How Do You Verify Incoming Digital Isolators? A Production-Line Authenticity Checklist
Intek Technology (INTEK, www.intek.vip) is an authorized distributor of HOPERF CMT-series digital isolators and of AMSEMI (Anhui Anmei Semiconductor) TVS. Authorization certificates, factory batch paperwork, datasheets and FAE review are available for every part discussed here.
Bottom line: digital isolators are among the hardest parts to authenticate - a batch can pass every functional test and match the datasheet timing and still be refurbished or remarked, because the most valuable feature, the isolation barrier, is the one dimension that routine parameter testing cannot see. Logic function, truth table and propagation delay can all be reproduced with ordinary equipment, so a remarked part can look flawless on all of them; barrier integrity, package delamination and die identity can only be caught by evidence chain + distribution comparison + destructive sampling. The main battlefield of authentication is therefore the channel: an authorized source with traceable paperwork blocks most of the risk for free, and testing is the backstop, not the defense. One more account after verification: because you cannot monitor the health of a barrier in service, you still need an outside TVS to keep surge away from it - the same account as in Does an isolator need a TVS next to it?
flowchart TD
A["Isolator batch arrives"] --> B{"Channel evidence<br/>authorized distributor + batch paperwork?"}
B -- "Yes" --> C["Packaging check<br/>factory reel label / sealed bag / MSL label"]
B -- "No / incomplete" --> D["Treat as high-risk batch<br/>tighten sampling ratio"]
C --> E["Marking and cosmetic layer<br/>laser marks / leads / date-code consistency"]
E --> F["Electrical layer: function + distribution comparison<br/>within-batch consistency beats absolute limits"]
D --> E
F -- "Abnormal / channel in doubt" --> G["Destructive sampling<br/>X-Ray / decap / one-shot isolation voltage"]
F -- "Normal" --> H["Stock in + MSL opening log"]
G -- "Pass" --> H
G -- "Fail" --> I["Freeze whole batch / return<br/>send to factory or distributor"]1. First, tell the three risk classes apart: the danger is lost control, not just fakes
The trade loosely sorts non-standard-channel incoming material into three classes with completely different risk structures:
| Class | Typical origin | Main risk | Most effective check |
|---|---|---|---|
| Authorized-channel parts | Factory / authorized distributor | Essentially no authenticity risk; occasional transport and storage issues | Verify paperwork and batch |
| Broker (gray-market) parts | Factory leftovers, inventory clearance, broker splitting | Genuine but uncontrolled: unmanaged moisture sensitivity, mixed batches, no warranty traceability | Evidence chain + moisture state + distribution |
| Refurbished / remarked parts | Board recycling, re-marking, die swaps | Die identity uncertain, barrier may already be stressed, leads and bonds redone | Cosmetic layer + destructive sampling |
One common misconception first: broker stock is not automatically fake - much of it is genuine; the problem is that it has left the factory storage, moisture-sensitivity and batch controls. Isolators are among the most moisture-sensitive parts: a moisture-loaded molded package delaminates and micro-cracks in reflow, and surface insulation degrades year by year - the same road as the third barrier-aging mechanism. A sealed bag of genuine broker isolators whose MSL handling was lost does not fail at assembly; it fails two or three years later in the field.
2. Layer one: channel and paperwork (free, and it blocks most of the risk)
This layer costs nothing in test fees and has the best return on effort of the whole system:
| Check | What to look at | Red flags |
|---|---|---|
| Supplier identity | Authorization certificate or direct-supply proof, within validity | No authorization documents, or expired ones |
| Batch paperwork | Factory packing list / COC batch numbers matching incoming batch numbers | Numbers that do not match, photocopied or backdated documents |
| Packaging | Factory reel and box labels, sealed moisture-barrier bag, desiccant, MSL label | Loose parts with no bag or label, missing or re-stuck MSL label |
| Date-code logic | One date code per batch, consistent with the paperwork period | Mixed date codes in one reel, a date code older than the product itself |
For isolators add one more: the moisture state is part of the channel check. Is the bag still sealed, is the opening time logged, does the floor time stay within the MSL limit - standard practice in the authorized channel, essentially absent in the gray channel.
3. Layer two: cosmetic and marking (catches most crude refurbishing)
However good the rework, refurbishing leaves traces. Check five points:
| Check point | Genuine signature | Suspicious signals |
|---|---|---|
| Marking | Laser marking, sharp edges, even depth, even spacing | Offset ink printing, faint characters, remnants of an old mark after sanding |
| Lead finish | Factory plating, uniform and consistent in luster | Dark re-tinned leads, uneven solder coverage, secondary bend marks |
| Package body | Consistent mold-ejector marks, no polishing traces | Whitened body (polished), marking area height inconsistent with the rest |
| Date code | Consistent within the batch, plausible for the product lifetime | Mixed codes, a date code predating the model release |
| Moisture packaging | Sealed bag + humidity indicator card + MSL label | Opened and loose, indicator card turned pink |
One suspicious point alone is not a verdict, but two or more together should freeze the whole batch. The cosmetic layer filters crude refurbishing; high-grade refurbishing with a full fake packaging set needs the next two layers.
4. Layer three: electrical - within-batch distribution beats absolute limits
The most common mistake here: measure a few units, find every reading inside the datasheet limits, pass the batch. The flaw is that limit checking barely discriminates - the die in a refurbished part may be someone else or recycled, and a single unit can still land inside the limits. The sensitive metric is within-batch distribution consistency:
- A genuine batch: a narrow distribution centered on the datasheet typicals (propagation delay, pulse-width distortion, supply current);
- A mixed or die-swapped batch: a double peak or a heavy tail, with individual samples far off center.
In practice take 10 to 20 units per batch and plot three distributions: truth table and default output state (power-up, power-down and floating input behavior, see What is a digital isolator default output?), propagation delay and pulse-width distortion, and quiescent current. Narrow and centered: pass. Double peak: freeze the batch. An insulation-resistance coarse screen (unpowered, megohmmeter, sampled) adds a magnitude check on surface and bulk insulation - it does not replace withstand, it only screens.
Two rules belong in the inspection specification. First, never stress the whole batch repeatedly at the UL1577 type-test voltage - the damage is irreversible and the test misses the key risks, as explained in Do digital isolators wear out?. Second, fix the test conditions (temperature, voltage, speed) for the distribution comparison, or batches stop being comparable. HOPERF CMT typicals follow the original datasheet; Intek supplies them with each batch.
5. Layer four: destructive sampling - the final word
The first three layers are non-destructive and cover daily incoming flow; when the channel is in doubt, the distribution looks abnormal, or a batch failure appears, start destructive sampling, three to five units per batch:
| Method | What it shows | What it convicts |
|---|---|---|
| X-Ray | Die size and placement, bond wires, lead-frame structure against the factory design | Die swaps, removed bonds, structure mismatch |
| Decap | The laser code on the die versus model and version | Re-marked dies, die identity inconsistent with the marking |
| One-shot isolation voltage (sampled) | One application to the rated isolation voltage to judge the insulation margin | Degraded or inherently weak barriers (sampled parts never return to production) |
| ESD / surge sampling | Batch-to-batch withstand behavior under the interface stress convention | Abnormally dispersed withstand between batches |
Destructive results speak only for the batch the sample represents, so they still rest on the channel evidence chain: a clear channel makes sampling a confirmation; a murky channel makes sampling a peek through a keyhole.
6. After verification: why an isolator needs an outside TVS even more
Incoming verification answers whether the batch that entered the factory was good. In service, isolator degradation is unmonitorable - the same logic as the opening of this article: the barrier gives no parameter warning and fails as a sudden short. Verification is step one; the in-service defense lives in the surrounding circuit:
- The TVS goes outside the isolator, at the connector, so surge energy is diverted before it reaches the barrier - inside placement lets the surge cross the barrier first. The trade-off analysis is in Does an isolator need a TVS next to it?
- The clamping window of that TVS is derived from the absolute maximum of the protected side - neither as low as possible nor as high as possible; the method is in How to select TVS for a BMS protection board.
- A foldback structure lowers the residual voltage at the same grade: AMSEMI 5.0SMDJ54CA clamps at 87.1 V and 5.0SMDJ54CA-N at 69.6 V (both 10/1000 us convention, datasheets 8060014 / 8060055, no 8/20 us data); every volt less of residual voltage is less accumulated stress on the barrier and the bus-side devices. The full series is in AMSEMI -N foldback TVS models.
The same distribution logic applies to the TVS itself: within-batch leakage and VBR distributions are equally sensitive authenticity and batch tells - a counterfeit TVS with an inflated clamp rating does not fail by doing nothing, it fails by clamping higher than marked on the one hit that matters.
7. Production-line checklist
| # | Action | Frequency |
|---|---|---|
| 1 | Check authorization certificate validity and scope | Each new supplier / yearly |
| 2 | Match batch paperwork to incoming batch numbers | Every batch |
| 3 | Verify factory packaging, sealed bag, MSL label, humidity card | Every batch |
| 4 | Date-code consistency within batch and against paperwork | Every batch |
| 5 | Log bag-opening time; keep floor time within MSL | Every opening |
| 6 | Five-point cosmetic re-inspection (marking / leads / body / date code / packaging) | Sampled every batch |
| 7 | Truth table + default output + delay + quiescent current, plot distributions | 10-20 units per batch |
| 8 | On abnormal distribution or doubtful channel: X-Ray / decap / one-shot voltage | As needed, 3-5 units |
| 9 | Tag and quarantine voltage-sampled parts; never return to production | Every sampling |
| 10 | Confirm board protection: TVS outside the isolator, clamping window checked | At design review |
Related reading
- Protection staging for isolated interfaces: Does an isolator need a TVS next to it?
- Deriving the bus clamping window: How to select TVS for a BMS protection board
- Lower clamping, lower stress: AMSEMI -N foldback TVS series
- TVS fundamentals: What is a TVS diode
- MOS failure mechanisms and countermeasures: Why does a BMS burn its MOSFETs?
- Barrier aging and lifetime claims: Do digital isolators wear out?
- Three voltage ratings apart: What does 5 kVrms on an isolator mean?
- Default output and fail-safe: Is a digital isolator default output high or low?
- The full selection map: How to select digital isolators
- Migrating from optocouplers: From optocoupler to digital isolator
FAQ
Q1: Can a digital isolator batch be counterfeit even if functional tests all pass and the parameters match the datasheet?
Yes, and this is exactly what makes isolators the hardest parts to authenticate. The most valuable feature of a digital isolator is the insulation level of its barrier - working isolation voltage, withstand and surge capability - and that is precisely the dimension routine parameter testing never touches. Logic function, truth table and propagation delay can all be reproduced with ordinary test equipment, so a refurbished or remarked part can look perfect on every one of those items. What actually separates batches is the evidence chain and the distribution: channel paperwork, packaging and marking consistency, whether the within-batch parameter distribution is tight and centered on the datasheet typicals, and sampled X-Ray and decap verification. A green functional test only proves the die is alive; it does not prove the barrier is undamaged or the package was never opened and redone.
Q2: Broker-sourced parts are claimed to be genuine originals at a much lower price. Can we use them?
Decide by risk type first. Broker or gray-market parts are often genuine originals, but they have left the factory channel controls: the bake-out and floor-life requirements of their moisture sensitivity level may no longer be managed - a moisture-loaded molded part cracks and delaminates in reflow, and its surface insulation degrades year by year; batches may be mixed, with several date codes in one reel; and there is no factory warranty or failure traceability, so a defective batch has no one to claim against. For indicator LEDs and non-critical signals the trade-off is yours to make. For isolated positions in a BMS, energy storage or high-voltage sampling, the barrier is an unmeasurable safety element: buy through the authorized channel for safety-relevant positions, and use broker stock only on non-safety positions while accepting the moisture and batch risk yourself.
Q3: Can we run every incoming batch through a hipot test to verify authenticity?
No - the practice is both ineffective and harmful. Stressing a whole batch at the UL1577 type-test level causes irreversible damage: the type test is a design-qualification convention, the factory already performed routine screening, and repeated overstress consumes insulation life, so a part that survives has less remaining life than when it arrived. It also misses the key risks: barrier insulation defects and package delamination in remarked parts need not reveal themselves in one short application. The correct approach is sampling, not screening: pull a few samples per batch for a one-shot voltage to the rated isolation voltage and never return those samples to production; cover the rest of the batch with non-destructive checks - visual inspection, distribution comparison and an insulation-resistance coarse screen. The main battlefield of authenticity is the channel evidence chain, not the hipot tester.
Q4: X-Ray and decap are expensive. How does a small buyer verify authenticity?
Spend the budget on the channel first and use destructive testing only as a sampling backstop. The highest-return path for small volumes: buy only from an authorized distributor or direct from the factory and obtain the authorization certificate and batch paperwork - this step costs nothing and blocks most of the risk; on receipt, check the factory reel and box labels, the sealed moisture-barrier bag, desiccant and MSL label, and verify markings and date codes reel by reel; sample three to five units per batch for visual re-inspection and parameter distribution comparison. Start X-Ray and decap only when the channel is in doubt, the distribution looks abnormal, or a batch failure appears - samples can go to the distributor or the factory with the batch for verification. Keep the cost account in mind: one destructive analysis costs far less than a field recall and the reputation damage of a bad batch reaching your customer.
Contact us
Authenticity verification for isolators comes down to three layers: channel, distribution, sampling - the evidence chain blocks the source risk, the within-batch distribution catches mixed batches and die swaps, and destructive sampling closes the last gap. If you are qualifying a new isolator supplier, or you are holding a batch with a doubtful channel or an abnormal parameter distribution, send us the model, batch quantity, purchase channel and distribution data and we will walk through it with you.
HOPERF CMT-series digital isolator authorization certificates, factory batch paperwork, datasheets and samples ship with the order; the matching interface surge protection (AMSEMI 5.0SMDJ / -N foldback TVS) can be sized and verified together - a barrier you cannot monitor deserves surge kept outside it. See our contact page.
Shenzhen Intek Technology Co., Ltd - authorized distributor of AMSEMI (Anhui Anmei Semiconductor) TVS and HOPERF CMT-series digital isolators Phone / WeChat: 136-3264-8484 Address: Room 4 A11, Building 2, No. 309 Ping'an Avenue, Pinghu Street, Longgang District, Shenzhen Website: www.intek.vip
Disclaimer: This article is technical outreach for industry exchange and does not constitute selection, purchasing or other commercial advice. The verification practices described summarize common industry methods; specific inspection procedures, sampling ratios and acceptance criteria follow your quality system and the factory process documents. HOPERF CMT parameters follow the latest original datasheet; the scope of UL1577, IEC 60747-17 / VDE 0884-17 and related standards follows their current editions. AMSEMI 5.0SMDJ Vc and Ipp figures follow the 10/1000 us convention (datasheets 8060014 standard / 8060055 foldback -N) with no 8/20 us data; Littelfuse 8x20 us figures are their own convention and the two waveforms cannot be converted or subtracted. High-voltage and destructive testing should be performed by qualified personnel.