

The amplitudes quoted for 3a are not consistent across the documents opened for this article, and the disagreement is large enough to deserve its own table rather than a footnote.

Pulse 4: the cranking dip

Pulse 4 is the one pulse that is not a spike. AMETEK defines it as the test that “simulates supply voltage reduction caused by energizing the starter motor circuits of internal combustion engines, excluding spikes associated with starting” — a deliberate exclusion, because the spikes are pulses 1 and 3 and the dip is a separate phenomenon with a separate duration.

ST’s level table gives pulse 4 as −6 V at level III and −7 V at level IV, applied once for 100 ms through a source impedance of 0.01 Ω. The 0.01 Ω figure is the important one: this is not a transient with a limited current behind it, it is the battery being dragged down by a starter, and anything on the rail sees the voltage collapse and then recover.

For a lamp driver, pulse 4 is a constraint on the low side rather than the high side. A constant-current stage that loses regulation at 9 V will drop out during a cold crank and come back when the engine catches; whether that is acceptable is a functional-status question, not a survival question.

The load dump: two conditions that are tested differently

The event is described consistently across sources: the battery is disconnected while the alternator is charging, and the alternator’s output, no longer loaded by the battery, rises until something limits it. TI lists cable corrosion, a poor connection and intentional disconnection with the engine running as causes, and notes that because the raw event is too energetic to protect against in every subsystem, manufacturers fit a clamp to the alternator, limiting the rail to a level that varies by manufacturer but is typically 30 V to 40 V. Littelfuse confirms that most alternators carry such a clamp and adds that it is still not sufficient on its own.

The test comes in two conditions, and the labels are where documentation goes wrong.

The unsuppressed condition assumes there is no working clamp on the alternator, so the full excursion reaches the lamp. The suppressed condition assumes the suppressor is present and functioning, so the lamp sees the clamped level. Littelfuse’s protection table maps the two onto the legacy numbers directly: 5a is the case where the alternator has no centralized suppression and therefore does need protection, and 5b is the case where a central suppressor is present and it does not. ST’s datasheet takes the same view from the component side, listing its 5b row as valid only when an external clamp limits the rail to 40 V maximum referred to ground.

That mapping is the reverse of what some summaries state, and the disagreement is not a detail. This article follows the sources it opened: three of them, from two manufacturers and one test house, place the suppressed condition on 5b. A buyer reading a document that says “5a, suppressed” should treat the label as unverified and ask for the waveform, because the number that matters is the clamped voltage, not the letter.

Condition	Label in the sources opened here	What the lamp sees	12 V values	24 V values
Unsuppressed	5a	Full alternator excursion, no clamp	79 V to 101 V, Ri 0.5 Ω to 4 Ω, td 40 ms to 400 ms	151 V to 202 V, Ri 1 Ω to 8 Ω, td 100 ms to 350 ms
Suppressed	5b	The alternator’s clamped level US*	US* 35 V	US* 65 V

Those values are the ISO 16750-2 ones, as reproduced by Littelfuse and in TI’s Test B table. Two further differences between the old and new homes for this pulse change how a part is qualified. Littelfuse states them as bullets: ISO 16750-2 defines 10 pulses in 10 minutes at one-minute intervals while the old ISO 7637-2 defined a single pulse, and ISO 16750-2 defines a higher US range than ISO 7637-2 did. The first means a suppressor must survive a sequence, not a hit. The second means an old qualification is not a conservative one.

Every pulse in one table, and where the documents disagree

The pulse shapes differ in more than amplitude, which is why one protection strategy cannot cover all of them.

The table below collects the level III and IV values for a 12 V system from ST’s automotive driver datasheet, the only opened source that lists both levels, the pulse count, the source impedance and the duration for every pulse in one place. The 24 V column uses Littelfuse’s level table where ST is silent.

Pulse	What it models	12 V level III	12 V level IV	24 V level III	24 V level IV	Source impedance	Pulses or duration
1	Supply disconnected from a parallel inductive load	−75 V	−100 V	−300 V	−600 V	10 Ω at 12 V, 50 Ω at 24 V	5000 pulses, 2 ms at 12 V
2a	Harness inductance interrupts a parallel load	+37 V	+50 V	+37 V	+50 V	2 Ω	5000 pulses, 50 μs
2b	DC motor acting as a generator after ignition off	+10 V	+10 V	+20 V	+20 V	not stated in the same table	10 pulses
3a	Negative burst from mechanical switching	−100 V	−150 V	−150 V	−300 V	50 Ω	1 h, burst 90 ms and 100 ms
3b	Positive burst from mechanical switching	+75 V	+100 V	+150 V	+300 V	50 Ω	1 h, burst 90 ms and 100 ms
4	Starter-motor cranking dip	−6 V	−7 V	not stated	not stated	0.01 Ω	1 pulse, 100 ms
5b	Dump with the alternator clamp in circuit	+65 V	+87 V	+123 V	+173 V	2 Ω	1 pulse, 400 ms

Now the disagreements, which are why this section exists rather than a single clean table.

Pulse and system	ST datasheet	TI TIDUC41A	Littelfuse 2023	Size of the disagreement
3a, 12 V, levels III and IV	−100 V and −150 V	112 V to 150 V, quoted as a range	−112 V and −220 V	Up to 70 V at level IV
2a, 12 V, level IV	+50 V	37 V to 50 V	+112 V	More than a factor of two
1, 24 V	not listed	−450 V to −600 V	−300 V and −600 V	150 V at level III
5a versus 5b	5b carries the external clamp note	pulse moved to ISO 16750-2	5a unsuppressed, 5b suppressed	Label, not value

None of these is a rounding difference, and none can be resolved without the standard itself, which is sold rather than published. What a buyer can do is refuse to accept a pulse number as a specification. “Pulse 3a” is a shape; “pulse 3a, level IV, −150 V, 50 Ω, 2011 edition” is a requirement. The gap between those sentences is where most lighting claims live.

12 V versus 24 V: what actually changes

A 24 V system does not simply double every number, and treating it as a scaled 12 V system produces the wrong protection. The voltage levels roughly double, the source impedances roughly double with them, and the result is that the current a suppressor must sink stays in the same range.

Quantity	12 V worst case	24 V worst case	Arithmetic
Pulse 1 peak current	10 A	12 A	100 V / 10 Ω and 600 V / 50 Ω
Unsuppressed dump peak current, no clamp	202 A	202 A	101 V / 0.5 Ω and 202 V / 1 Ω
Dump current with a 50 V clamp, Ri 4 Ω	13 A	38 A	(101 − 50) / 4 and (202 − 50) / 4
Suppressed level the driver must tolerate	35 V	65 V	ISO 16750-2 US*

The first two rows are the reassuring ones: a 24 V lamp is not facing ten times the stress of a 12 V one. The third row matters for a real design, and it is Littelfuse’s own worked example for 24 V, giving 38 A against a device rated 40.8 A for ten pulses at 220 ms. That is a 7 percent margin on paper, before any temperature derating, and Littelfuse explicitly warns that pulse withstand capability falls as ambient temperature rises.

The fourth row is where a 24 V lamp differs in the field: the suppressed level is 65 V, not 35 V, so every input-stage component has to stand off a voltage that would destroy a 12 V design. A “12 V and 24 V compatible” claim is therefore a statement about input-stage ratings, and it should come with a number.

What a pass is allowed to mean: functional status

A test result without a functional status is half a result, and this is the part of the standards that most product pages omit. AMETEK defines the five classes used across this family of tests.

Class	During the disturbance	After the disturbance
A	All functions perform as designed	All functions perform as designed
B	All functions perform as designed, but one or more can exceed tolerance	All functions return automatically to normal limits
C	One or more functions do not perform as designed	All functions return automatically to normal operation
D	One or more functions do not perform as designed	Normal operation only after a reset by simple operator action
E	One or more functions do not perform as designed	Cannot be restored without repair or replacement

For a lamp, class C during a cranking dip means the light goes out and comes back by itself. Class B means it may dim during the event but must not require anything from the driver. Class D means someone has to switch the lamp off and on again, which for an auxiliary lamp on a trail at night is a different experience from a minor inconvenience. Class E means the lamp is scrap.

The classes explain why two products can both claim “passed pulse 4”. One passed at status B and the other at status C, and only one suits a lamp that has to stay lit while the engine cranks.

Reverse voltage and the rest of the electrical-loads test

The dump gets the attention, but a lamp on a service vehicle meets reverse polarity more often, because reversing a battery is an ordinary human act. ISO 16750-2 clause 4.7 covers it, and TI’s reverse-polarity reference design gives both the voltages and the reasoning: 14 V is applied to a 12 V system and 28 V to a 24 V system, and the device is expected to withstand the condition for at least 60 s ± 6 s. TI’s explanation is that the alternator’s rectifier diodes survive reverse voltage for that long, so the rest of the electronics are expected to survive until the fuse or the alternator is replaced.

The same document records the choice that decides which protection a designer picks. A Schottky diode in series blocks reverse current completely, but at 5 A it drops about 0.4 V and dissipates about 2 W; TI’s MOSFET design dissipates about 0.1 W at the same current, twenty times less heat inside a sealed housing. The choice between a diode and a MOSFET is therefore a thermal decision as much as an electrical one.

Test	Clause	12 V condition	24 V condition	Duration
Reverse voltage	ISO 16750-2 4.7	14 V reverse	28 V reverse	at least 60 s
Dump, unsuppressed	ISO 16750-2 4.6.4	79 V to 101 V, Ri 0.5 Ω to 4 Ω	151 V to 202 V, Ri 1 Ω to 8 Ω	40 ms to 400 ms
Dump, suppressed	ISO 16750-2 4.6.4	US* 35 V	US* 65 V	same as unsuppressed
Cranking dip	ISO 7637-2 pulse 4, legacy	−6 V to −7 V	not stated in the same table	100 ms

Where UN R10 fits, and where it stops

UN Regulation No. 10 is the electromagnetic-compatibility type-approval regulation of the 1958 Agreement, and the published text is unusually clear about its own limits. It applies to vehicles of categories L, M, N and O and to the components fitted to them. Paragraph 6.1.1.1 draws the first boundary: a vehicle “shall be tested for radiated emissions and for immunity to radiated disturbances”, and “no tests for conducted emissions or immunity to conducted disturbances are required for vehicle type-approval”. Paragraph 6.1.1.2 puts the conducted work on the sub-assembly instead: ESAs are tested for radiated and conducted emissions and for immunity to radiated and conducted disturbances.

The conducted-transient tests then point at ISO 7637-2 by name. Paragraph 6.7 covers emission of transient conducted disturbances from an ESA on 12 V or 24 V supply lines, tested by the Annex 10 method with the levels in Table 1. Paragraph 6.9 covers immunity of an ESA to the same disturbances, again by the ISO 7637-2 method, with the levels in Table 2 and a functional status per pulse. The Table 2 text in the published regulation uses test level III, allows status C for pulse 2b, requires status A for 3a and 3b on immunity-related functions, and splits pulse 4 between status B and status C depending on whether the ESA must work during engine start.

R10 provision	What it establishes	What it does not establish
6.1.1.1	No vehicle-level conducted transient tests	Anything about a lamp on a bench
6.1.1.2	ESAs are tested for conducted emission and immunity	That an aftermarket lamp was tested
6.7	ESA transient emission per ISO 7637-2, Table 1 levels	Immunity
6.9	ESA transient immunity per ISO 7637-2, level III	The dump
2.12(c)(i)	Lighting functions are immunity related	A performance limit
3.2.9	Aftermarket parts unrelated to immunity functions need no approval	An exemption for lamps
Appendix 1	ISO 7637-2 second edition 2004 is referenced	Any reference to ISO 16750

Three consequences follow. First, lighting is explicitly inside the scope: paragraph 2.12(c)(i) lists “incorrect operation of e.g. direction indicators, stop lamps, end outline marker lamps, rear position lamp, light bars for emergency system” among the disturbances that must not occur. A lamp that flickers under a transient is not a cosmetic problem under this regulation.

Second, the aftermarket exemption in paragraph 3.2.9 does not read as it is often paraphrased. The text exempts components “if they are not related to immunity related functions”, and requires a manufacturer’s declaration that the ESA meets the regulation’s limits. Because lighting is on that list, a lamp does not fall into the exemption by virtue of being sold as an accessory.

Third, Appendix 1 lists the standards the regulation references, and that list of twenty items includes ISO 7637-2 in its second edition of 2004. It does not include ISO 16750. The dump is therefore outside the R10 gate, which is why manufacturers specify it themselves: Ford’s FMC 1278 specification carries a section headed “Immunity from load dump: CI 222”. An R10 mark on a lamp says something real about pulses 1 to 4 and nothing at all about the dump.

Why an LED lamp is less forgiving than a filament bulb

The reason a driver needs protection at all is that an LED lamp is not a resistor with a filament in it. A filament is a piece of wire that heats up: its resistance rises with temperature, its thermal mass absorbs a short overvoltage, and a negative pulse simply reverses current through a symmetric load for a few microseconds.

An LED lamp contains a semiconductor string driven by a switching regulator, and every part of that sentence is a vulnerability. The LEDs are diodes with a forward characteristic that is exponential in voltage, so a modest overvoltage becomes a large current unless something regulates it. The regulator is a switching stage with an inductor, a switch and a control loop compensated for a particular input range. Before both sits an input filter, usually a capacitor, whose job is to smooth a 13.5 V rail and which is now asked to hold off 35 V, 65 V or 101 V depending on the condition being tested.

None of that is a defect in LED lighting. It is the normal consequence of replacing a self-regulating resistive load with a regulated electronic one, and it is why the same transient a sealed-beam halogen unit shrugged off can end a lamp bar’s life in one event. The relevant question for a buyer is how the input stage is protected, not whether the lamp “has protection”. The site’s explanation of driver topologies covers why a buck, boost or buck-boost stage behaves differently when the input rises, and that difference decides whether an overvoltage becomes a shutdown or a destroyed switch.

Failure mode one: the constant-current driver

The first failure mode is the driver itself, and it is worth separating the mechanisms because they produce different symptoms and different evidence.

An input capacitor rated for 25 V in a 12 V lamp has 10 V of margin over the charging system and none at all against a 35 V suppressed dump. When it breaks down it usually fails short, taking the rail down and blowing the fuse; the lamp is dead and the fuse is the only visible clue. A capacitor that fails open is worse, because the lamp keeps working with a degraded filter until the next transient.

A switching device that exceeds its rating fails in a way that depends on energy: a brief excursion may be absorbed by an avalanche rating, a long one will not be. TI’s measurements on its own reference design show the shape of the problem. With a 101 V unsuppressed dump at the input, the output of the protection stage peaked at 41.75 V and fell to 7.5 V, both inside the operating range of a 12 V system, which is what allowed the design to claim class A operation. That result came from an active stage with a MOSFET, a controller and a TVS, not from a single component.

The control loop is the third mechanism and the least visible. A constant-current regulator asked to hold current while its input collapses or spikes will either maintain regulation, run out of duty cycle and let current fall, or overshoot when the loop recovers. The last case produces the field report “it flickered and then a section went dark”, because the overshoot passes through strings that were already at rated current.

This is a mechanism argument rather than a measurement, and it should be labelled as one: the sources cited here describe the pulses and the clamp, not the inside of any particular lamp. What the argument does establish is which questions are worth asking, and those are listed below.

Failure mode two: the clamp, and how to size it

The second failure mode is the protection component, and here the sources are unusually specific. Littelfuse defines the terms a sizing calculation needs: reverse standoff voltage as the maximum peak voltage that may be applied with no significant current flow, breakdown voltage as the voltage at a test current typically of 1 mA, peak pulse power as IPP multiplied by VCL, and maximum clamping voltage as the voltage across the protector at the maximum peak pulse current.

The calculation is short. Peak pulse current is the peak transient voltage minus the clamping voltage, divided by the source impedance. Littelfuse’s worked example for 24 V uses a 202 V open-circuit peak, a 4 Ω alternator resistance and a 50 V target clamp, giving (202 − 50) / 4 = 38 A, below the 40.8 A the chosen device handles for ten pulses at 220 ms. TI reaches the same relationship from the other side, noting that peak power in the TVS depends on the clamping voltage, the peak transient voltage and the alternator resistance, and that parts must be selected on the energy dissipated at the peaks.

The counter-intuitive part, and the one that separates a competent design from a plausible one, is what happens when the clamp is set too low. Littelfuse states it directly: if a designer adds a TVS with a lower clamp voltage than the alternator’s own suppressor, the centralized suppressor is “by-passed (or shorted)” and the added TVS takes the energy. The waveform the lamp then sees is no longer the suppressed condition but the unsuppressed one, because the lamp’s own clamp has become the only clamp in the circuit. The recommendation follows: if the suppressed level US* falls inside the TVS protection range, choose a low-power TVS whose working voltage sits a little above US*, and it will also cover pulses 1, 2a, 3a and 3b. Only when US* is too high for the downstream electronics does the designer need the high-energy series, and then the whole dump energy has to be budgeted.

Design situation	Correct response	What goes wrong otherwise
Alternator has a working clamp at US*	TVS with working voltage slightly above US*, low power	A lower clamp bypasses the alternator suppressor and takes the whole dump
No working clamp on the alternator	High-energy TVS sized on US, Ri and td together	An undersized part fails short on the first event
One TVS cannot take the energy	Series string of equal-power parts, each with a third of the breakdown voltage	Parallel parts do not share current evenly
Operating temperature above 25 °C	Derate the pulse capability before choosing	The datasheet number is measured at room temperature

Littelfuse also documents how the part fails, and it is the sentence that makes a TVS defensible in a lamp rather than a fuse with extra steps: “the TVS Diode failure mechanism is a short circuit; if the TVS Diode fails due to a transient greater than the datasheet specification, the circuit will still be protected.” A shorted clamp takes the rail down and blows the fuse. The lamp stops working, but it does not pass the transient on to the driver and it does not start a fire inside a sealed housing.

Failure mode three: reverse polarity and jump starts

Reverse polarity is the failure a user can cause in a car park, and the series element that protects against it is also the element that sets the forward loss, as the reverse-voltage test above shows. Jump starting is the adjacent event a lamp owner is more likely to meet. TI lists jump start among the transients a battery-connected circuit must handle, and its stated reason for the whole exercise is worth repeating: during the life of a car the alternator may be replaced with a non-original part, and a replacement “may have different or no” suppression. That is the mechanism by which a lamp that survived its own vehicle’s dump can fail on a vehicle whose alternator was swapped.

Failure mode four: overvoltage lockout and the lamp that comes back

Not every transient kills a lamp, and the ones that do not are harder to diagnose. A driver with overvoltage protection shuts its output down when the input exceeds a threshold and restarts when the input returns to range, so a 400 ms event looks like a flicker rather than a fault. A driver with latching protection stays off until the supply is cycled, which produces the report “the bar went out and came back after I switched it off and on”: class D behaviour, and a different defect from a dead output stage.

The diagnostic consequence is that “it flickers” is not a useful symptom on its own. What separates a lockout from a failing driver is whether the lamp recovers without a power cycle, whether it recovers with one, and whether the event correlates with a load switching.

What the test bench actually looks like

The gap between a laboratory pass and a vehicle failure is easiest to understand by looking at how the test is set up, because several setup details are things a vehicle does not do.

AMETEK lists the conditions: ambient temperature of 23 ± 5 °C, a continuous source with an internal resistance below 0.01 Ω, and a test voltage of 13.5 V or 27 V. The generator must be connected directly to the reference ground plane, and for pulses 3a and 3b that connection must not be longer than 100 mm because of the frequencies involved. The device and its wiring sit typically 50 mm above the reference plane, and the oscilloscope is disconnected during the test, having been used only to verify the pulse at the generator output with no battery voltage applied.

Setup element	Test condition	Real vehicle difference
Ambient temperature	23 ± 5 °C	Engine bay and lamp housing are far hotter
Supply source impedance	below 0.01 Ω	Harness, connectors and fuses add resistance
Generator bonding	direct to the plane, 100 mm maximum for 3a and 3b	The lamp’s return is the chassis, at an unknown impedance
Device position	50 mm above the reference plane	Whatever the bracket and body allow
Instrumentation	oscilloscope disconnected during the test	Nothing monitors the lamp in service
Pulse verification	at the generator output, no battery voltage	The pulse at the lamp terminal is not measured

The last row is the one worth carrying into a conversation with a supplier. The test verifies the generator, not the lamp terminal. A lamp wired through a long harness, a relay and a switch sees a different waveform from the one the generator produced, and no part of the procedure claims otherwise.

The bench also has an alternative. AMETEK documents Annex F, which replaces the single pulses with composite waveforms A1 and A2 for inductive loads above 1 A and below 0.4 A, B1 and B2 for low-side switching with reversed polarity, and C for the secondary arc across switch contacts. Two laboratories can both claim ISO 7637-2 compliance while applying physically different stresses.

Why a laboratory pass is not a vehicle guarantee

A pass is evidence about a specimen under defined conditions, and the conditions are narrower than a vehicle. Three gaps account for most of the distance.

The first is the harness. The standard defines a wiring length and a ground reference; a vehicle defines neither. A lamp on a roof bar with a 4 m run to the battery, a relay in the cabin and a chassis return through a painted bracket is not the circuit that passed, and the difference appears nowhere in the report.

The second is the specimen. A qualification is normally run on a small number of samples, and a transient failure is statistical, driven by component tolerance. The capacitor that breaks down at 40 V in one unit may hold 45 V in the next, which is why a lamp that passed a 35 V suppressed dump is not thereby shown to survive an unsuppressed one. Durability testing covers the mechanical and environmental stresses that act on the same hardware; voltage transients are a separate axis and a separate report.

The third is the specification hierarchy. ISO 7637-2 and ISO 16750-2 are international standards, and manufacturers do not stop there. TI’s overview lists LV124, the group standard used by Audi, BMW, Porsche and Volkswagen among others, alongside SAE J1113-11 and JASO A-1, noting that specifications are not limited to that list and that although changes are typically in only a few parameters, “the essence of the requirements are the same”. Ford’s FMC 1278 goes further and writes its own dump immunity section. A supplier who tested to an ISO pulse may still be short of a specific customer’s requirement.

What a supplier claim can and cannot prove

This is where the article turns into something a buyer can use, and the rule is simple: a claim is checkable only if it names the things that changed the result.

Claim as written	What it actually establishes	What to ask for
“ISO 7637-2 compliant”	Nothing specific; the standard has seven pulse families and four levels	Which pulses, which levels, which edition
“Load dump protected”	Nothing sizably; two conditions and a range of source impedances	US, Ri, td, and whether the test was suppressed or unsuppressed
“12 V and 24 V compatible”	A statement about input rating, not about testing	The maximum continuous input voltage and the clamp level
“Reverse polarity protected”	Usually true, and usually a series element	The reverse voltage and duration tested, and the forward drop
“TVS protected”	That a component exists	Its standoff, breakdown and clamping voltages, and its pulse rating
“Tested to 100 V”	A number without a shape	The pulse, the source impedance, and the duration at that voltage

Two of these deserve a longer note. “Load dump protected” is the claim most often made and least often specified, and the reason is that the two conditions differ by a factor of nearly three in voltage. A lamp designed for a 35 V suppressed level is a different product from one designed for a 101 V unsuppressed excursion, and the second costs more because the clamp must absorb more energy.

“Tested to 100 V” is the claim that sounds most rigorous and is the least informative, because 100 V of pulse 1 at 10 Ω and 100 V of an unsuppressed dump at 0.5 Ω differ by a factor of twenty in current. The voltage is the part that fits on a badge; the impedance is the part that decides survival.

Buying and installing: the rules that follow

The engineering converts into decisions a buyer can act on without a laboratory.

Decision	What to do	Why
Circuit protection	Size the fuse to the conductor, not to the lamp	A transient that trips a fuse is one the lamp survived
Return path	Bond the lamp’s negative to the chassis with a short, clean connection	A long return adds impedance and shifts what the clamp sees
Relay placement	Keep the relay close to the battery and the lamp close to the relay	Loop area and harness length both feed the transient
Reverse connection	Assume it will happen during a service	The test condition is 14 V or 28 V for 60 s, not a momentary touch
Jump starting	Treat a jump start as an unsuppressed event on the donor side	The donor’s alternator may have no clamp
Alternator replacement	Ask what the replacement’s suppression is	TI names this as a real path to an unprotected dump
Suppressor voltage	Compare any added clamp with US* before adding it	A lower clamp becomes the only clamp and takes the energy

The last row reverses intuition, so it is worth restating outside the table. Adding a clamp to a circuit that already has a better one upstream does not add protection; it moves the energy. A lamp with a 24 V standoff TVS on a vehicle whose alternator clamps at 35 V will conduct through that TVS during every dump event, and it will fail on the first one carrying enough energy. The correct TVS for that vehicle is one whose standoff sits above 35 V, not below it.

The same reasoning applies to any lamp wired to a vehicle rail, from a single pod to a 40 inch slim bar, and the LED light bars category shows how differently the same questions are answered at different price points.

Two boundaries belong here as well. This article covers 12 V and 24 V systems, the scope of ISO 7637-2’s second part; 48 V systems, high-voltage traction systems and the shielded high-voltage tests in ISO/TS 7637-4 are separate documents, and a lamp rated for a 12 V vehicle should not be assumed to behave on any of them. And nothing here substitutes for a test report on the specific article number being purchased.

Frequently asked questions

Does a 12 V LED light bar need dump protection?

If the lamp is connected to the vehicle’s battery or charging circuit rather than to a separately regulated supply, then yes, because the dump reaches whatever sits on that rail. Whether it needs an unsuppressed or a suppressed rating depends on the alternator: a vehicle with a working clamp presents 35 V on a 12 V system, and one without presents up to 101 V behind 0.5 Ω.

What is the difference between ISO 7637-2 and ISO 16750-2?

ISO 7637-2 defines the fast conducted transients on the supply line and how to test immunity to them. ISO 16750-2 defines electrical loads, the slower supply-voltage conditions a device must operate through, and since the 2011 edition of ISO 7637-2 it also carries the cranking pulse and the dump pulse. For a lamp, the first is about surviving switching events and the second is about surviving the charging system.

Is 5a or 5b the suppressed pulse?

In the sources opened for this article, 5b is the suppressed condition and 5a the unsuppressed one. Littelfuse states it as a table row, AMETEK describes 5b as the same generator setting with a suppressor connected, and ST conditions its 5b row on an external clamp of 40 V maximum. Some summaries state the opposite, so the label alone is not evidence; the waveform and the clamp voltage are.

What test level should I ask a supplier for?

Ask for the level and the numbers. For pulse 1 on a 12 V system, level IV is −100 V through 10 Ω; for pulse 3b it is +100 V through 50 Ω; for the suppressed dump it is 65 V to 87 V through 2 Ω with an external clamp, or 35 V as the ISO 16750-2 suppressed level. A supplier who can quote those numbers has read a report; one who cannot has read a brochure.

Can a TVS diode alone protect an LED lamp?

It can protect the lamp against a defined pulse at a defined source impedance, which is what the standards specify, and it fails short when overstressed, which keeps the driver behind it alive. It cannot protect against energy it was not sized for, and it cannot compensate for an input capacitor rated below the clamp voltage. The TVS is the last element in a chain that also includes the fuse, the reverse-polarity device and the driver’s own input rating.

Does an E-mark or R10 approval cover voltage transients?

R10 covers transient immunity of an ESA at the levels in its Table 2, tested by the ISO 7637-2 method, and it lists lighting functions as immunity related. It does not cover the dump, and its Appendix 1 does not reference ISO 16750. An R10 approval is real evidence about pulses 1 to 4 and no evidence at all about the dump. The regulation and the EMC emission standard do different jobs, which is covered in the comparison of R10 and CISPR 25.

What does functional status A mean on a report?

Status A means all functions performed as designed during and after the disturbance. Status B allows a function to leave its tolerance during the event provided it recovers by itself; status C allows it to stop during the event provided it recovers by itself; status D requires operator action; status E means it did not recover. For a lamp, A and B are the useful results, and C is acceptable for a cranking dip only if the light coming back by itself suits the task.

Can a transient kill a lamp without blowing a fuse?

Yes, and it is a common outcome. A switching device or an LED string can fail open while input current stays below the fuse rating, and the lamp then draws nothing. The fuse protects the wiring, not the electronics, which is why a dead lamp with an intact fuse is a normal finding rather than evidence that no transient occurred.

Editorial note

Written by the SAEC Lighting Engineering Team, Foshan. Last updated: 2026-10-17. This article describes ISO 7637-2 and ISO 16750-2 as reported by the manufacturer application notes, datasheets and regulatory text listed under Sources. Both standard texts are sold rather than published and were not obtained for this article; where a source quotes a clause number or a parameter, the value is attributed to that source, and disagreements between sources are shown rather than resolved. No test report for any SAEC product is cited, and nothing here should be read as a claim that a particular lamp passed a particular test.

Sources

• Texas Instruments, Automotive 12- and 24-V Battery Input Protection Reference Design (TIDUC41A), 2016, revised 2017 — https://www.ti.com/lit/ug/tiduc41a/tiduc41a.pdf
• Texas Instruments, Automotive Reverse Polarity Protection Reference Design (TIDUC42), 2016 — https://www.ti.com/lit/ug/tiduc42/tiduc42.pdf
• Littelfuse, TVS Diodes to Meet Automotive Load Dump Standard, application note, 2020 — https://www.mouser.com.gt/datasheet/2/240/Littelfuse_TVS_Diode_Meet_Automotive_Load_Dump_Sta-1889011.pdf
• Littelfuse, Automotive Circuit Protection using Littelfuse Automotive TVS Diodes, application note, 2023 — https://www.littelfuse.com/assetdocs/tvs-diode-sad-silicon-avalanche-diode-application-note?assetguid=e4e5626b-a7b4-4098-8b9d-49424e2cfdcc
• STMicroelectronics, VND5E008MY-E double channel high-side driver datasheet, Doc ID 018699 Rev 3, 2012 — https://www1.futureelectronics.com/doc/STMicroelectronics/VND5E008MYTR-E.pdf
• AMETEK CTS, Automotive ISO 7637-2 seminar material — https://www.volta.it/wp-content/uploads/2025/10/Ametek-CTS-Italy-Automotiv-ISO-7637.pdf
• Regulation No 10 of the Economic Commission for Europe of the United Nations, Uniform provisions concerning the approval of vehicles with regard to electromagnetic compatibility, Official Journal of the European Union L 41, 17 February 2017 — https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:42017X0260
• Ford Motor Company, FMC1278 Electromagnetic Compatibility Specification, 4th edition, 2021 (copy hosted by TI E2E) — https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/6/_8F797972_EMC_0768C651_.pdf
• SAEC, UN R10 vs CISPR 25: Vehicle Lighting EMC Explained — https://saec-lights.com/standards-safety/un-r10-vs-cispr-25-vehicle-lighting-emc/
• SAEC, Automotive LED Lamp Durability Testing — https://saec-lights.com/standards-safety/automotive-led-lamp-durability-testing/
• SAEC, Automotive LED Drivers Explained: Buck, Boost and Buck-Boost Circuits — https://saec-lights.com/lighting-basics/automotive-led-driver-topologies/
• SAEC, LED Light Bars category — https://saec-lights.com/product-category/off-road-lights/led-light-bars/
• SAEC, Jungle Series 40 inch Slim Straight LED Light Bar — https://saec-lights.com/product/jungle-series-40-inch-slim-straight-led-light-bar/