How Do You Verify the TVS on a BMS Protection Board? What to Test on the Bench and How to Judge Pass or Fail
One-line takeaway: bench verification must answer not "did the TVS survive" but three questions — (1) at the surge current you actually have to block, is the true clamping voltage of the TVS below the withstand ceiling of the downstream device; (2) are its leakage and temperature rise at the maximum steady-state voltage and high ambient acceptable; (3) does it drift after repeated surges. A datasheet only guarantees clamping voltage at one current point, one waveform, a single pulse and room temperature, while a real surge differs in waveform, amplitude, repetition and starting temperature. So the part must be measured on the board, and the result must be judged against criteria — "we hit it a few times and nothing blew" is not a pass.
1. Why datasheets that "line up" still need measurement
Everyone checks Vc at selection time. The problem is that the datasheet Vc is a point, not a curve.
| Aspect | What the datasheet guarantees | What the real world delivers |
|---|---|---|
| Current | Vc specified at the single Ipp point only | Surge current from tens to hundreds of amps; the operating point is elsewhere |
| Waveform | One convention only (AMSEMI 5.0SMDJ: 10/1000us) | Customer specs and whole-machine tests may use 8/20us, 10/1000us or a combination wave |
| Repetition | Single, non-repetitive pulse | Over-current protection trips and contact chatter produce repeated hits |
| Temperature | Test conditions TA = 25 C, 0.6 x 0.6 inch copper pad | Battery compartment at 60-70 C, with less ideal pads and traces |
Clamping voltage climbs with current along the dynamic resistance — the more current, the higher the clamp. The datasheet only guarantees the calibrated point; beyond it, nothing is promised. "Datasheet Vc below the MOSFET rating" therefore does not mean "below it during a real surge" — the gap in between can only be measured.
Convention reminder: for the AMSEMI 5.0SMDJ series (document 8060014 standard, 8060055 for the -N foldback) the Vc and Ipp figures are exclusively 10/1000us; there is no 8/20us data in the datasheet. Littelfuse's published 8x20us figures (for example 273.8 A at the equivalent grade) are Littelfuse's own convention for Littelfuse parts and must be attributed to Littelfuse whenever quoted. The same part shows a completely different Vc under the two waveforms, so subtracting one from the other or converting between them is forbidden — see What Does a 5000 W TVS Rating Actually Mean?.
2. The four questions bench verification must answer
Split "verification" apart and it is really four questions. Each maps to one test class; miss one and the job is not done.
| Test objective | How to test | Pass criterion |
|---|---|---|
| Is the clamp low enough | Current sweep, log the Vc-versus-current curve across the TVS | At the current you have to block, Vc is no more than 0.85 x the measured typical withstand of the downstream MOSFET |
| Does it get in the way | Apply the maximum steady-state voltage at high ambient, measure leakage and device temperature rise | Leakage-induced error stays inside the accuracy budget; temperature rise stays inside the derating line |
| Can one hit kill it | Apply one to many surges, re-measure VRWM and Vc before and after | Parameter drift stays within allowance and the failure mode is predictable (short circuit dominates) |
| What does the downstream actually see | Measure directly across the MOSFET drain-source | Peak at the MOSFET stays below its no-avalanche ceiling (0.85 x measured typical withstand; 0.8 for automotive) |
The last question is the one most often skipped: the TVS clamping does not mean it clamps at the MOSFET. Trace inductance adds L di/dt on top of the clamp voltage, so the further the TVS sits from the MOSFET and the larger the loop, the higher the peak the MOSFET actually sees.
3. Verification flow
flowchart TD
A["Define the protected device: MOSFET part\n+ parallel count + surge waveform and current class"] --> B["First check the three-stage voltage window\nVRWM / VBR / Vc"]
B -- "Window does not exist" --> C["Change the selection or topology first\n(raise MOS rating / use low-clamping TVS)\nDo not rush to the bench"]
B -- "Window exists" --> D["Build the board: TVS right at the\nMOSFET drain-source, short wide traces"]
D --> E["Current sweep: measure the Vc-versus-current\ncurve and compare with the datasheet Ipp point"]
E --> F{"At the target current,\nVc no more than 0.85 x measured MOS withstand?"}
F -- "No" --> G["Use a lower clamp grade / raise MOS rating\n(foldback -N pulls Vc down about 20 percent)"]
F -- "Yes" --> H["Measure the actual drain-source peak at the MOSFET\nand confirm trace inductance does not push it up"]
H --> I["Long-term energised at high ambient:\nleakage and temperature rise"]
I --> J["Single to repeated surges, re-measure drift\nand confirm the failure mode"]
J --> K["Issue the report: waveform, current point, Vc,\nMOSFET peak, thermal drift and parameter drift"]4. How to run each of the four test classes
Class 1: clamping voltage curve. Do not fire a single point. Sweep several current levels (for example 50 percent to 150 percent of the target) and log Vc at each, then plot the Vc-versus-current curve and mark the datasheet Ipp point on it. The slope is the order of magnitude of the dynamic resistance and tells you where the clamp climbs if the current doubles.
Class 2: leakage and temperature rise. Hold the device at the maximum steady-state bus voltage for a long period and measure reverse leakage at room temperature and at 60-70 C, together with case or solder-joint temperature rise. Leakage roughly doubles every 10 C, so climbing from microamps to tens or hundreds of microamps in a hot enclosure is normal; the question is whether it breaks the accuracy budget or self-heats. High-impedance nodes such as sampling lines must be checked here — see Should BMS Cell-Sampling Lines Have a TVS?.
Class 3: repeated surges and parameter drift. Passing once is not reliability. Fire the customer-specified number of hits (tens, hundreds) and re-measure VRWM and Vc at the same current point every so often, looking for monotonic drift. The typical TVS failure mode is a short circuit, so the drift phase is usually accompanied by a clear rise in leakage — the earliest signal worth catching.
Class 4: system-level verification (measure at the MOSFET). The first three classes measure across the TVS; this one must measure across the downstream device. Trace inductance between the TVS and the MOSFET raises the peak at the MOSFET, and paralleled devices share the energy unevenly. The criterion is the measured peak at the MOSFET against its no-avalanche ceiling, not a good reading at the TVS terminals.
Note: a power MOSFET rated voltage is a minimum (a 100 V part is only guaranteed not to be below 100 V), and production typicals run about 10 percent higher, so a 100 V part often measures around 110 V. State which convention is used in the criteria — see TVS Fitted, MOSFETs Still Blow: The 23S / 24S Voltage Budget.
5. Choosing the injection waveform
The most common bench error is not the equipment but the waveform convention. Fix three things before testing: which waveform, what current class, how many shots.
| Waveform | Typical use | Convention ownership |
|---|---|---|
| 10/1000us | Component-level parameter verification, matching the TVS datasheet convention | AMSEMI 5.0SMDJ and -N series datasheets (8060014 / 8060055) |
| 8/20us | Whole-machine surge immunity testing, current wave | The current wave of the IEC 61000-4-5 combination wave; Littelfuse public data uses this convention |
| 1.2/50us + 8/20us combination wave | Injected at the system immunity level (voltage wave plus current wave) | The IEC 61000-4-5 standard combination wave |
| Customer or OEM specification | Waveform and repetition specified for the actual environment | The customer specification document |
Why the separation matters: the same part can take nearly an order of magnitude more peak current at 8/20us than at 10/1000us. Judging a 10/1000us calibrated clamp curve with an "8/20us handles hundreds of amps" figure, or the reverse, produces the wrong conclusion. Use the datasheet waveform for component-level verification and the standard or customer waveform for system-level acceptance, and report the two results separately rather than in one table.
6. How to set the criteria (four conditions; missing one is not a pass)
| Condition | Requirement | Basis |
|---|---|---|
| Clamp gate | At the target current, Vc no more than 0.85 x the measured typical downstream withstand (0.8 automotive) | Upper gate of the three-stage voltage window |
| Start gate | VBR minimum no more than 0.85 x the measured typical withstand | Middle gate of the window, the one skipped most often |
| Leakage gate | IR at maximum steady-state voltage and high ambient does not break the accuracy budget or cause self-heating | Device IR specification plus system accuracy budget |
| Drift gate | After the specified number of surges, VRWM and Vc drift within allowance and the failure mode is predictable | Customer shot count plus before-and-after re-measurement |
Of the four, the start gate and the drift gate are the ones most easily hidden by "it did not blow". The first decides whether the MOSFET entered avalanche the moment the surge arrived; the second decides whether the part a year later is still the same part.
7. Common pitfalls
| Pitfall | Symptom | Correct approach |
|---|---|---|
| Firing a single current point | The calibrated point passes but the real high-current clamp is far higher | Sweep to get the Vc curve and see the dynamic-resistance rise |
| Measuring at the TVS terminals and calling it done | TVS reading passes, the MOSFET still blows | Measure the actual peak across the MOSFET drain-source |
| Mixing the two waveform conventions | Comparing 8/20us and 10/1000us figures directly | Only one convention is comparable; AMSEMI is 10/1000us, Littelfuse 8x20us must be attributed to Littelfuse |
| Firing once only | Single-shot passes; repeated surges in production punch through | Fire the specified shot count and re-measure drift |
| Measuring leakage at room temperature only | Passes cold, leakage doubles in a hot battery compartment and accuracy drifts | Measure IR and temperature rise at maximum steady-state voltage and 60-70 C |
| Using "a higher power tier" instead of verification | Assuming an 8000 W tier clamps lower and calling it done | At any voltage grade the 3 kW, 5 kW and 8 kW tiers have identical VBR and Vc; only surge current capability rises (54 V grade Ipp 34.4 A / 57.5 A / 91.8 A). The clamp still needs to be verified |
How to use power tiers correctly: the AMSEMI -N foldback (snap-back) series comes in 3000 W, 5000 W and 8000 W tiers; at the same voltage grade the VBR and maximum Vc are identical and only Ipp differs. So a higher tier means "harder to destroy", not "lower clamp" — the residual voltage the downstream MOSFET sees is unchanged and no margin re-calculation is needed. Full models and grades: AMSEMI Foldback TVS: Full Series and Cross-Reference.
Related articles
- Bus-level BMS TVS selection overview: How to Select TVS for BMS Protection Boards
- Bus-side three-stage voltage window: TVS Fitted, MOSFETs Still Blow: The 23S / 24S Voltage Budget
- Failure analysis when a TVS is already fitted: Why BMS Boards Still Burn MOSFETs With a TVS and BMS MOSFET Failure Mechanisms (Chinese)
- Protection for a high-impedance node such as a sampling line: Should BMS Cell-Sampling Lines Have a TVS?
- Surge test levels and staged protection: Surge Protection for Energy Storage / BMS
- Reading the power rating and derating: What Does a 5000 W TVS Rating Actually Mean?
- Telling a genuinely low-clamping part apart: Which Chinese Brands Make Low-Clamping TVS?
FAQ
Q1: A BMS protection board has a TVS fitted — how do I verify it actually works?
Bench testing must answer three questions: (1) at the surge current you have to block, is the true clamping voltage below the downstream MOSFET withstand ceiling; (2) are leakage and temperature rise at the maximum steady-state voltage and high ambient acceptable; (3) does the part drift after repeated surges. A pass means a curve, a measurement at the MOSFET and a thermal plus drift conclusion — not "we hit it a few times and nothing blew".
Q2: The datasheet Vc is below the MOSFET rating, so why does the measured clamp still exceed it?
Because the datasheet only specifies Vc at one current point, one waveform, a single pulse and room temperature, while clamping voltage climbs with current along the dynamic resistance. Real surge current is often far above the calibrated Ipp point, and high battery-compartment temperature plus trace inductance add on top, so the peak the MOSFET actually sees is considerably higher than the datasheet Vc. That is why on-board measurement is mandatory, and why it must be taken across the MOSFET drain-source.
Q3: Which waveform should the bench use, and can 10/1000us and 8/20us be compared?
No. Component-level parameter verification uses the 10/1000us convention that matches the datasheet (the AMSEMI 5.0SMDJ and -N datasheets only carry this one convention, with no 8/20us data); system-level immunity acceptance follows the IEC 61000-4-5 combination wave or the customer specification. Littelfuse's published 8x20us figures are Littelfuse's own convention and must be attributed to Littelfuse. The same part can take nearly an order of magnitude more peak current at 8/20us, so subtracting or converting between the two produces the wrong conclusion.
Q4: If I upgrade the TVS from 5000 W to 8000 W, will the clamp drop and the test pass?
No. The AMSEMI -N foldback series 3000 W, 5000 W and 8000 W tiers have identical VBR and maximum Vc at the same voltage grade; only surge current capability differs (54 V grade Ipp 34.4 A / 57.5 A / 91.8 A). A higher tier makes the device harder to destroy; it does not lower the clamping voltage or change the residual voltage the downstream MOSFET sees. To pull the clamp down, move to a lower-clamping grade (the foldback -N series is about 20 percent lower at the same grade).
Contact us
The hard part of bench verification is not the equipment but the order of the judging criteria: confirm the three-stage voltage window exists first, then do the current sweep, the measurement at the MOSFET, and the thermal and drift re-tests on the board, and finally issue a traceable report.
Send us your cell count and maximum working voltage, the downstream MOSFET part and parallel count, and the surge waveform and current class you need to block (or your specification test conditions). We will work the criteria through with you and can run bench measurements to your specification waveform, issuing a surge test report covering waveform, current point, Vc, MOSFET-side peak, thermal drift and parameter drift. Datasheet pages and samples for the AMSEMI 5.0SMDJ standard series and the -N foldback (snap-back) series, plus FAE review, are available on request. Contact us.
Shenzhen Intek Technology Co., Ltd — authorized distributor of AMSEMI (Anhui Anmei Semiconductor) TVS and HOPERF CMT series digital isolators Tel / WeChat: +86 136-3264-8484 Address: Room 4F, Building 2, Jingwei Center, No. 309 Ping'an Avenue, Pinghu Street, Longgang District, Shenzhen, China Website: www.intek.vip
Disclaimer: This article is technical and industry information for reference only and does not constitute selection, procurement or any other business advice. Device parameters (clamping voltage, breakdown voltage, peak pulse current, leakage current, holding voltage and so on), waveform conventions and derating ratios are subject to the latest datasheet from each manufacturer and to whole-system measurements. The AMSEMI 5.0SMDJ series (documents 8060014 / 8060055) quotes Vc and Ipp on a 10/1000us convention with no 8/20us data; Littelfuse 8x20us figures are Littelfuse's own convention. The 0.85 factor (industrial) and 0.8 (automotive) are engineering practice, not mandatory standards, so actual designs must be verified against the original documentation and measurements for the part selected. Work involving high voltage and surge testing must be carried out by suitably qualified personnel.