How to Select Charge and Discharge MOSFETs for a BMS Protection Board
Bottom line: MOSFET selection for a BMS runs in four steps — set the voltage rating from cell count, then size on-resistance and package from load current, then calculate temperature rise from power dissipation, and finally verify SOA and the paralleling scheme. Low-voltage BMS designs (e-bikes and e-trikes, portable power stations, residential storage low-voltage packs) almost always use N-channel low-voltage MOSFETs in the 20 V to 150 V range. For continuous high current the contest is RDS(on) and package thermal capability; for transients such as short circuits and surges it is SOA and turn-off response. The nominal current rating Id is not the deciding figure.
First: what do the charge and discharge MOSFETs actually do?
The protection board puts power MOSFETs in series with the main current path as switches, controlled by the protection IC:
- Discharge MOSFET: conducts while the load draws current, and turns off on overcurrent, short circuit or undervoltage to disconnect the pack from the load.
- Charge MOSFET: conducts while charging, and turns off on overcharge, overcurrent or reverse connection.
- On e-bikes and e-trikes, portable power stations and low-voltage residential storage packs, load current reaches tens of amperes or more. A single MOSFET usually cannot handle it, so several are paralleled to share current and heat.
Charge and discharge MOSFETs spend most of their life in the on state, so conduction loss (I squared times RDS(on)) dominates. During transients such as short circuits and load dump, the MOSFET must carry a large current briefly before it turns off, which tests the safe operating area (SOA) and turn-off speed. Confusing these two requirements — watching only the nominal current and ignoring thermal behaviour and SOA — is what produces the classic field failure: a prototype that is fine at room temperature, runs hot at full load in summer, and blows on the first short circuit or surge.
Selection flow
flowchart LR
A["Cell count and voltage<br/>(max charge voltage sets rating)"] --> B["Current demand<br/>continuous / peak / short"]
B --> C["On-resistance and package<br/>RDS(on) x thermal resistance"]
C --> D["Single device or parallel<br/>count / sharing / drive"]
D --> E["Transient check<br/>SOA / avalanche / turn-off time"]
E --> F["Board-level measurement<br/>temp rise / surge / short"]Step 1: set the voltage rating from cell count
The maximum charge voltage of the pack is the lower bound for rating selection. Take 4.2 V per cell at full charge for NMC or lithium manganese chemistries and 3.65 V for LiFePO4. In practice leave a 1.5x to 2x margin over the maximum charge voltage, then add the system transients (load dump, charge-port surge, motor back-EMF) into the assessment. Confirm the result against the complete product design and the datasheet.
| Platform | Typical cells | Max charge voltage (typ.) | Common rating | Typical application |
|---|---|---|---|---|
| 1S to 2S | 1 to 2S | 4.2 to 8.4 V | 20 to 30 V | TWS earbuds, electric toothbrushes, small packs |
| 3S to 5S | 3 to 5S | 12.6 to 21 V | 30 to 40 V | Power tools, vacuum cleaners, lighting |
| 6S to 10S | 6 to 10S | 25.2 to 42 V | 40 to 60 V | Scooters, garden tools, some e-bikes |
| 48 V e-bike / e-trike | 13S NMC / 16S LiFePO4 | 54.6 / 58.4 V | 75 to 100 V | E-bikes, e-trikes, battery swap packs |
| 48 V portable / residential storage | 13 to 16S | 54.6 to 67.2 V | 75 to 100 V | Portable power stations, residential storage packs |
The table shows typical pairings, not absolute rules. A higher rating on the same platform is acceptable, but on the same process a higher voltage rating means higher on-resistance and higher cost; too low a rating breaks down on the first surge or transient. The final rating follows the cell specification plus the transient design of the protection board, and part parameters follow the manufacturer datasheet.
Steps 2 and 3: current, then on-resistance, then temperature rise
- Calculate conduction loss first: steady-state dissipation is approximately I squared times RDS(on). The MOSFET sits in series in the main path, so every position carries the current. RDS(on) must be taken as the hot value: the on-resistance of a low-voltage MOSFET rises noticeably with junction temperature, and at 125 degrees Celsius it is typically about 1.5 to 2 times the 25 degree Celsius nominal figure (device dependent, per the datasheet temperature curve). Using the cold 25 degree value seriously underestimates temperature rise.
- Then fix the package and heatsinking: the package sets thermal resistance and the power dissipation ceiling. Medium currents commonly use SOP-8 or DFN (5x6) / PDFN. High-current parallel designs commonly use DFN, TO-252, D2PAK or TOLL-class packages with a large pad, combined with large copper areas for heatsinking. The dissipation figure in the datasheet assumes a specific PCB condition, so copper area and airflow must be part of the board-level calculation.
- Verify junction temperature: keep steady-state junction temperature at full load to roughly two thirds of the specification limit (commonly 150 degrees Celsius), that is around 100 degrees Celsius with margin, calculated at the highest ambient temperature. Overheating is often not a bad part; it is insufficient convection and copper.
| Parameter | What to check | Common mistake |
|---|---|---|
| VDS (drain-source rating) | Max charge voltage times margin, plus transients | Using the nominal pack voltage instead of the full-charge voltage |
| RDS(on) (on-resistance) | Use the hot value for loss and temperature rise | Treating the 25 degree Celsius value as the operating value |
| Id (drain current) | Reference only; the real limit is package dissipation | Looking only at nominal current and ignoring heatsinking |
| SOA (safe operating area) | Check short-circuit and surge operating points | Doing steady-state checks only, no transient check |
| Qg / VGS(th) (gate charge / threshold) | Match the protection IC drive capability, especially when paralleling | Insufficient drive after paralleling, slower switching |
| EAS (avalanche energy) | Where unclamped inductive turn-off energy exists | Standing outside the rating even with a TVS or clamp fitted |
Step 4: one device or several in parallel
When continuous current is high and a single package cannot dissipate the power, paralleling several identical MOSFETs is standard practice in low-voltage BMS design. Both the gain and the cost need to be counted:
| Item | Single MOSFET | N devices in parallel |
|---|---|---|
| Effective on-resistance | RDS(on) | Approximately RDS(on) divided by N; conduction loss falls roughly in proportion |
| Heatsinking area | One package pad | Spread over several packages, heat more distributed |
| Gate drive | Simple | Total Qg multiplied by N; driver peak current and switching time must be rechecked |
| Current sharing risk | None | Requires same lot and part number, symmetrical PCB layout, individual gate resistor per device |
| Short-circuit turn-off | Fast | Turn-off takes longer and short-circuit energy grows; must be validated with the protection IC |
Four rules for parallel design:
- Device consistency: same lot and same part number, so VGS(th) and RDS(on) match as closely as possible.
- Self-sharing comes from the positive temperature coefficient: on-resistance rises with temperature, so a hotter device automatically carries less current. This is a natural negative feedback, but only if the devices are thermally coupled in the same way — the PCB layout must be symmetrical, with no device sitting next to a heat source while another gets all the airflow.
- Individual gate resistor per device (commonly a few ohms to around ten ohms, tuned against real ringing and switching waveforms) to suppress circulating currents and parasitic oscillation between paralleled devices.
- Drive capability: total gate charge multiplies, so confirm that the protection IC can still drive the gates fast enough. Otherwise turn-off slows down and short-circuit energy runs away.
Do not blame the MOSFET: split "blown devices" into four causes
When a MOSFET fails in the field, identify the mechanism before changing parts, or you will cycle through parts forever:
- Overvoltage breakdown (VDS): rating too low, or port surge and load-dump spikes too high. The fix is a higher voltage rating plus clamping on the charge and discharge ports — for port TVS selection see How to Select a Low-Clamping TVS for a BMS Protection Board (Chinese). The lower the residual voltage, the more margin the MOSFET keeps.
- Gate damage (VGS / ESD): the gate oxide is punctured by static or drive spikes. Usually an ESD and gate-drive circuit problem, tied to the gate resistor and layout.
- Overcurrent or short circuit beyond SOA: the protection IC responds too slowly or the detection threshold is wrong, and the MOSFET burns outside its SOA. Compare protection IC turn-off time against the MOSFET SOA.
- Accumulated overheating: insufficient heatsinking or unequal sharing between paralleled devices, causing long-term high-temperature ageing. Revisit the temperature rise calculation and PCB heatsinking.
Whether a surge or short circuit kills the MOSFET first or the sense circuit and AFE first is often a protection-chain problem as a whole; the debugging approach in Why Does a BMS Board Still Blow MOSFETs With a TVS Installed helps locate it.
One BOM for the whole power path
The charge and discharge MOSFET is one link in the main path, not an isolated choice: controller (sensing and protection strategy) + MOSFET (power switch) + TVS (port protection) must be decided together. For controller selection see Which MCU for a Storage BMS? GD32 Selection Guide (Chinese); for power device trade-offs at higher voltages (IGBT, high-voltage MOSFET, SiC) see How to Select Power Devices for Residential Storage and Inverters (Chinese).
Intek Technology distributes CR Micro MOSFETs (from low voltage through superjunction platforms) and AMSEMI protection devices, and can produce a combined "charge/discharge MOSFET + low-clamping TVS" cross-reference table and paralleling recommendation from your cell count, continuous current and short-circuit requirement. Part numbers, RDS(on) grades and SOA curves follow the manufacturer datasheet.
FAQ
Q1: Which MOSFET part number should I use for BMS charge and discharge switching?
There is no universal part number. Set the voltage rating from cell count, size RDS(on) and parallel count from continuous current, choose the package from the thermal budget, then verify SOA. Intek Technology can match CR Micro low-voltage MOSFET part numbers to your platform and load current and issue a cross-reference table; parameters follow the manufacturer datasheet.
Q2: Why do BMS boards parallel several MOSFETs, and how many is right?
A single device cannot carry the conduction loss or dissipate the heat. Paralleling N devices reduces effective on-resistance to roughly RDS(on) divided by N and spreads the heat. The count is derived backwards from target effective RDS(on), per-device power budget and package thermal capability; there is no fixed number. Paralleling requires same part number and lot, symmetrical PCB, an individual gate resistor per device, and confirmation that the driver can supply the multiplied gate charge.
Q3: What voltage rating do I need for a 48 V pack with 13 to 16 cells?
Use the maximum charge voltage, not the nominal 48 V: 13S NMC charges to about 54.6 V and 16S LiFePO4 to about 58.4 V, and the industry normally picks a 75 V to 100 V rating for margin. A higher rating is safer, but on the same process both on-resistance and cost rise with voltage rating. Confirm the final choice against the product transient design and the datasheet.
Q4: The charge and discharge MOSFETs keep overheating or blowing. How do I debug it?
Identify the mechanism first: overvoltage breakdown on VDS (port surge and clamping), gate ESD on VGS, overcurrent or short circuit beyond SOA (protection IC response time), or accumulated overheating (heatsinking and sharing). For overheating, recalculate conduction loss and temperature rise with the hot RDS(on). For blown devices, check first whether a low-clamping TVS is fitted and how long short-circuit turn-off takes. Do not simply move to a higher-current part.
Contact us
For a low-voltage MOSFET cross-reference table, a combined charge/discharge MOSFET plus low-clamping TVS proposal, a parallel design review or sample testing, contact us.
Shenzhen Intek Technology Co., Ltd — electronic component distributor and system solution provider Phone / WeChat: 136-3264-8484 Address: 4F, Building 2, Jingwei Center, No. 309 Ping an Avenue, Pinghu Street, Longgang District, Shenzhen Website: www.intek.vip
Disclaimer: This article is technical education and industry exchange for reference only and is not selection, procurement or other commercial decision advice. The 1.5x to 2x voltage margin, the hot RDS(on) multiplier and the two-thirds junction temperature rule are engineering practice, not mandatory standards, and vary by manufacturer, process and lot. Final selection must follow the latest manufacturer datasheet and measurement on your own board. Work on high-voltage and high-current circuits must be carried out by qualified personnel.