English

简体中文

English

简体中文

Appearance

What Is CMTI in a Digital Isolator? Why an Isolator Still Misfires and Gets Damaged ​

Bottom line: an isolator separates steady-state potential difference, not transient energy. When an interface fitted with a 5 kVrms isolator still fails, there are only two paths. Path 1, common-mode transient: an extremely high dV/dt between the two grounds drives a displacement current across the barrier, so the output glitches or the communication link produces errors. This is what CMTI (common-mode transient immunity), quoted in kV/us, governs. Path 2, surge energy: lightning, hot-plugging, inductive kickback or ESD enters through the interface wiring and punches through the barrier or burns the bus-side pins. This is what interface TVS diodes govern, and the energy must be diverted before it reaches the barrier. CMTI, isolation withstand voltage (kVrms) and TVS clamping are three specifications of different nature; reading them interchangeably guarantees a wrong conclusion: high CMTI does not mean surge immunity, 5 kVrms is not a continuous working voltage, and a TVS placed on the wrong side of the barrier does nothing. Isolation figures below follow HOPERF CMT official material and the original datasheet; TVS figures are quoted together with their waveform convention.

mermaid
flowchart TD
    A["Isolated interface misbehaves:<br/>errors, false triggering, damaged parts"] --> B{"Is the device damaged?"}
    B -- "No, recovers when traffic resumes<br/>occasional bad bits, output glitches" --> C["Path 1: common-mode transient<br/>CMTI margin too small or coupling path too open"]
    B -- "Device permanently damaged<br/>barrier breakdown, burnt bus pins" --> D["Path 2: surge energy<br/>interface TVS missing or misplaced"]
    C --> C1["1. Compare CMTI minimum, not typical"]
    C --> C2["2. Inspect the coupling path:<br/>barrier keep-out, stray capacitance, shield"]
    C --> C3["3. Give common-mode current<br/>a return path that avoids the barrier"]
    D --> D1["1. TVS outside the isolator, at the connector"]
    D --> D2["2. Coarse clamp, decoupling, then fine clamp"]
    D --> D3["3. Isolator transient rating is only a backstop"]
    C1 --> E["Verify: common-mode transient injection + surge bench"]
    C2 --> E
    C3 --> E
    D1 --> E
    D2 --> E
    D3 --> E

Two questions engineers ask most ​

Is CMTI the same thing as a 5 kVrms isolation rating?

No. CMTI (common-mode transient immunity) describes how fast the voltage can move: the maximum voltage slew rate across the two grounds, with a specified amplitude and rise time, at which the device output still holds the correct logic state. It is quoted in kV/us or V/ns. An isolation rating such as 5 kVrms describes how high, and how long: it is a type-test value in the spirit of UL1577, applied for about one minute without breakdown. One is immunity to fast events, the other to high voltage. Neither implies the other. A device with an excellent CMTI can still be destroyed by surge energy, and a device with a high isolation rating still glitches when the common-mode step is steep enough. The dual convention of isolation ratings, type test versus continuous working voltage, is covered in What Does 5 kVrms on an Isolator Mean.

Why does an isolated interface still produce errors or false triggering?

Separate the two paths first. If the device is intact and errors are intermittent, with output glitches that clear when traffic resumes, the cause is a common-mode transient coupling across the barrier: check the CMTI margin and the coupling path (barrier layout, stray capacitance across the barrier, shield and grounding arrangement, common-mode current return path). If the device is destroyed, the barrier is shorted, or the same part keeps failing, the cause is surge energy: check the protection placement and level (is the TVS outside the isolator, is the protection staged, does the energy rating match the system requirement). The two paths call for completely different fixes, so decide the path before changing anything, and skip the reflex of swapping in an isolator with a higher isolation rating.

Three specifications that must not be confused ​

SpecificationWhat it resistsUnit / formHow to read it
CMTIHigh dV/dt between the two grounds (false switching)kV/us or V/ns, static and dynamicCompare the minimum, not the typical; must hold at the highest data rate
Isolation withstandHigh voltage held continuously across the barrier (insulation, safety)kVrms type test plus working isolation voltageType test is not a working voltage; use the working isolation voltage for busbar checks
Surge withstandTransient energy impact (breakdown, destruction)Peak voltage, peak current, waveform (for example 8/20us)Handled by the interface TVS; do not compare across waveform conventions

The single most common mistake sits in the first row: minimum versus typical. Isolator datasheets often quote both static and dynamic CMTI, and often only a typical value. A design calculation must use the worst-case minimum, valid simultaneously at the highest data rate, the highest ambient temperature and the largest common-mode swing. Using the typical value treats margin as if it were a result.

Where does the common-mode transient come from? ​

A common-mode transient is fundamentally a fast voltage step between the two grounds. Four sources cover most cases.

SourceTypical situationSignatureFirst countermeasure
Power-stage switchingInverter legs, PCS, bidirectional DC-DC bridgesPeriodic, tracks the switching frequencyBarrier keep-out, isolated secondary supply, physical separation
Long-cable inductionField bus or sampling harness near heavy current loopsRepeats with load activityShield and ground arrangement, common-mode choke, series limiting
Ground potential shiftMulti-point grounded systems, ground rise during lightningLarge amplitude, high energy, often with surgeInterface TVS to dump energy, plus a single-point grounding strategy
Common-mode part of a surgeCommon-mode content of lightning, hot-plug, ESDSingle event, high energy, very steep edgeLet the TVS remove the energy first, then talk about CMTI

The order of countermeasures matters: cut the energy and the coupling first, and only then pick a device with a higher CMTI. Common-mode current will find a way home; if the board leaves the barrier as the only route, no CMTI figure will save the layout. Barrier keep-out, ground planes that do not bridge the barrier, stray capacitance across it and the common-mode return path are covered in Isolation Barrier PCB Layout and Isolated Supply.

How to read CMTI: static versus dynamic, typical versus minimum ​

  • Static CMTI: the common-mode dV/dt a device tolerates while the output sits at a fixed level with no data traffic. This figure is the more forgiving one.
  • Dynamic CMTI: the dV/dt tolerated while valid data is being transmitted, without corrupting a bit or flipping the output. This is the real operating case and is usually the smaller figure.
  • Read the swing and the edge together: CMTI is always quoted at a specified common-mode amplitude and rise time. A number without its test conditions means nothing.
  • Coupling to data rate: at a higher rate the bit window is shorter, so the same disturbance is more likely to land inside a valid bit window. Fast links such as CAN FD and high-speed RS-485 therefore need more CMTI margin than slow signals.
  • Gate-driver positions are a different magnitude: an isolated gate driver faces the dV/dt generated by the power stage itself, far above that of a communication interface. HOPERF CMT8602X datasheet figures quote a CMTI typical value in the 150 kV/us range (refer to the original datasheet). But even a high CMTI only guarantees no false switching; it dissipates no energy - busbar surge must still be handled by a TVS. This division of labour matches How to Select an Isolated Gate Driver.

Counter-intuitive but important: CMTI is an immunity figure, not a survivability figure. A common-mode transient that corrupts the output usually leaves the device physically unharmed, while surge energy is what destroys it. So the reasoning that a 150 kV/us part must be tough is simply wrong.

The surge path: energy must be removed before the barrier ​

CMTI stops false switching; the TVS stops destruction. Three hard rules apply on this path.

  1. The TVS belongs outside the isolator, as close to the connector as possible. Surge enters from the cable, is clamped and dumped at the connector side, and only then, attenuated by the decoupling element, reaches the isolator and transceiver. Placing the TVS on the control side lets the surge cross the barrier first, which defeats the protection entirely.
  2. Stage the protection when the energy is large: coarse clamp at the front, a decoupling element, then a fine low-capacitance clamp. High-speed buses need the fine stage capacitance kept low, and the decoupling element separates the high energy of the front stage from the low capacitance of the back stage. The complete parameter alignment is covered in TVS Next to an Isolator: Surge Protection Coordination.
  3. The isolator transient rating is only a backstop. General-purpose isolator datasheets quote a transient overvoltage withstand in the region of 8 kV peak, as a one-shot or limited-event figure that cannot be used as repeated-surge protection (refer to the selected part datasheet).

How to verify: inject common-mode transients, then run the surge bench ​

  • Common-mode transient injection: while the device is communicating normally, inject a common-mode pulse of specified amplitude and rise time through a coupling capacitor between the two reference grounds, and watch for glitches, false switching or dropped packets. The pass criterion is uninterrupted communication, not a part that stays cool.
  • Worst-case combination: verify CMTI margin with the highest ambient temperature, the highest data rate and the largest common-mode swing all applied together. Avoid concluding anything from room-temperature low-speed tests; dynamic CMTI is normally below static CMTI.
  • Surge bench: apply the waveform and level required by the system safety specification at the interface, measure at the protected device pins rather than across the TVS, confirm the residual voltage stays below the downstream absolute maximum, and re-check parameter drift after repeated strikes.
  • State the waveform convention: component-level verification uses the convention of the datasheet. AMSEMI 5.0SMDJ Vc and Ipp figures follow 10/1000us (documents 8060014 and 8060055, with no 8/20us data); Littelfuse 8x20us figures are Littelfuse's own convention and must be attributed to Littelfuse. The two must never be mixed or subtracted.

Design notes and common pitfalls ​

Symptom or habitUsual misdiagnosisCorrect approach
Occasional bad bits, output glitchesSwap in an isolator with a higher isolation ratingCheck the CMTI minimum and the coupling path first; the isolation rating has nothing to do with CMTI
The same isolator keeps burningBlame the batch qualityCheck where the interface TVS sits and what energy level it handles
Using the typical CMTIThe datasheet says 150 kV/us, that is plentyUse the minimum under the worst-case rate, temperature and swing
Reading CMTI but ignoring layoutA high enough figure solves everythingGive common-mode current a return path; fix barrier keep-out and stray capacitance
A high-power TVS on a high-speed busMore power is always saferCapacitance ruins the fast edge; stage it as coarse clamp, decoupling, fine clamp

FAQ ​

Q1: What does CMTI mean, and how is it different from a 5 kVrms isolation rating? ​

CMTI (common-mode transient immunity) is the ability of an isolator to hold the correct output logic state while a fast voltage step (high dV/dt) appears between the two grounds, quoted in kV/us. It is an immunity-to-speed figure. A 5 kVrms figure is a type-test isolation rating in the spirit of UL1577, judged by applying the voltage across the barrier for about one minute without breakdown, so it is an immunity-to-amplitude figure. Neither implies the other: a high-CMTI device can still be destroyed by surge energy, and a high-rating device still glitches under a steep common-mode step. Checking a system busbar requires the working isolation voltage, not the 5 kVrms type test. Refer to the selected part datasheet for actual figures.

Q2: Why does an isolated interface still produce errors or false triggers, and how do I tell the two causes apart? ​

Start by asking whether the device is damaged. An intact device with intermittent errors or glitches that clear when traffic resumes points to a common-mode transient: check the CMTI minimum margin and the coupling path, including barrier keep-out, stray capacitance across the barrier, shield and grounding, and the common-mode return path. A destroyed device, a shorted barrier or a part that keeps failing points to surge energy: check whether the interface TVS is outside the isolator, whether the protection is staged, and whether the energy rating matches the system requirement. The two causes need entirely different fixes, so decide which one applies before redesigning.

Q3: The isolator is rated at 150 kV/us CMTI, so do I still need surge protection? ​

Yes. CMTI is an immunity figure that only guarantees the output will not be corrupted by a common-mode dV/dt event, and it dissipates no energy. Surge is an energy event, and an interface TVS must divert it before the barrier. Isolator datasheets quote a transient overvoltage withstand in the region of 8 kV peak (refer to the selected part datasheet) as a one-shot or limited-event figure, so it is a backstop only. The right split is that CMTI keeps the link from false switching while the TVS keeps the device from being destroyed.

Q4: How is a common-mode transient test performed, and why does a room-temperature low-speed test pass while the field still fails? ​

The test injects a common-mode pulse of specified amplitude and rise time through a coupling capacitor between the two reference grounds while the device is communicating normally, and looks for glitches, false switching or dropped packets. Room temperature, low data rate and a small common-mode swing form the most forgiving combination, and many problems appear only when the highest ambient temperature, the highest data rate and the largest common-mode swing are applied together (dynamic CMTI is normally below static CMTI). Accept the design against that worst-case combination, and require the datasheet figure to be a minimum rather than a typical.

Contact us ​

For matching selection support across HOPERF CMT digital isolators (digital isolators, isolated CAN, isolated RS-485, isolated gate drivers, isolated sensing) and AMSEMI TVS diodes, for reviews of isolated-interface immunity and surge protection, and for samples or certification documents, please contact us.

Shenzhen Intek Technology Co., Ltd - authorised distributor of AMSEMI protection devices and HOPERF CMT digital isolators Tel / WeChat: +86 136-3264-8484 Address: Room 4F, Building 2, Jingwei Center, 309 Pingan Avenue, Pinghu Street, Longgang District, Shenzhen, China Website: www.intek.vip

Disclaimer: This article is provided for technical exchange and selection reference only. It is not design, procurement or commercial advice. Device figures are subject to the original datasheet and to board-level measurement; verify every application against your own operating conditions.