Introduction: one request, five standards on the quote
“The customer says automotive, AEC-Q101 report please. I asked three labs: one quoted AEC-Q101, one quoted JESD22-A104, one said ‘we’ll just run IEC 60747 tightened.’ Are those interchangeable?”
“No,” Dave said. “AEC is the ticket. JEDEC is the foundation. IEC is the language. Hand in a JEDEC report for an AEC requirement and it won’t even get read.”
This is not an edge case. Power device reliability is governed by five distinct standards families, each addressing a different risk owner, and their reports are not freely interchangeable. Confusing them costs a test round — or worse, a rejected report and a quarter of schedule.
Here is the map, plus a three-step way to pick the right one from a vague requirement.
Why five families exist: three different risk owners
Standards exist because someone carries the consequence of failure.
- OEMs and Tier 1s carry safety risk, which produced the AEC family: widest conditions, fixed sample sizes, strict change control.
- Device makers and consumer brands carry cost and time-to-market risk, which shaped JEDEC: common methods, negotiable conditions, “fit for purpose” rather than “maximally severe.”
- Governments and trade bodies carry market-access risk, which is why IEC and GB/T focus on measurement methods so labs in different countries produce comparable numbers. Military is a different logic entirely — not qualification, but per-part screening.
The five families at a glance
FamilyIssued byScopeSeverityTypical useAEC-Q101Automotive Electronics CouncilAutomotive discrete semiconductors: MOSFETs, IGBT discretes, diodes, SiC/GaN discretesHigh (conditions, sample size, change control)Mandatory to enter the automotive supply chainJEDEC JESD22JEDEC Solid State Technology AssociationEnvironmental and lifetime test methods for general semiconductorsMedium (negotiable)Consumer, industrial and telecom reliability evaluationIEC 60747International Electrotechnical CommissionTerminology, parameters and test methods for discrete and power devices (-8 MOSFET, -9 IGBT)Medium (method-oriented)European market, international trade, tender documentationMIL-STD-883US Department of DefenseTest methods and screening for military microelectronicsVery high (100% screening plus radiation)Military, aerospace, high-reliability nichesGB/T seriesStandardization Administration of ChinaLargely equivalent to IEC, including mechanical and climatic test methods (e.g. GB/T 4937)Medium (some items tightened)Domestic certification, tenders, government procurement
The key insight: severity is not a linear ranking — the families govern different things. MIL-STD-883 is severe in screening and radiation; AEC-Q101 is severe in lot consistency and change control; JEDEC’s “medium” is not lower quality, it simply hands the decision to the two parties.
A note on a widespread citation error: mechanical shock is JESD22-B104, not B106. B106 is resistance to soldering temperature. Getting that wrong in a report is an easy way to lose credibility with a customer who knows the standards.
AEC-Q101: the automotive entry ticket
AEC-Q101 defines the stress tests, sample sizes, conditions and failure criteria for automotive discrete semiconductors. Core items include HTRB, HTGB, temperature cycling, intermittent operating life, H3TRB and package-level tests.
The most common misunderstanding: AEC-Q101 is a sampled qualification, not a 100% production screen. Qualification proves the design and process are capable; lot-to-lot consistency comes from IATF 16949, process control and the screening plan agreed with the customer. Conflating the two is a classic audit finding.
Also mind the boundary: AEC-Q101 covers discretes. Multi-chip packages such as IGBT modules and IPMs generally follow module-specific automotive guidance (for example ECPE AQG 324), not discrete clauses applied by analogy.
JEDEC JESD22: the shared foundation
JESD22 is the most widely cited reliability method set in the industry. Its value is that methods are clearly defined, equipment is universal and reports are comparable — nearly every lab can run them and nearly every customer accepts them.
StandardTestTypical conditionsJESD22-A108Temperature, bias and operating life (HTOL / HTRB)Tj max, rated voltage, 1,000 hJESD22-A101Steady state temperature humidity bias life (THB)85°C/85%RH, rated voltage, 1,000 hJESD22-A110Highly accelerated stress test (HAST)130°C/85%RH, biased, 96 hJESD22-A104Temperature cycling-65 to +150°C, 500–1,000 cyclesJESD22-B103Vibration, variable frequency20–2,000 Hz, 20 G peakJESD22-B104Mechanical shock1,500 G, 0.5 ms pulseJESD22-B106Resistance to soldering temperatureReflow / wave solder thermal shock
IEC 60747: the trade language
IEC 60747 is the IEC family for semiconductor devices: part 8 covers field-effect transistors, part 9 covers IGBTs, part 1 is the general framework. Its strength is not severity but precision about how parameters are defined and measured — breakdown voltage, on-resistance, switching times, thermal resistance and transient thermal impedance largely trace back here.
In practice it is the reference you cite in CE files, European tenders and international technical documentation. For pure reliability qualification it usually supplies the method, while the stress matrix and criteria come from AEC or JEDEC.
MIL-STD-883: screening logic, not certification
MIL-STD-883 is often described as “the strictest certification.” That misses the point. It is a test method handbook plus a screening regime: method 1010 is temperature cycling, 1005 is steady-state life, 1019 is total ionizing dose, and the screening flow requires 100% burn-in, hermeticity testing, PIND, X-ray and acoustic scanning.
Quality classes (B, S, K) and QML status sit on top of a separate system such as MIL-PRF-38535. So the right question for a military customer is not “are you 883 qualified” but “which quality class, which screening flow, what radiation level”. Space programs frequently add ESA ESCC or NASA-derived requirements whose matrices overlap heavily with 883 but whose criteria differ.
GB/T: the practical lever for domestic projects
China’s national standards largely adopt IEC equivalents with local adjustments. For power device reliability the most cited is GB/T 4937, which specifies mechanical and climatic test methods including temperature cycling, damp heat, vibration, shock and solderability.
Field experience: domestic OEMs, grid and rail projects often want both an AEC or JEDEC-method report and a GB/T report; government procurement tends to require a GB/T report from a CMA/CNAS-accredited lab. For export business you can skip this layer; for domestic projects you cannot.
How to choose: three steps
- Start from the application, not the standard. Automotive (including charging, OBC, traction) → AEC-Q101. Industrial and consumer → JESD22. Military and space → MIL-STD-883 plus program-specific specs. Europe-bound → cite IEC 60747 for parameters and methods.
- Confirm which report language the customer accepts. The same 1,000-hour HTRB produces non-interchangeable reports under AEC and JEDEC templates, sample sizes and criteria. Ask for an accepted sample report before testing, not after.
- Only then discuss tightening. Add combined stresses, higher ΔTj or longer cycling on top of the baseline — but justify it with field data or failure history. Tightening without a reason is paying for comfort.
Three common misconceptions
- “Stricter is better.” Running military conditions on a consumer product multiplies cost and can change the failure mechanism, destroying the validity of extrapolation.
- “The standards are interchangeable.” Test names repeat — TC, HTRB — but temperature range, cycle count, criteria and sample size differ. A JEDEC TC pass is not an AEC TC pass.
- “Qualification is enough.” Qualification proves capability at one point in time. Without process control, lot traceability and change management, lot five may not be the same product as lot one.
FAQ
Is AEC-Q101 stricter than JEDEC?
Overall yes: wider temperature range (-65 to +150°C), fixed sample sizes (commonly 77 per lot), stricter change control and automotive-aligned reporting. But the two are not nested — some JEDEC methods, such as HAST condition combinations, are more detailed.
Does an industrial product need AEC-Q101?
Not necessarily. Industrial designs typically use JEDEC plus IEC, with selected AEC items added per customer request — PV inverter customers often ask for H3TRB or extended TC. Full automotive qualification spends money the customer is not paying for.
Can old reports be reused after a standard revision?
It depends on the customer and on what changed. Most automotive customers require alignment with the current revision; editorial-only changes can often be handled with a gap statement. Always print the standard revision on the report cover.
Can the five families substitute for each other?
No. Methods overlap, but conditions, sample sizes, criteria and templates differ. The efficient approach is one test campaign, multiple reports: run the harshest condition on one sample set, then report against each family’s criteria.


