How to Choose an ESD Protection Diode for an Ethernet Port: 0.3-1.2 pF Capacitance, Discharge Ratings and Package Fit | VOOHU

Choosing an ESD protection diode for a differential port is rarely about finding a part. It is about what happens after the part goes on. Two failures come up again and again: protection is fitted and the unit still resets or drops link during ESD testing, or — less expected — ESD testing passes but the gigabit link starts negotiating badly and throwing errors, and removing the diode fixes it.

Both point at the same thing. A device sitting across a signal pair has to shunt energy within nanoseconds and stay invisible the rest of the time. Selection has to satisfy both ends, not just the kV number on the datasheet.

1. Junction capacitance is the first gate on a differential pair

An ESD device across a pair looks like a capacitor to ground as far as the signal is concerned. It pulls down the impedance of that line and attenuates the high-frequency content, and the faster the link, the more visible that becomes.

1.1 The tolerable figure changes by an order of magnitude with rate

A 100 Mb/s port has limited bandwidth and usually tolerates around 1 pF. At gigabit, four pairs run full duplex and the band that matters extends to hundreds of MHz, where 1 pF starts eating return-loss margin. At 2.5G, 5G and 10G the figure generally has to come down to sub-picofarad, or insertion and return loss look poor at the top of the band.

1.2 Asymmetry between the two lines hurts more than a large value

If the two lines of a pair carry devices with different capacitance, or one line runs longer or through more vias, differential energy converts to common mode and shows up directly in radiated emissions. Use the same channel group of the same part on both lines, and place it symmetrically in the layout.

1.3 If removing the diode fixes it, look here first

When a unit negotiates down, drops packets or shows errors with the ESD device fitted and recovers without it, suspect capacitance or asymmetry before touching PHY registers.

2. Match the voltage rating to the real level on that line

The standoff voltage must sit above the highest voltage the line sees in normal operation. Below it, the device conducts slightly during normal working, which shows up as leakage, heating and attenuated signal.

2.1 Start from what the line actually carries

Ethernet pairs run at low common-mode levels, where a 3.3 V part is fine. RS485 and CAN have a far wider common-mode range and need a higher rating. PoE pairs carry DC power on top of the data, and a rating chosen for data alone gets driven straight through.

2.2 Unidirectional or bidirectional follows the signal swing

A signal that stays positive can use a unidirectional part. Differential and bus signals swing either side of zero and need a bidirectional device, or the negative half gets clamped away.

2.3 Check whether the kV figure is contact or air

Published kV ratings come from contact and air discharge, and the two values for one part are usually different. Align the test method before comparing a device rating against a product-level requirement rather than comparing the larger number.

3. What ESD diodes, GDTs and MOVs each handle

These three often get lumped together as protection, but they address different energy levels at different speeds and work as a set.

3.1 ESD diodes: fast, but limited energy

Fastest response and lowest capacitance, which is what makes them suitable directly on signal lines against fast low-energy human-body events. They are not meant to absorb the long, high-energy surge a lightning-induced event produces.

3.2 GDTs: high energy capability, slower, with a breakdown threshold

A gas discharge tube carries far more current than an ESD diode, which suits it to the position nearest the port on cables that leave the building. The trade-off is slow response and a high breakdown voltage, too late to protect downstream silicon on its own, so it goes first and an ESD diode behind it does the fine clamping.

3.3 MOVs: power input only, never signal lines

A varistor has high capacitance and would swamp a high-speed signal. Its place is the power input, absorbing overvoltage on the supply line.

3.4 Indoor and outdoor need different depth

For equipment whose cabling stays inside a cabinet, one appropriate ESD device on the signal line is often enough. Where cable runs outdoors, between floors, or out to an external camera or access point, a GDT takes the first hit and the power input gets its own MOV. They divide the work rather than replace one another.

4. VOOHU protection device selection

4.1 On signal lines: match capacitance to rate

These fiveTVS/ESDparts cover the range from high-speed differential pairs to wide-voltage buses, with capacitance graded from 0.3 pF to 1.2 pF:

Working rule: 2.5G and above start at the 0.3 pF grade; gigabit sits between 0.3 pF and 0.8 pF depending on layout margin; 100 Mb/s and slower interfaces can go out to 1.2 pF. The 30 V bidirectional WHALD30E2B suits the wide common-mode range of RS485 and CAN. WHTB058VA is the only bidirectional TVS of the five and its 58 V rating targets PoE pairs and the supply side; note that its discharge rating and capacitance are not published, so request the datasheet before committing it to a signal line.

4.2 At the port front end: pick a GDT by line voltage

These fourGDTparts span 90 V to 400 V DC breakdown. The rule is that breakdown must sit above the highest voltage the line can legitimately reach, with margin:

The pin count matters. The three 2-pin parts shunt a single line to ground. WHGT090V1P0A is a 3-pin device where both lines share one discharge chamber, so the two fire more closely together, which suits a differential pair. Its upper operating temperature is also a grade higher than the 2-pin parts, worth noting for wide-temperature designs.

4.3 On the power input: MOV

This is currently the only MOV in theprotection devicerange, for absorbing overvoltage at the power input. Keep MOVs off signal lines; their capacitance will swamp a high-speed link.

5. Confirming the choice on the board

5.1 Take a baseline with the protection unpopulated

Build the same board two ways, protection depopulated and fitted, and measure link negotiation, error rate and return loss for each. The delta between them is what the device actually costs you, which is far more useful than one set of absolute figures.

5.2 Run ESD at product level, not just at the port

Do contact and air discharge separately and watch whether the unit self-recovers, needs a restart, or is damaged. Self-recovery means the energy went somewhere but disturbed the logic; a restart or damage means the shunt path is not good enough.

5.3 Confirm the GDT actually fires

Where a GDT is fitted, verify it conducts first during a surge. If it does not and the full energy reaches the ESD diode behind it, that diode punches through — the symptom is a port that goes dead after the first surge test.

5.4 Retest EMI

If emissions get worse after adding protection, check that both lines of the pair carry matching devices and symmetric routing, and that each device's ground via lands close by on the correct plane.

6. FAQ

When choosing an ESD protection diode, is lower capacitance always better?

No. Lower capacitance means less effect on the signal, but within a given technology it usually also means less shunt capability and a higher price. Set the capacitance ceiling from the interface rate, then choose discharge rating and voltage grade within that ceiling rather than reaching for the smallest number available.

A gigabit port already has an isolating LAN transformer. Is an ESD diode still needed?

Yes. The transformer isolates DC and common-mode voltage; it does little against a fast ESD pulse, and that pulse can couple in through the connector shell, the shield or the PHY side. The two components solve different problems.

Can a GDT or an ESD diode be used alone?

It depends on the cabling. Indoors, with cable staying inside the cabinet, one suitable ESD device is usually enough. Where cable goes outdoors or between floors, lightning-induced energy far exceeds what an ESD diode can take, so a GDT has to absorb it first — otherwise the first surge destroys the diode.

Can a low-capacitance part drop straight into a footprint built for a 1 pF-plus device?

Electrically it often can, but three things need checking together: the standoff voltage grade, whether the structure is uni- or bidirectional, and whether package and pinout are compatible. Packages across the five parts above run from DFN2510-10L to DO-214AB, so footprints are not interchangeable and a change of part usually means a layout change.

What is the difference between the -40 to +85 C and -40 to +125 C grades?

Choose by the real ambient plus the rise inside the enclosure. Outdoor cabinets, automotive and industrial environments, and positions close to a heat source, all favour the 125 C grade. Among the four GDTs, the three 2-pin parts top out at 85 C while WHGT090V1P0A reaches 125 C — worth watching in wide-temperature designs.

Conclusion

The order for choosing an ESD protection diode on an Ethernet port is: let the interface rate set the capacitance ceiling, let the real line level set the standoff voltage and the uni- or bidirectional structure, then check that discharge rating, package and operating temperature fit the layout. Add a GDT in front where cable leaves the building, and give the power input its own MOV.

VOOHU carries stock across ESD, GDT and MOV, with capacitance graded from 0.3 pF to 1.2 pF and GDT breakdown from 90 V to 400 V. Send over the interface type, rate, cable routing and the product test standard and we will walk that order with you. For clamping voltage, surge current capability and waveform data beyond what is published, request the datasheet for the specific part.

This article was prepared by the technical engineering team at Suzhou VOOHU Electronic Technology Co., Ltd. (VOOHU).