How to Select a TVS for an FPV ESC: Clamp Voltage and MOSFET Withstand on a 12S Four-in-One ESC

Bottom line: a 12S pack is 50.4 V at full charge, so the power MOSFETs can only be 60 V or 80 V parts. A standard TVS clamps at 93.6 V or more in every grade from 58 V to 85 V and therefore cannot be fitted inside an 80 V withstand window; only the foldback (snap-back) parts in the 58 V and 60 V grades bring the clamp down to 74.8 V and 77.4 V. This single calculation is what TVS selection for an FPV ESC really comes down to.

Intek Technology (INTEK, www.intek.vip) is an authorized distributor of AMSEMI (Anhui Anmei Semiconductor Co., Ltd) and handles supply and selection support across the full AMSEMI product line. Every device parameter cited in this article comes from AMSEMI public datasheets (5.0SMDJ series 8060014, 5.0SMDJ-N series 8060055, SMBJ series 8060058, GSD/GSM series 8180072/8180084), quoted on the 10/1000 µs waveform convention. Figures that are engineering estimates rather than direct datasheet values are marked as such in the text.

1. First, See What Kinds of Stress the Power Train Carries

An FPV power train is short: pack, connector, four-in-one ESC, four brushless motors, plus a BEC branch feeding the flight controller, video transmitter and receiver. Short does not mean clean. This chain carries at least five stresses of completely different natures, and treating all of them with one TVS is a common source of misjudgement.

flowchart TD
  B["LiPo pack<br/>4S ~ 12S"] --> X["XT30 / XT60<br/>hot-plug arcing"]
  X --> E["Four-in-one ESC<br/>4 x 3-phase bridge / 24 MOSFETs"]
  E --> M["4 x brushless motor<br/>commutation spikes / regen"]
  E --> C["BEC<br/>5 V / 9 V / 12 V"]
  C --> F["Flight controller / VTX / receiver"]
  F --> A["Antenna port<br/>5.8G / 2.4G / 915M"]

Node	Dominant stress	Who should absorb it
Pack input (VBAT)	Hot-plug arcing, cable inductance kickback, soft-start failure	High-power TVS (the subject of this article)
MOSFET drain-source / phase leads	Commutation spikes, motor back-EMF, braking regen	TVS + RC snubber + MOSFET avalanche capability
BEC output	Back-feed from downstream devices, hot-plug	TVS + Schottky OR-ing
Signal lines (DShot / telemetry / USB)	ESD, ground bounce	ESD diode (capacitance chosen by data rate)
Antenna port	ESD, human contact discharge	Ultra-low-capacitance ESD only; general-purpose parts do not work

The most important row is the last one: it shows that "protection device" is not one category. Treating the antenna port and the pack input as the same problem is the most common flaw in FPV hardware reviews.

2. After 12S, the MOSFET Voltage Window Is Only 9.6 V

Do the voltage arithmetic first; every later decision rests on these numbers.

Cells	Full charge (4.2 V/cell)	Usual MOSFET rating	Absolute headroom
4S	16.8 V	40 V	23.2 V
6S	25.2 V	40 V or 60 V	14.8 V / 34.8 V
8S	33.6 V	60 V	26.4 V
12S	50.4 V	60 V or 80 V	9.6 V / 29.6 V

A 12S pack at 50.4 V behind a 60 V MOSFET leaves only 9.6 V: that is the entire space between normal operating voltage and device breakdown. Hot-plug arcing, the inductive kickback of tens of centimetres of pack cabling, and motor braking regen each exceed that magnitude without effort.

If the pack is high-voltage lithium (4.35 V/cell), 12S full charge is 52.2 V and the headroom on a 60 V part drops further to 7.8 V. This directly changes the grade conclusion in section 3, yet many selection tables still assume 4.2 V/cell.

So on a 12S platform there are only two realistic MOSFET choices: 60 V or 80 V. 60 V is cheaper with lower Rds(on); 80 V costs one grade more and has slightly higher Rds(on). Which one you can use depends on whether the surge can be held below the device rating — which is exactly the TVS job.

3. Put the TVS Clamp and the MOSFET VDS in the Same Table

All figures below come from AMSEMI datasheets on the same waveform convention (10/1000 µs), so the grades are directly comparable.

VRWM grade	Standard Vc @ Ipp	Foldback Vc @ Ipp	Verdict against an 80 V MOSFET (worst case)
58 V	93.6 V @ 53.5 A	74.8 V @ 53.5 A	Standard exceeds rating by 17%, foldback leaves 6.5%
60 V	96.8 V @ 51.7 A	77.4 V @ 51.7 A	Standard exceeds by 21%, foldback leaves 3.3%
64 V	103.0 V @ 48.6 A	82.4 V @ 48.6 A	Both exceed
70 V	113.0 V @ 44.3 A	90.4 V @ 44.3 A	Both exceed
75 V	121.0 V @ 41.4 A	96.8 V @ 41.4 A	Both exceed
85 V	137.0 V @ 36.5 A	110.0 V @ 36.5 A	Both exceed

Three conclusions follow, and each one is counter-intuitive:

First, a standard TVS has no solution on 12S — this is not a matter of picking the wrong grade. To avoid conduction at 50.4 V, VRWM must be at least 58 V; and the standard part in the 58 V grade already clamps at 93.6 V. Going up in grade only raises the clamp; going down (54 V, 48 V) puts VRWM below the full-charge voltage, so the part conducts during a normal charge. For a standard TVS on 12S, "the correct grade" does not exist.

Second, a foldback part fits the 80 V window, but only in the 58 V and 60 V grades. The 58 V grade clamps at 74.8 V, leaving 6.5% margin on an 80 V MOSFET; the 60 V grade clamps at 77.4 V, leaving only 3.3%. Engineering practice favours the 58 V grade, provided the pack is counted at 4.2 V/cell for 50.4 V. With 4.35 V/cell high-voltage lithium (52.2 V), the VRWM margin on the 58 V grade falls to 11%, and you are trading "VRWM margin" against "clamp margin" with neither side comfortable — that decision needs the real pack and the real surge waveform.

Third, a 60 V MOSFET on 12S cannot be rescued by a TVS. To clamp below 60 V you would need a foldback part with VRWM under 45 V (45CA-N clamps at 58.1 V), but 45 V VRWM sits below the 50.4 V full-charge voltage, so it conducts during a normal charge. The combination "12S + 60 V MOSFET + TVS" therefore does not exist; you either use an 80 V part with a TVS, or a 60 V part with no TVS and rely on MOSFET avalanche.

One widely repeated claim needs correcting here: a low-clamp TVS does not buy you a lower MOSFET grade. On 12S, even with a foldback part clamping at 74.8 V you cannot enter a 60 V window, and all 24 devices still have to be 80 V. What a TVS buys on 12S is reliability — an 80 V part that does not enter avalanche and does not accumulate damage — not BOM cost. The scenarios where low clamping really does buy a grade are places with a lower bus voltage or with tightly spaced withstand grades, such as 100 V / 135 V / 150 V on BMS protection boards, where one grade step is worth far more.

4. Three Places the Arithmetic Goes Wrong

4.1 Assuming a Higher Power Rating Lowers the Clamp

This is the most expensive misconception. Within one grade, the power rating decides how much current the part can take, not how many volts it clamps at.

Part	Rated power	VRWM	Vc	Corresponding Ipp
SMBJ58CA (SMB)	600 W	58 V	93.6 V	6.5 A
5.0SMDJ58CA (SMC)	5000 W	58 V	93.6 V	53.5 A

Same grade, same clamp, 8x difference in surge current capability. Moving from SMB to SMC solves "the TVS itself is destroyed by the surge", not "the downstream circuit is destroyed by the clamp". To lower the clamp you must change the process (foldback) or change the grade; there is no third route.

4.2 Assuming the Clamp Voltage Is a Fixed Number

The Vc in a datasheet is the maximum clamp at the rated Ipp, not a constant. Below the rated surge current the clamp is lower; above it the clamp is higher and the device may already be overstressed.

Two datasheet points give a rough dynamic resistance: Rd ≈ (Vc − VBR(max)) / Ipp. For the 58 V grade:

• Standard 5.0SMDJ58CA: (93.6 − 71.2) / 53.5 ≈ 0.42 Ω
• Foldback 5.0SMDJ58CA-N: (74.8 − 71.2) / 53.5 ≈ 0.067 Ω

A factor of six. This tells you that foldback is not "a slightly lower marked value" but a dynamic resistance an order of magnitude lower once the junction is in breakdown — the larger the surge current, the wider the gap. It must be stated that this Rd is a linear estimate from two datasheet points; a foldback I-V curve drops to a holding voltage after breakdown and then rises, so linear extrapolation only indicates magnitude and cannot be used as datasheet data. For accurate values you need the measured Vcl-Ipp curve.

4.3 Assuming a Clamp Below the Device Rating Is Safe

A 74.8 V clamp against an 80 V device looks like 5.2 V of margin, but that 5.2 V is easily consumed by three things:

1. Trace inductance. Every extra 10 mm from the TVS to the MOSFET drain adds roughly 10 nH; at a transient di/dt of 1000 A/µs, V = L·di/dt is 10 V — more than the 5.2 V you started with.
2. TVS turn-on delay and trigger spread. The transition from blocking to clamping is not instantaneous, and VBR itself has a ±5% batch distribution (datasheets give min/max, not typical).
3. MOSFET withstand is a 25 °C figure. Rated VDS falls with junction temperature, and avalanche capability falls with it.

Engineering practice therefore keeps the peak within 80% to 90% of the rating (this is not a clause in any standard but each company's own derating rule — follow your own). At 80%, the target clamp for an 80 V part is 64 V, and at that point even the 74.8 V of the 58CA-N is not enough. This is why "TVS fitted and it still blew up" persists in real products: a TVS only moves the stress from "certain breakdown" to "enters avalanche but may survive". It is not an insurance policy.

5. Six Locations, One by One

5.1 Pack Input (VBAT): the Main Battleground

This is the only location that genuinely needs a high-power TVS, and it is the subject of every calculation above. Three practical points:

• Package. On a 30.5 × 30.5 mm four-in-one ESC, SMC (DO-214AB) usually fits, but keep it out of the thermal spread of the high-current copper. Where space is really tight, SMB drops Ipp from 53.5 A to 6.5 A in the same grade — enough for ESD-level energy, not for hot-plug arcing.
• Unidirectional or bidirectional. The pack is DC, so unidirectional is sufficient in principle; but reversed connection and negative ringing on long cables push the device the other way, so most FPV ESCs use bidirectional. That trade-off is covered in Unidirectional vs Bidirectional TVS.
• Grade. Follow the table in section 3: the 58 V foldback grade is the default answer for 12S, unless the pack is high-voltage lithium.

5.2 Phase Leads and MOSFET Drain-Source: Not the Only Option, and Often Not the Best

Commutation spikes and braking regen travel through the motor windings. Three approaches are common here: an RC snubber, a TVS, or nothing and rely on MOSFET avalanche. The problem with a TVS at this location is that it references to ground while the three phases float, so you need three devices (one per phase) and both cost and board area climb.

What you should confirm first is whether the device avalanche capability (EAS) is sufficient. The mechanism of inductive-load turn-off spikes is covered in Suppressing Relay Coil Turn-Off Spikes; the division of labour between MOSFET avalanche and a TVS is covered in What Is MOSFET Avalanche Breakdown.

5.3 BEC Output and OR-ing: A Schottky Job, Not a TVS Job

The BEC delivers 5 V / 9 V / 12 V at 1 to 3 A to the flight controller and video transmitter. What is needed here is low forward drop, not clamping:

Location	Device	Key parameters
5 V / 9 V / 12 V rectification	Schottky SS52B to SS5100B (SMB, 5 A)	VF 0.60 to 0.90 V, IFSM 100 A
Dual-BEC redundant OR-ing	Same, one per branch in parallel	Reverse leakage rises with temperature; check high-temperature IR
BEC input overvoltage	Low-power TVS (SMA/SMB)	Grade chosen against BEC input rating

One easily missed point: Schottky reverse leakage rises exponentially with temperature, so two OR-ing branches will start to back-feed each other when hot, raising both standby loss and heat. Read the 125 °C IR in the datasheet, not the 25 °C figure.

5.4 Antenna Port: ESD Here Cannot Be Chosen Casually — a General-Purpose Part Is a Short Circuit

FPV antenna ports run at 5.8 GHz and 2.4 GHz, with 915 MHz for long-range links. An ESD diode sits in parallel with the signal line, so its junction capacitance is directly across the RF path. A quick capacitive reactance calculation shows the consequence:

Junction capacitance	Reactance at 5.8 GHz	What it means for the RF path
0.3 pF	≈ 91 Ω	Acceptable; common on high-end links
0.5 pF	≈ 55 Ω	Marginal; check the insertion-loss curve
1 pF	≈ 27 Ω	Clearly degrades VSWR and sensitivity
200 pF	≈ 0.14 Ω	Effectively shorts the antenna port to ground

The existing AMSEMI GSD series (SOD-323) and GSM series (SOT-23) have CT in the 200 to 450 pF range (GSD05C 200 pF max, GSD03C 450 pF max), with ESD capability of IEC 61000-4-2 ±30 kV and 350 W (8/20 µs). They suit power ports, keys, LEDs and low-speed GPIO, and do not suit 5.8G/2.4G antenna ports. This must be stated plainly: ultra-low-capacitance ESD for the antenna port needs a separate selection, and a general-purpose part should never be substituted.

Conversely, this is also a test of whether a supplier is technically serious: one who opens with "our ESD parts work on any port" has most likely never looked at the insertion-loss requirement of an RF path.

5.5 Bootstrap and Gate: What Is Needed Here Is Speed, Not Clamping

Each of the four three-phase bridges in a four-in-one ESC has three bootstrap diodes, twelve across the board (some designs halve that with dual devices). Selection looks at reverse recovery time (trr) and withstand voltage (above bus voltage plus gate voltage), not clamp voltage. A TVS here only slows the switching edges.

5.6 USB / UART / Receiver Signals: Choose Capacitance by Data Rate

USB 2.0 high speed (480 Mbps) requires ESD capacitance within a few pF, so a 200 pF general-purpose part is again unusable. UART, SBUS and CRSF at a few hundred kbps to a few Mbps can use general-purpose parts. The criterion is data rate, not whether the line is a signal line.

6. Layout and Measurement: Why a Board With a TVS Still Burns

The same TVS with different layouts can show measured clamps more than 20 V apart. Four points:

1. Mount the TVS next to the MOSFET drain-source loop, not at the board-edge connector. The surge current flows through the segment from connector to TVS just the same, and its inductance adds the spike right back. The correct order is "connector → input capacitor → TVS → MOSFET", with the TVS and input capacitor grounded at the same point.
2. Ground through a via array, not a single via. Under a 50 A-class pulse, the inductance of one via in an SMC footprint is enough to raise the clamp by more than ten volts.
3. Probe the clamp across the MOSFET drain-source, not across the TVS. The trace drop between the two is real stress on the device; measuring at the TVS systematically underestimates it.
4. Use a differential or isolated probe with at least 100 MHz bandwidth. Measuring with a standard passive probe and its ground clip gives you ground bounce plus clamp, not clamp.

7. Selection Checklist

Step	What to confirm	Criterion
1	Maximum pack voltage	4.2 or 4.35 V/cell — this sets the VRWM floor
2	MOSFET withstand	60 V or 80 V — this sets the clamp ceiling
3	TVS grade	VRWM above full-charge voltage, and Vc below device rating
4	Process	If a standard part will not fit, move to foldback; do not just raise the power rating
5	Surge current magnitude	Check Ipp and the power-versus-pulse-width curve against the real waveform, not the headline power rating
6	Layout	TVS in the same loop as the MOSFET, multiple ground vias, probe at the device

Related Solutions

• What Is a Foldback (Snap-Back) TVS? Definition, I-V Reading and Latch-Up Assessment
• Foldback vs Standard TVS: Clamp Comparison (Vishay / Littelfuse Benchmarks)
• TVS Fitted, MOSFETs Still Blow: The 23S / 24S Voltage Budget
• AMSEMI 5.0SMDJ-N Full Voltage Grade and Parameter Matrix
• TVS and Bus Protection for BLDC Motor Drives

FAQ

Q1: Can a 60 V MOSFET on a 12S ESC be saved by adding a low-clamp TVS?

No. Clamping below 60 V requires a foldback part with VRWM under 45 V (45CA-N clamps at 58.1 V), but 45 V is below the 50.4 V full charge of a 12S pack, so it conducts during a normal charge. The route "12S plus 60 V MOSFET" does not work even with a TVS: use an 80 V MOSFET with a TVS, or no TVS and rely on MOSFET avalanche.

Q2: On 12S, how much difference does a foldback TVS make compared with a standard one?

In the same 58 V grade on the same 10/1000 µs waveform, the standard 5.0SMDJ58CA clamps at 93.6 V maximum and the foldback 5.0SMDJ58CA-N at 74.8 V, a difference of 18.8 V. Against an 80 V MOSFET the standard part exceeds the rating by 17% while the foldback part leaves 6.5% margin — and that difference decides whether the device is punched through or merely enters avalanche and may survive.

Q3: Does moving from a 600 W to a 5000 W TVS lower the clamp?

No. Within one grade the clamp is set by the process and the voltage grade, not by the power rating: SMBJ58CA (600 W) and 5.0SMDJ58CA (5000 W) both clamp at 93.6 V maximum, and the difference is the peak current they survive, 6.5 A versus 53.5 A. Raising the power rating solves the TVS being destroyed, not the downstream circuit being destroyed by the clamp.

Q4: Can a general-purpose ESD part be used on an FPV antenna port?

No. Antenna ports work at 5.8G/2.4G/915M, and the ESD diode junction capacitance sits directly across the RF path: 0.5 pF is about 55 Ω at 5.8 GHz, 1 pF about 27 Ω, and 200 pF only 0.14 Ω, which effectively shorts the antenna port to ground. Antenna ports need ultra-low-capacitance ESD (typically 0.5 pF or less); the existing AMSEMI GSD/GSM series at 200 to 450 pF suits power ports and low-speed signals and needs a separate selection for RF ports.

Contact Us

If you want the TVS grade checked against your cell count, MOSFET part number and actual surge waveform, send us the relevant part of the schematic and we will work through it item by item against the datasheet.

Shenzhen Intek Technology Co., Ltd — electronic components distributor and system solutions provider Tel / WeChat: 136-3264-8484 Address: 4F, Building 2, Jingwei Center, No. 309 Pingan Avenue, Pinghu Street, Longgang District, Shenzhen, China Web: www.intek.vip

---

Disclaimer: The data in this article is taken from AMSEMI public datasheets (5.0SMDJ series 8060014 Rev. 2025-05-22, 5.0SMDJ-N series 8060055, SMBJ series 8060058, GSD series 8180072, GSM series 8180084); clamp voltages and peak currents are on the 10/1000 µs waveform convention and must not be converted directly to 8/20 µs figures. Dynamic resistance, capacitive reactance and voltage margins in this article are engineering estimates derived from datasheet values, intended for order-of-magnitude judgement only, and do not constitute a device selection commitment. Actual designs must be based on the latest original datasheets, measured waveforms and whole-product validation. This is technical exchange content and does not constitute purchasing or commercial advice.