Does the TVS on a BMS Leak Enough to Drain the Battery? How to Work Out Leakage Current, Standby Drain and Long-Term Reliability
One-line conclusion: it does leak, but at the microamp level — the AMSEMI 5.0SMDJ standard series and its -N foldback version are both specified at 5 uA maximum reverse leakage at the 54 to 95 V grades (25 C, at VRWM, maximum-value convention). That is 43.8 mAh per year, about 0.04 percent per year on a 100 Ah pack, so at room temperature it is rarely the main reason a pack goes flat (cell self-discharge runs at percent-per-month, an equivalent current two orders of magnitude larger). The two things that actually matter are: leakage varies enormously with voltage grade and temperature — one datasheet lists 1000 uA at the 5 V grade against 5 uA above 54 V, and leakage grows at roughly a doubling per 10 C — and it keeps growing after high-temperature long-term bias and repeated surges. Low leakage comes from chip process (GPP glass passivation versus OJ acid etching) and HTRB behaviour, not from how low the clamping voltage is.
1. Two frequently asked questions
"The TVS on my BMS leaks all the time — will it drain the battery?"
At room temperature, almost certainly not to the point of flattening the pack, but "almost" has conditions. A TVS sits across the pack terminals and stands off the bus voltage continuously; in that state only reverse leakage IR flows, drawn from the pack and dissipated in the die. It is a continuous draw, not a transient. At the 54 to 95 V grades the datasheet limit is 5 uA: 5 uA for a full year is 43.8 mAh, which is 0.04 percent per year on a 100 Ah pack and can be ignored. On a 10 Ah e-bike pack the same 5 uA is 0.44 percent per year, already inside the standby budget.
"If it is that small, why bother checking it?"
Because it grows, and it grows fast. Leakage is strongly temperature dependent (a doubling per 10 C is the usual rule of thumb), and a battery bay in summer sits at 60 to 70 C while an outdoor storage cabinet or a sun-exposed e-bike pack runs hotter still. At 85 C that same 5 uA reaches roughly 320 uA (multiplier and real curve per the manufacturer data), which is 2.8 Ah per year on a 100 Ah pack — enough on a small pack to show up within a storage or shipping cycle. Add the parameter drift a TVS accumulates after surges, and leakage keeps climbing. This is the classic path to "passed at room temperature, drains in the field".
2. Working the numbers: how much does TVS leakage draw in a year
You do not need a simulation. Multiplying leakage by time settles the magnitude (leakage in uA times 8760 h divided by 1000 gives mAh per year; the percentages are against nominal pack capacity):
| TVS leakage (uA) | Per year | Share of a 100 Ah pack | Share of a 10 Ah pack |
|---|---|---|---|
| 5 | 43.8 mAh | 0.04% | 0.44% |
| 20 | 175 mAh | 0.18% | 1.8% |
| 80 | 701 mAh | 0.70% | 7.0% |
| 320 | 2.80 Ah | 2.8% | 28% |
These figures are pure arithmetic to establish magnitude. Real self-discharge also includes cell self-discharge, BMS standby draw and any balancing current, and must be confirmed by measuring the assembled unit at rest.
Seen as a breakdown, the TVS is only one contributor — and usually not the largest:
| Contributor | Typical order | Temperature sensitivity | How to isolate it |
|---|---|---|---|
| Cell self-discharge | percent per month | Medium (faster when warm) | Let the bare cells rest without the BMS |
| BMS standby draw (MCU / AFE) | microamps to milliamps | Medium to high | Break the supply rails and measure in sections |
| Passive balancing (while active) | milliamps | Low | Check whether balancing is stuck on |
| TVS reverse leakage | microamps (grade dependent) | High (doubling rule) | Remove the TVS and repeat the measurement |
The table gives typical orders of magnitude only; real designs vary widely. Deciding which item is responsible for a fast-draining pack requires staged measurements, not estimates.
3. Why one number cannot be reused: how to read the IR column
A datasheet gives leakage as one column, but it is a single figure at a specified VRWM, at 25 C, as a maximum — it does not hold at other grades or temperatures. In the AMSEMI 5.0SMDJ family (document 8060014 for the standard series, 8060055 for the -N foldback series) the pattern is worth remembering: the low-voltage grades are two to three orders of magnitude worse than the high-voltage grades.
| Voltage grade VRWM | IR at VR, max (25 C) | Vc max: standard 5.0SMDJ | Vc max: -N foldback |
|---|---|---|---|
| 5.0 V | 1000 uA | 9.2 V | not offered in -N |
| 9.0 V | 10 uA | 15.4 V | not offered in -N |
| 10 V | 5 uA | 17.0 V | not offered in -N |
| 54 V | 5 uA | 87.1 V | 69.6 V |
| 75 V | 5 uA | 121.0 V | 96.8 V |
| 90 V | 5 uA | 146.0 V | 116.8 V |
Convention note: the Vc figures above follow the 10/1000us waveform and come from AMSEMI documents 8060014 and 8060055, which contain no 8/20us data at all. Littelfuse 8x20us figures are Littelfuse own convention for its own parts and must be attributed to Littelfuse explicitly; the two conventions must never be mixed or subtracted. The IR column is the "Maximum Reverse Leakage IR at VR" line in the datasheet, a maximum at 25 C.
Three points on reading it:
- It is a maximum, not a typical. Use the limit for a worst-case standby budget; for trends (is this batch leaking more than before), compare against your own historical measurements rather than the datasheet.
- It varies drastically with grade. A 5 V-grade part for small-signal protection can be specified as high as 1000 uA, while the 54 V and above grades used on BMS buses come in at 5 uA. Do not carry one grade experience over to another.
- At the same grade the foldback and standard parts share the same IR limit (both 5 uA). The advantage of the AMSEMI -N series is a lower clamping voltage (about 20 percent lower Vc at the same grade), not lower leakage — low clamping and low leakage are two different things, set by two different parts of the design.
The datasheet also quotes Tj as -55 to +150 C (same in both documents). That is the device temperature rating, not a promise that leakage stays at 5 uA across that range; the two must not be conflated.
4. The real variables: temperature and ageing
First, leakage is exponential in temperature. Reverse leakage in a PN junction is dominated by carrier generation and rises exponentially with junction temperature; a doubling per 10 C is the usual rule of thumb. On that basis (a rule of thumb, not a datasheet guarantee — use the manufacturer temperature data for real designs):
| Ambient / junction temperature | Multiplier vs 25 C | Estimated leakage for a 5 uA part | On a 100 Ah pack per year |
|---|---|---|---|
| 25 C | 1x | 5 uA | 43.8 mAh (0.04%) |
| 45 C | 4x | 20 uA | 175 mAh (0.18%) |
| 65 C | 16x | 80 uA | 701 mAh (0.70%) |
| 85 C | 64x | 320 uA | 2.80 Ah (2.8%) |
This is why storage cabinets, e-bike battery bays and outdoor telecom supplies — everything that lives hot — cannot be specified from room-temperature data alone. Leakage that is irrelevant at 25 C becomes a continuous load and a heat source at operating temperature, and self-heating pushes leakage further: a positive feedback loop. The voltage-window article TVS Fitted, MOSFETs Still Blow: The 23S / 24S Voltage Budget touches on this loop; this article works the numbers out properly.
Second, repeated surges shift leakage and breakdown voltage. Every surge leaves microscopic damage in the junction. The macroscopic result: a slowly rising leakage limit, a drifting VBR and a higher clamping voltage at high current (a larger dynamic resistance). So "it survived a few hits" does not mean it will behave identically after a few hundred. Bench verification has a dedicated drift criterion for exactly this, described in How Do You Verify the TVS on a BMS Protection Board?.
Third, HTRB is the key reliability metric for continuously biased nodes. HTRB (high temperature reverse bias) measures precisely what a TVS on a battery bus actually sees: leakage and failure behaviour under a standing reverse voltage at elevated temperature. Industry sources agree that OJ acid-etched parts are limited to roughly 100 C, while GPP glass-passivated planar parts still perform well at 150 C. For a part that will be biased 8760 hours a year, that belongs in the specification alongside the clamping voltage.
5. Low leakage comes from the process, not from a low clamp
Two TVS parts with the same package (DO-214AB/SMC) and the same voltage grade can differ markedly in leakage and high-temperature behaviour, because the difference is in how the PN junction is passivated:
| Item | GPP glass passivation | OJ acid etching |
|---|---|---|
| Junction protection | Glass sintered to the die at about 800 C, fused into one body | Protective coating over the surface only (cured at about 200 C) |
| Room-temperature leakage | Lower | Higher |
| HTRB high-temperature reverse bias | Still excellent at 150 C | Limited to roughly 100 C |
| Suitable for | Continuously biased, hot, long-life designs | General consumer conditions |
The AMSEMI -N foldback (snap-back, low-clamping) series that Intek supplies is built on GPP glass-passivated planar die — low clamping is not paid for with worse leakage or worse high-temperature reliability. That is why, on a node that is energised 8760 hours a year, process and HTRB data belong in the selection criteria together with the clamping voltage, rather than picking whichever vendor has the prettier Vc number. Process details: Why Are AMSEMI TVS Reliable? GPP Glass Passivation Explained (Chinese).
6. Finding out why a pack drains while at rest
flowchart TD
A["Symptom: fast drain at rest / storage drain / SOC creeping down"] --> B["Measure standby current at room temperature<br/>cells plus BMS, record current and time"]
B --> C{"Which order does the current fall in?"}
C -- "milliamps" --> D["Check BMS standby draw and balancing<br/>MCU / AFE idle current, balancer stuck on"]
C -- "microamps" --> E["Break the rails and isolate<br/>remove TVS / cut AFE supply / cut comms supply"]
E --> F{"Does current drop clearly with the TVS removed?"}
F -- "no" --> G["TVS is not the culprit<br/>back to cell self-discharge and system standby"]
F -- "yes" --> H["Repeat the measurement hot<br/>60 C / 85 C, look at the multiplier"]
H --> I["Remedy: move to a GPP part<br/>raise VRWM margin / rework the sampling design"]
D --> HStep one: measure cold. Let the assembled unit rest (cells plus BMS only) and record the current over time. A milliamp-level draw has essentially nothing to do with the TVS — check BMS standby draw and whether the balancer is stuck on.
Step two: isolate by breaking rails. Remove the TVS, then cut the AFE supply and the comms supply, measuring after each. Whichever change drops the current identifies the culprit.
Step three: repeat hot. This is the step most often skipped and the one that exposes the problem — passing cold does not mean passing hot. Repeat the same test at 60 C and 85 C and check whether the current multiplier matches the doubling-per-10 C order.
Step four: act on the result. If the TVS dominates, first move to a part with GPP processing and better HTRB behaviour, then re-check whether the VRWM grade leaves enough margin (the maximum bus voltage must not sit at VRWM). If the leakage is on a cell-sampling line rather than the bus, the remedy is entirely different — a sampling line is continuously energised, high impedance and feeding a narrow AFE window, so leakage there turns straight into measurement error: see Should BMS Cell-Sampling Lines Have a TVS?.
7. Five common pitfalls
| Pitfall | Symptom | Correct approach |
|---|---|---|
| Choosing on clamping voltage alone, ignoring IR and process | Protection looks fine but standby or hot-weather consumption is out of spec | Put the IR limit, HTRB and the clamp into the criteria together for continuously biased nodes |
| Assuming every TVS leaks a few microamps | Measurements at low-voltage grades differ by two orders of magnitude | Read IR at VR for the grade you are using; never carry over across grades |
| Treating the datasheet maximum as a typical | False alarm about a part "degrading" | The datasheet is a limit; judge trends against your own history |
| Accepting the design at room temperature only | Passes cold, drains in the field | Repeat the standby measurement at the worst-case unit temperature |
| Believing low clamping means low leakage | Assuming a foldback part is automatically tighter | At the same grade foldback and standard parts share the same IR limit; low leakage comes from process and HTRB |
Related articles
- Selection overview (bus level): How to Select TVS for BMS Protection Boards
- Full series models and grades: AMSEMI Foldback TVS: Full Series and Cross-Reference
- Checking the three gates: TVS Fitted, MOSFETs Still Blow: The 23S / 24S Voltage Budget
- Why sampling-line leakage is a separate calculation: Should BMS Cell-Sampling Lines Have a TVS?
- Drift criterion and bench verification: How Do You Verify the TVS on a BMS Protection Board?
- Why low clamping and high reliability coexist: Why Are AMSEMI TVS Reliable? GPP Glass Passivation (Chinese)
- Reading the power rating and derating: What Does a 5000 W TVS Rating Actually Mean?
- Four causes of blown MOSFETs and the debugging order: Why BMS Boards Still Burn MOSFETs With a TVS
FAQ
Q1: Does the TVS on a BMS leak enough to drain the battery?
At room temperature it is usually not the main cause. A TVS sits across the pack terminals and stands off the bus voltage continuously, drawing only its reverse leakage; the AMSEMI 5.0SMDJ standard and -N foldback series are both specified at 5 uA maximum at the 54 to 95 V grades (25 C, maximum-value convention), which is 43.8 mAh per year or about 0.04 percent per year on a 100 Ah pack. Hot conditions change the order of magnitude — at roughly a doubling per 10 C, 85 C gives about 320 uA, or 2.8 Ah per year on a 100 Ah pack — so small packs must include it in the standby budget. Confirm against a measurement of the assembled unit at rest.
Q2: How large is TVS leakage current, and how do I read the IR column?
IR at VR is a maximum at a specified VRWM and 25 C — not a typical, and not valid across the temperature range. In the AMSEMI 5.0SMDJ family (documents 8060014 and 8060055) it is 5 uA at the 54 to 95 V grades and much higher at low grades (1000 uA at the 5.0 V grade, 10 uA at 9.0 V), a spread of two to three orders of magnitude within one datasheet. So reading it takes three habits: confirm which grade you are looking at, remember it is a limit, and re-estimate for elevated temperature using the exponential trend.
Q3: Why does TVS leakage rise so much with temperature, and does it get worse with age?
Both are true. Reverse leakage is dominated by carrier generation and rises exponentially with junction temperature — a doubling per 10 C is the usual rule of thumb. On top of that, every surge leaves microscopic damage in the junction, so over life the leakage limit rises, VBR drifts and the clamping voltage at high current increases. That is why continuously biased nodes are judged on HTRB (high temperature reverse bias): GPP glass-passivated parts still perform well at 150 C, whereas OJ acid-etched parts are limited to roughly 100 C. Confirm against the manufacturer datasheet and reliability reports.
Q4: How do I tell whether a fast-draining pack is caused by the TVS?
Four steps: measure standby current at room temperature with cells plus BMS only — a milliamp-level draw points at BMS standby consumption and balancing first; then break the TVS, the AFE supply and the comms supply in turn and see which removal drops the current; then repeat at the worst-case unit temperature (60 to 85 C) and check whether the multiplier matches the leakage temperature rule; finally act on the result — if the TVS dominates, move to a GPP part with better HTRB behaviour and re-check the VRWM margin, while leakage on a sampling line must be handled against the measurement accuracy budget instead.
Contact us
Selecting a TVS for a continuously biased node means answering three questions at once: is the voltage window valid, is the leakage still acceptable at the worst-case temperature, and will the part stay inside its parameters over a ten-year life. Answering only the first tends to produce designs where protection looks fine but standby drain and temperature behaviour do not line up.
Send us your cell count and maximum working voltage, worst-case unit temperature, measured standby current (or drain rate), and the downstream MOSFET / AFE parts and ratings. We will work the voltage window, the standby budget and the leakage-versus-temperature numbers through with you and give a definite answer on which grade to use, whether a different process class is needed and whether the sampling design has to change. Datasheet pages, samples and FAE review for the AMSEMI 5.0SMDJ standard series and the -N foldback (snap-back) series 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 (leakage current IR, clamping voltage Vc, breakdown voltage VBR, peak pulse current Ipp and so on) and waveform conventions are subject to the latest datasheet from each manufacturer and to whole-system measurements. The AMSEMI 5.0SMDJ series (documents 8060014 standard and 8060055 foldback -N) quotes Vc and Ipp on a 10/1000us convention with no 8/20us data; Littelfuse 8x20us figures are Littelfuse own convention for its own parts. Self-discharge conversions, temperature multipliers and leakage orders of magnitude are engineering estimates (rules of thumb) rather than datasheet guarantees; real designs must be based on manufacturer data, your own circuit calculations and measurements on the assembled unit. Work involving high voltage and battery packs must be carried out by suitably qualified personnel.