A TVS diode protects a rail by clamping an overvoltage, and getting the protection right is a window problem: the standoff voltage must sit high enough that normal operation never trips the device, and the clamping voltage must sit low enough that the protected circuit survives the clamp. Between those two lines, the breakdown voltage and the tolerance stack decide whether the protection works or fails silently. This article sets the margin between standby, trip, and system limit, translates the VRWM / VBR / Vc alphabet, sizes the margin with real tolerance math, handles the repetitive-pulse case, and works three rails — 12 V, 24 V, and 48 V — end to end.
The Three Lines: Standby, Trip, and the System Limit
Every TVS selection is a window bounded by three voltages. The standby line is VRWM, the maximum working reverse voltage the rail can carry without the TVS conducting at all. The trip line is VBR, the breakdown voltage at which the TVS starts to conduct. The system limit is the maximum voltage the protected circuit can survive — the part’s clamping voltage Vc must sit below it, or the protection arrives too late to help.
The window is the space between VRWM and the system limit, and inside it the breakdown and clamping lines must be placed so that normal operation stays below VRWM, the clamp triggers before the system limit, and the margin on both sides absorbs the rail’s tolerance, the TVS’s own tolerance, and the temperature swing. A rail that runs at 12 V with a 15 V system limit has a narrow window and a demanding selection; a rail with a 5 V tolerance band and a 30 V system limit has a wide window and an easy one.
Reading the window first is the correct way to start, because it forces the selection to state the goal before a part number is opened. The protection family comparison places the TVS against the other clamp families; this article owns the margin math inside the TVS window, and the two read together.
VRWM vs VBR vs Vc: The TVS Alphabet in One Place
The three-letter alphabet is the source of most TVS confusion, and the relations are consistent enough to remove the guesswork. VRWM, the working or standoff voltage, is the highest steady voltage the TVS blocks without conducting — in practice the rail’s normal maximum plus a small guard band. VBR, the breakdown voltage, is where the TVS begins to conduct, and it always sits above VRWM. Vc, the clamping voltage, is the peak the TVS holds during a transient, and it always sits at or above VBR.
The ordering VRWM < VBR ≤ Vc is fixed, and the ratios are usable for estimation. The ratio VBR/VRWM is typically around 1.1 to 1.2, and the ratio Vc/VBR adds another factor of the same order, so a 12 V rail TVS might list VRWM 12 V, VBR around 14 V, and Vc around 20 V. The protected circuit must survive Vc, because Vc is the worst-case voltage the rail ever sees during a clamp — a designer who sizes only for VRWM and ignores Vc selects a part that protects on paper and destroys on the bench.
The rectifier voltage ratings guide is a useful parallel, because the VRRM / VRMS / VDC ordering on a rectifier follows the same discipline: read the rating that matches the actual waveform, not the largest number on the sheet.
How Much Margin Is Enough: 10%, 20%, and the Tolerance Stack
The margin between the rail’s normal maximum and the TVS standoff is the first tolerance question, and the answer is not a single percentage but a stack. The rail has its own tolerance — a “12 V” rail may run 10.8 V to 13.2 V across a switcher’s regulation band and load steps. The TVS has its own tolerance on both VRWM and VBR. Temperature moves the breakdown. The chosen margin must clear all of them simultaneously.
A practical starting rule is 10% to 20% margin between the worst-case rail maximum and VRWM, with the exact number set by the stack. A tightly regulated rail with a ±2% spec can use the low end of the range; a rail with a wide spec and a hot environment needs the high end to keep the TVS from tripping on the rail’s own extremes. The honest method computes the stack — rail worst case, TVS tolerance, temperature — rather than picking a percentage out of habit.
The same stack logic appears on the power side in the rectifier thermal design approach: both are tolerance and margin disciplines, and both reward computing the worst case over trusting a single fixed number. A TVS margin computed from the stack is defensible in a design review; a number pulled from a rule of thumb is not.
Repetitive Pulses: When the First Trip Is Not the Last
A TVS is a clamp, not an infinite energy sink, and the repetitive-pulse case is where the selection stops being a one-shot. The first transient clamps and absorbs energy; a second transient that arrives before the junction cools finds the TVS hotter, closer to its limit, and dissipating the same energy into less thermal headroom. The repetition rate and the energy per pulse together set the de-rating, and the datasheet’s pulse capability curve is the number that decides whether the part survives the actual duty.
The de-rating read looks at the pulse’s peak power and the repetition rate: a pulse that is perfectly safe at one event per minute can overheat the junction at one per millisecond. The thermal reset time — how long the junction needs to shed the absorbed energy and return to a safe starting temperature — sets the ceiling on how fast the events may repeat. Selecting a TVS on the single-pulse rating alone, without checking the repetition rate against the thermal reset, is how a protection part that survives the first hit dies on the hundredth.
The clamping margin and the energy budget are two halves of the same selection, and both must be satisfied. Board placement is the third practical factor that the margin math assumes but does not calculate: a TVS with a long trace to the protected rail clamps late, because the parasitic inductance of the trace delays the current rise and lets the overvoltage propagate. The selection closes only when the TVS sits close to the connector or the vulnerable node, with a short, wide return path — otherwise the clean clamping window on paper degrades into a late clamp in the real layout. The TVS protection failure review covers the layout, rating, and repetition mistakes that make a correct-looking clamping window fail in the field, and it is the natural follow-on to the voltage-margin work in this article.
The margin discipline extends to the selection’s failure modes, which are worth naming because they are the opposite of the failures a designer expects. The common instinct is that a TVS that is “too weak” is the risk — a part that clamps too high and lets the overvoltage through. That risk is real, and it is the upper-line failure the 24 V example showed. But the equally common silent failure is a TVS that is “too strong” for the rail — a part whose standoff voltage sits so far above the rail that it never trips on a legitimate transient, so the protected circuit absorbs the overvoltage the TVS was meant to clamp. The margin window is a two-sided budget: too little margin trips the TVS on normal operation, too much margin lets real transients through. Both are design errors, and both are avoided by computing the window rather than by choosing the “biggest” or “most protective” part in the catalog.
The same two-sided reasoning governs how a designer reviews an existing TVS selection. When a new rail inherits a TVS from a previous board, the first check is not the part’s energy rating but its window: does the standoff clear the new rail’s worst-case maximum with margin, and does the clamping stay under the new system limit? A TVS that was correct on a 12 V rail can be dangerously misapplied on a 24 V rail with the same board, purely because the window moved. The margin method converts a reusable part into a fresh selection decision every time the rail voltage changes, which is the disciplined habit the TVS selection guide and this article are both building.
Temperature is the second factor that a fixed percentage margin fails to capture, because the TVS breakdown voltage moves with junction temperature and the rail itself sags and swells with load. A margin computed at 25 °C can vanish at 85 °C ambient, and a part that trips cleanly in the lab can false-trip in the field on a hot day. The margin stack in this article includes temperature as an explicit line, not an afterthought, and a design review that does not state the operating temperature range is reviewing an incomplete selection. The same temperature awareness runs through the thermal design guide on the power side, and it is the reason both disciplines insist on worst-case conditions rather than nominal ones.
Sizing Worked Example for a 24 V Rail, 12 V Rail, and 48 V Telecom Rail
Three worked examples apply the window method to real rails, and they show both the success and the failure directions.
A 12 V rail with a 5% spec runs 11.4 V to 12.6 V and has a 20 V system limit. VRWM at 12 V or 13 V clears the 12.6 V worst case with a small guard band; the breakdown near 14 V and the clamping near 20 V sit under the 20 V system limit — a tight but workable window. The 10%–20% margin rule lands at the low end here because the rail is tightly regulated, and the selection closes cleanly.
A 24 V rail with a 10% spec runs 21.6 V to 26.4 V and has a 35 V system limit. VRWM at 26 V clears the worst case with margin, but the clamping near 42 V is above the 35 V system limit — the selection fails on the upper line. This is the classic hidden failure: a part that passes the lower margin and looks right on paper fails because its clamping exceeds what the circuit can survive. The fix is a lower-clamping family or a system-limit review, not a bigger part with a higher Vc.
A 48 V telecom rail with a wide transient budget and a 75 V system limit. VRWM at 48 V to 51 V clears the nominal, the breakdown sits near 58 V, and the clamping near 75 V sits at the system limit — a window that needs the tightest control and the highest-margin TVS family.
The margin stack becomes actionable as a short table that a designer fills in for any rail:
| Rail | Worst-case normal max | System limit | VRWM | VBR | Vc | Verdict |
|---|---|---|---|---|---|---|
| 12 V, 5% | 12.6 V | 20 V | 13 V | ~14.5 V | ~19 V | Pass, tight window |
| 24 V, 10% | 26.4 V | 35 V | 26 V | ~28 V | ~42 V | Fail — Vc over limit |
| 48 V telecom | 51 V | 75 V | 51 V | ~58 V | ~74 V | Pass, high-margin family |
The table is the window method in one view: each row carries the rail’s worst case, the system limit, and the TVS’s three-line response, and the verdict falls out of whether the lower and upper lines both clear. The table also makes the review auditable — a reviewer can see at a glance which line failed, rather than trusting a conclusion stated in prose. The three rows mirror the three worked examples and give the reader a reusable template for a new rail.
The final verification on the board completes the selection. After the TVS is placed, a transient injection test across the rail should show the clamp triggering below the system limit and returning to the standoff level once the event ends; if the scope shows the rail exceeding the system limit during the clamp, the Vc is too high and the selection must be revisited. The same scope discipline that reads rectifier behavior in ripple and transient troubleshooting verifies the TVS clamp in the real layout, closing the loop between the paper margin and the operating board. A TVS selection is finished not when the part is chosen, but when the clamp is verified at the operating temperature and the real transient — the margin method produces the numbers, and the bench confirms them. In short, the standoff margin is the honest bridge between a catalog part and a protected rail: it turns a vague bigger-is-safer instinct into a two-line window that is computable, reviewable, and verifiable at every rail voltage.
The TVS SMD category and the axial TVS family list parts whose three-line numbers fit each window, and the selection closes by confirming the chosen part’s Vc stays under the system limit with the tolerance stack already counted. The three examples together show that a TVS selection is complete only when both lines clear — a part that trips on the rail’s normal maximum fails the lower line, and a part whose clamping exceeds the system limit fails the upper one.