How to Choose a Power Semiconductor Supplier in China
A five-point audit for buying power semiconductors from China - die source, packaging capacity, reliability evidence, engineering support, change control.
OCENME Engineering · Published November 15, 2026 · 6 min read
Choosing a power semiconductor supplier in China is a capability audit, not a price negotiation. Check five things — die source transparency, in-house packaging, reliability evidence, engineering response and change control. A supplier packaging 50 million units per month with an on-site reliability laboratory running JESD22 tests can answer a failure question in days, while a trader needs weeks.
Key points
- Die source transparency comes first — ask which foundry and which wafer generation a quote is based on
- In-house packaging capacity decides lead time and lot traceability, not just unit price
- A reliability claim without test conditions and sample size is not evidence
- A change notification agreement is as important as any electrical parameter
The cheapest quote is the wrong starting point. Choosing a power semiconductor supplier is a capability audit: five questions, asked in order, each of which eliminates a class of supplier before you spend money on samples. Get through all five with written answers and you have a supplier you can build a product on. Fail at any one and the price advantage will cost you more in field returns than it saved on the bill of materials.
This article gives you the five questions, what a competent answer looks like, and what each answer predicts about your experience over the following two years.
Why the usual checklist fails
Most supplier evaluations start and end with a price list and a certificate scan. Both are nearly useless for power semiconductors:
- Certificates describe a management system, not a device. A quality system certificate tells you the factory has procedures. It says nothing about whether the 650 V die in the quote is a current generation or a repackaged older one.
- Datasheet comparisons mislead when conditions differ. Two parts at the same headline RDS(on) can differ by a factor of two in gate charge — see how to read a MOSFET datasheet.
- Price lists hide the die. The same headline specification can be met with different die sizes and process generations, and that is the single biggest cost and reliability variable — what drives the price of a power MOSFET explains the mechanics.
So the audit has to go below paperwork. Here are the five questions.
The five-point audit
| # | Question to ask | A competent answer sounds like | An immediate red flag |
|---|---|---|---|
| 1 | Who makes the die, and which generation? | Names the wafer source and process node, states whether the die is their own design or a licensed one | A vague answer about mature supply chains, or refusal to say whether the die is self-designed |
| 2 | Where does packaging and test happen? | Own packaging line, stated monthly capacity, own test programme with 100% testing of critical parameters | Assembly outsourced with no line audit offered, or test limited to a final functional check |
| 3 | What reliability evidence exists for my part number? | Named reports — HTRB, H3TRB, temperature cycling — with duration, sample size and read points, for the exact ordering code | Family-level claims, or reports that cannot be shared because they belong to someone else |
| 4 | Who answers my technical question, and how fast? | A named application engineer who replies with calculations or curves at your operating point, typically within days | Sales-only contact, or answers that quote the datasheet back at you |
| 5 | What happens when you change something? | A written change notification agreement covering die revision, mould compound, lead frame and assembly site | A promise that nothing ever changes, or silence on the question |
Two of these deserve expansion, because they are the ones buyers most often skip.
Die source transparency (question 1). You are not asking for secrets — wafer foundry relationships are normal and openly discussed in this industry. You are asking whether the supplier can tell you what is inside the package they ship. A supplier that designs and packages but sources wafers from a named foundry can still answer precisely. One that cannot name its die source cannot control what arrives in six months, and neither can you.
Change control (question 5). An unnotified die change is the most common reason a field failure cannot be reproduced on the original sample. The mechanism is described in why power supplies fail in the field — the part number stays the same while the behaviour changes. If a supplier will not sign a notification agreement before you place the first order, they have told you what the next two years look like.
What in-house packaging actually buys you
Packaging capacity is not just a lead-time number. A supplier with its own assembly and test line — for example, one packaging 50 million units per month — controls three things a reseller cannot:
- Lot traceability. The date code on the reel maps back to a wafer lot and an assembly batch, so when one lot misbehaves the rest of it can be contained instead of discovered.
- Screening depth. In-house test means parameters like leakage or gate threshold can be 100% screened on every unit, not sampled by an external subcontractor you will never audit.
- Failure analysis turnaround. When you return a failed unit, an in-house laboratory with X-ray and acoustic microscopy can open the part and tell you what happened in days. A supplier without these tools sends the unit to a third party and comes back in three weeks with a less specific answer.
The same logic applies at the product level. A supplier that can discuss the trade-offs in its own high-voltage MOSFET range or explain where a SiC MOSFET needs negative gate bias has engineers reading its own test data. One that only forwards PDFs does not.
When to use it — and when not to
Use the five-point audit when you are selecting a supplier for a design-in you intend to ship for several years, qualifying a second source for an existing product, or moving volume away from a distributor-only channel to a direct factory relationship. It is also the right framework for any order above roughly ten thousand units per month, where your own qualification effort is large enough that supplier stability dominates the economics.
Do not use it when you are buying a few hundred pieces of a standard part for a prototype, filling a one-time gap in distribution stock, or sourcing through a franchise distributor of a major international brand. In those cases the supplier is not being asked to stand behind the design for years, and the audit overhead exceeds the risk it manages. Buy on availability and price, and note that the part was procured opportunistically in your own records.
Limits and scope
This audit addresses the supplier relationship, not the device itself. It does not replace board-level qualification — a supplier that passes all five questions can still ship a part that misbehaves in your layout, which is what second-source qualification is for. It also does not apply to automotive-grade procurement with its own formal framework, and it does not certify that any specific supplier passes: the point of the questions is that you run them and read the answers yourself. Finally, certificate lists vary by company and change over time — always ask a supplier for its current documentation rather than assuming it from this article.
This article describes general manufacturing practice for machined and fabricated components. It is not a substitute for a drawing-specific engineering review — tolerances, material selection and process capability depend on the individual part.