Zener Diode Resistor Calculation: The Worst-Case Method • The Science 360

Q: How do you calculate the resistor for a zener diode?

Use the worst case: RS = (Vin(min) minus Vz(max)) divided by (Iz(min) plus IL(max)). Then check the opposite corner, maximum input with no load, for zener current and power. For 10.8 to 13.2 V in, a 1N4733A 5.1 V zener, 0 to 20 mA load and 10 mA minimum zener current, RS comes out at 181 ohms, so 180 ohms.

Q: What wattage resistor do I need for a zener regulator?

The resistor dissipates (Vin minus Vz) squared over RS whenever the zener is regulating, regardless of load. In the 12 V to 5.1 V example with 180 ohms that is 0.39 W at 13.2 V in, so use a 1 W part. If the output can be shorted, the resistor sees the full input: 13.2 squared over 180 is 0.97 W.

Type "12 V in, 5.1 V zener, 20 mA" into a typical online zener calculator and it returns 345 Ω, then tells you to round down to the nearest standard value: 330 Ω. Now hang a 20 mA load on the output. At exactly 12 V the zener is left with 0.9 mA, below the 1 mA knee test current on the datasheet, and if the supply sags 10% the zener stops regulating altogether. The formula was right. It was simply applied to the wrong operating point.

Almost every result on the first page of a search for this topic is a calculator built on $R = (V_{in} - V_Z)/I_Z$ . That equation sizes a resistor for one zener current at one input voltage with no load. A real shunt regulator has an input that moves, a load that moves and a zener whose voltage depends on its current. This post does the calculation the way onsemi's own zener handbook lays it out: two corners, two checks, and then an honest look at how well the result actually regulates.

Quick answerCalculate the series resistor at the worst corner:

R_S = (V_{in(\min)} - V_{Z(\max)})/(I_{Z(\min)} + I_{L(\max)})

. Then check the opposite corner, maximum input with no load, where the zener current and power peak. For 10.8 to 13.2 V in, a 5.1 V 1N4733A and a 0 to 20 mA load, that gives 180 Ω and 0.25 W in the zener.

Infographic: What resistor for a zener diode? A schematic shows a 180 ohm resistor feeding a 5.1 V zener and a load, its 30 mA splitting 10 mA and 20 mA. Sized at the worst corner, 10.8 V in, the zener at 5.355 V and a 20 mA load, the resistor is 181.5 ohms, so 180. At 13.2 V with no load the zener takes 46.4 mA and 249 mW, and the resistor 0.388 W, so fit a 1 W part. A chart of zener current against load current shows the 180 ohm design keeping 10.25 mA at full load, while a calculator's 330 ohm answer leaves only 0.9 mA, below the 1 mA knee. — Tap the image to view it full size. View the accessible, editable SVG.

How a zener shunt regulator splits the current

The circuit has three parts (Figure 1). A resistor R_S drops the difference between input and zener voltage. The zener sits across the output with its cathode at the positive node, reverse biased, and the load hangs in parallel with it. Because the output is pinned near V_Z, the resistor current is set almost entirely by the input voltage:

\begin{aligned}I_S &= (V_{in} - V_Z)/R_S \\ I_Z &= I_S - I_L\end{aligned}

Read those two lines carefully, because everything else follows from them. The resistor does not know or care what the load is doing. It delivers a fixed current, and the zener absorbs whatever the load leaves behind. That makes the zener a variable dump, not a source. It can only regulate while I_Z stays positive and comfortably past its knee.

Schematic of a zener shunt regulator: a 10.8 to 13.2 volt source feeds a 180 ohm series resistor RS; at the output node a 1N4733A 5.1 volt zener diode connects cathode up to ground, in parallel with a load resistor RL drawing 0 to 20 milliamps. Arrows show IS through the resistor splitting into IZ through the zener and IL through the load. — Figure 1. The basic zener shunt regulator. I_S is fixed by the input voltage and R_S; the zener takes $I_S - I_L$ . Component values are the worked example below.

It is the same current-steering logic behind a resistor in series with an LED, with one twist: here a second branch, the load, competes for the current. It is also why a zener regulator behaves so differently from the plain voltage divider people try to use as a power supply. The divider's output sags in proportion to load; the zener branch gives up its own current first and holds the voltage until it runs out.

Which corner sets the resistor value?

Two things threaten regulation: too little zener current and too much. They happen at opposite corners of the operating range.

Too little zener current happens when the input is at its lowest, the load draws its maximum and the zener voltage is at the top of its tolerance (a higher V_Z leaves less voltage across R_S). That corner sets the largest resistor you can use. The onsemi "Zener Theory and Design Considerations" handbook writes it as:

R_S = \frac{V_{in(\min)} - V_{Z(\max)}}{I_{Z(\min)} + I_{L(\max)}}

Too much zener current happens at maximum input, minimum load (usually zero, because loads get unplugged) and the lowest zener voltage. That corner sets the zener's power rating:

\begin{aligned}I_{Z(\max)} &= \frac{V_{in(\max)} - V_{Z(\min)}}{R_S} - I_{L(\min)} \\ P_{Z(\max)} &= I_{Z(\max)} \times V_{Z(\max)}\end{aligned}

Multiplying a current computed with V_Z(min) by V_Z(max) is slightly pessimistic, which is what you want from a worst case. This is also where page-one calculators go wrong most often: they compute the zener's power at your stated "typical" current, which is the one condition the zener is guaranteed never to stay in.

Worked example: 12 V to 5.1 V at up to 20 mA

The job: power a small sensor board that draws up to 20 mA at about 5 V from a 12 V adapter that can sit anywhere from 10.8 to 13.2 V (±10%). The load can be unplugged, so I_L(min) = 0. The zener is the common 1N4733A, a 5.1 V, 1.3 W part in the Vishay DO-41 series. From its datasheet:

\begin{aligned}V_Z &= 5.1\ \mathrm{V}\ \text{nominal at } I_{ZT} = 49\ \mathrm{mA} \\ &\pm 5\% \to 4.845\text{ to }5.355\ \mathrm{V} \\ Z_{ZT} &= 7\ \Omega\ (\text{typical})\ \text{at } 49\ \mathrm{mA} \\ Z_{ZK} &= 550\ \Omega\ (\max)\ \text{at } 1\ \mathrm{mA} \\ P_{tot} &= 1.3\ \mathrm{W} \\ R_{thJA} &= 110\ \mathrm{K/W}\end{aligned}

P_tot applies with leads 4 mm from the body held at ambient.

Step 1: choose the minimum zener current. The datasheet's 1 mA is a knee test point where the impedance can be 550 Ω, not a recommended operating current. I will set I_Z(min) = 10 mA, which keeps the impedance down in the tens of ohms (Figure 2). More on that choice in the next section.

Step 2: size R_S at the starved corner.

\begin{aligned}R_S &= \frac{10.8\ \mathrm{V} - 5.355\ \mathrm{V}}{10\ \mathrm{mA} + 20\ \mathrm{mA}} \\ &= 5.445\ \mathrm{V}/0.030\ \mathrm{A} = \mathbf{181.5\ \Omega}\end{aligned}

Nearest E12 value at or below: 180 Ω. Check:

\begin{aligned}I_S &= 5.445/180 = 30.25\ \mathrm{mA} \\ I_{Z(\min)} &= 30.25 - 20 = \mathbf{10.25\ mA}\end{aligned}

Step 3: check the flooded corner.

\begin{aligned}I_{Z(\max)} &= (13.2\ \mathrm{V} - 4.845\ \mathrm{V})/180\ \Omega - 0 \\ &= 8.355/180 = \mathbf{46.4\ mA} \\ P_{Z(\max)} &= 46.4\ \mathrm{mA} \times 5.355\ \mathrm{V} = \mathbf{249\ mW} \\ \Delta T_J &= 0.249\ \mathrm{W} \times 110\ \mathrm{K/W} \approx \mathbf{27\ K}\end{aligned}

The last line is the junction rise, with ideal lead cooling.

Well inside the 1.3 W rating. A 500 mW glass zener would also survive on paper, but at half its rating it would run hot and drift, which is a poor trade for a few cents.

Step 4: size the resistor's own power. This is the step most calculators skip, and the resistor is usually the hotter part:

\begin{aligned}P_R &= (V_{in(\max)} - V_{Z(\min)})^2/R_S \\ &= 8.355^2/180 = \mathbf{0.388\ W}\end{aligned}

Output shorted:

P_R = 13.2^2/180 = \mathbf{0.968\ W}

A 0.25 W resistor would cook. A 0.5 W part would run at 78% of its rating, which in a closed box is too close. Use a 1 W resistor, which also survives a shorted output indefinitely.

Step 5: tolerance sanity check. A 5% resistor may be 189 Ω. At the starved corner that gives $5.445/189 = 28.8\ \mathrm{mA}$ , leaving 8.8 mA for the zener: a little under the 10 mA target but far from the knee. At 171 Ω and the flooded corner the zener takes 48.9 mA and 262 mW. Both are fine, which is the point of designing with margin.

Figure 3 shows what the arithmetic means across the whole load range. Each line is $I_Z = (V_{in} - V_Z)/R_S - I_L$ , a straight line with slope −1. The design lives in the gap between the purple line (starved corner) and the green one (flooded corner). The dashed orange line is the calculator's 330 Ω answer, and you can see it cross into the knee right at the rated 20 mA load.

Plot of zener current against load current from 0 to 30 milliamps. Three solid lines for a 180 ohm resistor at 13.2, 12 and 10.8 volts input start at 46.4, 38.3 and 30.3 milliamps and fall with slope minus one; at 20 milliamps load the 10.8 volt line is at 10.25 milliamps. A dashed line for a 330 ohm resistor at 12 volts starts at 20.9 milliamps and reaches 0.9 milliamps at 20 milliamps load, inside a shaded region below 1 milliamp marked not regulating. — Figure 3. Zener current versus load current. Solid lines: the 180 Ω design at low, nominal and high input, each with its worst-case V_Z. Dashed: the 330 Ω value a single-point calculator suggests for "12 V, 5.1 V, 20 mA". Computed from $I_Z = (V_{in} - V_Z)/R_S - I_L$ .

There is also a lower bound on R_S. If you derate the 1N4733A to, say, half its rating (650 mW), the zener current must stay below $650/5.355 = 121\ \mathrm{mA}$ , so $R_S \ge 8.355\ \mathrm{V}/0.121\ \mathrm{A} = 69\ \Omega$ . Anything from 69 to 181 Ω is a valid design; the high end wastes the least power.

What is the minimum zener current, really?

Zener datasheets list two currents, and neither is labeled "minimum operating current". The 1N4733A has I_ZT = 49 mA, the point where the 5.1 V ± 5% is guaranteed, and I_ZK = 1 mA, a point down in the knee where Vishay guarantees only that the impedance is below 550 Ω. onsemi's handbook describes I_ZT as usually at the quarter-power point and I_ZK as "an arbitrary low value in the knee region".

Why the impedance matters: the zener is not a perfect voltage source. Its slope in breakdown, $Z_Z = \Delta V/\Delta I$ , is its output resistance, and it falls steeply with current. The handbook notes that between I_ZK and I_ZT, impedance against current is close to a straight line on log-log axes, so the two datasheet points let you estimate the curve in between. Figure 2 does that for three Vishay parts.

Log-log plot of zener dynamic impedance in ohms against current in milliamps. Straight lines join datasheet points for the 1N4728A 3.3 volt (400 ohms at 1 milliamp to 10 ohms at 76 milliamps), 1N4733A 5.1 volt (550 ohms to 7 ohms at 49 milliamps) and 1N4736A 6.8 volt (700 ohms to 3.5 ohms at 37 milliamps). A square marks the BZX79C5V1 maximum of 60 ohms at 5 milliamps. The design range of 10 to 46 milliamps is shaded. — Figure 2. Dynamic impedance versus current, interpolated on log-log axes between the Vishay datasheet values at I_ZK (maximum) and I_ZT (typical), as the onsemi handbook suggests. The interpolation is ours and is not valid outside the two end points.

Plugging numbers into that interpolation for the 1N4733A gives roughly 90 Ω at 5 mA, 42 Ω at 10 mA, 19 Ω at 20 mA and 9 Ω at 40 mA. The 500 mW BZX79C5V1 from onsemi, specified at 5 mA, has a maximum of 60 Ω there, which lands on the same trend. So "10 mA minimum" for a 1 W part is not superstition: below that the zener is a soft spring rather than a stiff one.

There is a second, quieter consequence. The printed voltage is only promised at I_ZT. Integrate the impedance curve from 49 mA down to 10 mA and the 1N4733A's voltage could be around 0.6 V lower than at 49 mA. That is a large number, and probably pessimistic, because the curve mixes a typical value with a maximum one.

Source note The Vishay 1N4733A datasheet guarantees V_Z only at 49 mA. I could not find a manufacturer's figure for its voltage at 5 to 20 mA. The 0.6 V estimate above is our own integration of the log-log impedance line, which uses a maximum value at 1 mA and a typical value at 49 mA, so treat it as an upper bound on the sag rather than a prediction. If the exact output voltage matters, measure your parts at your operating current, or choose a zener characterised at that current, such as the BZX79 series at 5 mA.

How well does a zener actually regulate?

Now that the design keeps the zener in breakdown, we can ask how much the output still moves. The onsemi handbook gives the full expression; for practical purposes it boils down to two small-signal rules (our simplification), with R_Z the zener's dynamic impedance at the operating point:

Line regulation:

\Delta V_{out} \approx \Delta V_{in} \times \frac{R_Z \parallel R_L}{R_S + R_Z \parallel R_L}

Load regulation:

\Delta V_{out} \approx -\Delta I_L \times (R_S \parallel R_Z)

Line regulation in our example. At full load, swinging the input from 10.8 to 13.2 V moves the zener current from about 11.7 to 25 mA. Using the datasheet's best-case 7 Ω gives the optimistic bound; the interpolated curve gives the pessimistic one:

Optimistic (R_Z = 7 Ω):

\Delta V_{out} \approx 2.4\ \mathrm{V} \times 7/187 = \mathbf{0.09\ V}

Pessimistic (integrated curve, 11.7 to 25 mA):

\Delta V_{out} \approx \mathbf{0.30\ V}

So a 2.4 V swing on the input becomes somewhere between 0.09 and 0.3 V on the output: an improvement of 8 to 27 times, not the near-perfect clamp many tutorials imply. The same ratio applies to ripple. A 1 V peak-to-peak ripple from a rectifier and filter capacitor comes out at roughly 0.1 V peak-to-peak at full load with R_Z near 20 Ω.

Load regulation. At 12 V in, going from no load to 20 mA moves the zener from 38.3 to 18.3 mA:

Optimistic:

\begin{aligned}\Delta V_{out} &\approx 20\ \mathrm{mA} \times (180 \parallel 7\ \Omega) \\ &= 20\ \mathrm{mA} \times 6.7\ \Omega = \mathbf{0.13\ V}\end{aligned}

Pessimistic (integrated curve, 38.3 to 18.3 mA):

\Delta V_{out} \approx \mathbf{0.27\ V}

Add temperature. The zener's own dissipation changes with load, so its junction temperature moves by up to about 27 K in this design. onsemi's BZX79C5V1 lists a temperature coefficient of −2.7 to +1.2 mV/K, which over 27 K is up to about 0.07 V. The 5.1 V region happens to be near the crossover between the zener mechanism (negative coefficient) and the avalanche mechanism (positive coefficient), which is why it is relatively stable thermally; the BZX79C6V8 is +1.2 to +4.5 mV/K.

Is a higher-voltage zener a better regulator?

Around 6 to 7 V, yes. The onsemi handbook says zeners below about 5 V break down by field emission with high impedance and a negative temperature coefficient, those above about 8 V by avalanche, and both mechanisms act in between. It notes that zeners around 7 V "will have remarkably low impedance". You can see it in Figure 2: the 6.8 V 1N4736A reaches 3.5 Ω where the 3.3 V 1N4728A is still at 10 Ω, and in the BZX79 series the 6.2 V part is specified at 10 Ω maximum against 95 Ω for the 3.3 V part, both at 5 mA. A low-voltage zener at a few milliamps is a poor regulator. If you need 3.3 V, think twice before reaching for a 3.3 V zener.

Why the zener runs hottest when nothing is connected

Figure 4 plots the power in each part at the flooded corner. The resistor's dissipation is a flat line: as long as the zener regulates, the voltage across R_S is fixed, so its power is fixed. The zener's dissipation is a falling line that starts at its maximum with no load. The total power drawn from the supply never changes with load at all.

Plot of power in milliwatts against load current from 0 to 25 milliamps at 13.2 volts input. The resistor line is flat at 388 milliwatts. The zener line falls from 249 milliwatts at no load to about 115 milliwatts at 25 milliamps. The load line rises from zero to about 121 milliwatts. — Figure 4. Power in the resistor, zener and load at the worst-case input (13.2 V, V_Z at minimum for current, maximum for zener power). Computed from $P_R = (V_{in} - V_Z)^2/R_S$ and $P_Z = I_Z V_Z$ .

This is the least intuitive property of a shunt regulator and the root of most fried zeners. Someone builds the circuit, measures a comfortable zener temperature with the load running, then unplugs the load for the night. The zener now absorbs every milliamp the load was using.

It also explains the efficiency. At 12 V nominal and full load:

\begin{aligned}P_{in} &= 12\ \mathrm{V} \times (12 - 5.1)/180\ \mathrm{A} \\ &= 12 \times 38.3\ \mathrm{mA} = 0.46\ \mathrm{W} \\ P_{out} &= 5.1\ \mathrm{V} \times 20\ \mathrm{mA} = 0.102\ \mathrm{W} \\ \text{Efficiency} &= 0.102/0.46 = \mathbf{22\%}\end{aligned}

Efficiency is 0% at no load, with the same 0.46 W drawn.

The linear regulator heat problem is the same physics in a different package, except that a series regulator at least stops drawing current when the load does.

When a zener and resistor is the wrong tool

A zener shunt regulator earns its place as a reference, a clamp, or a supply for a few milliamps that rarely changes. It is a poor choice once the load is tens of milliamps and varies, for the reasons above. Three upgrades, in increasing order of effort:

Add an emitter follower. Put the zener on the base of an NPN transistor and take the output from the emitter. The zener now supplies only base current, roughly $I_L/\beta$ , so its current barely changes with load and R_S can be much larger. The output sits one V_BE (about 0.65 V) below V_Z. Sizing that base drive is the same exercise as a transistor base resistor calculation, and the gain mechanism is covered in how a transistor amplifies.

Use a TL431 shunt reference. It drops into the same position as the zener but contains an amplifier. TI's datasheet gives a dynamic impedance of 0.2 Ω typical and 0.5 Ω maximum, and a minimum cathode current of 1 mA maximum. That is 35 times lower impedance than the 1N4733A's best-case 7 Ω, and it keeps regulating down to about a milliamp. Our 20 mA load step would move the output by about $20\ \mathrm{mA} \times 0.2\ \Omega = 4\ \mathrm{mV}$ .

Use a real regulator. An LDO draws only what the load needs plus a small quiescent current. For larger drops or currents, a buck converter stops turning the difference into heat at all.

Whatever you pick, a 100 nF decoupling capacitor across the zener is cheap insurance. onsemi's handbook notes that zeners generate noise in the knee region, which falls off as current rises, and a capacitor shunts the high-frequency part to ground.

The short version for your bench:

R_S = (V_{in(\min)} - V_{Z(\max)})/(I_{Z(\min)} + I_{L(\max)})

. Choose I_Z(min) of 5 to 10 mA or more, not the 1 mA knee figure. Check the zener at maximum input with no load, and the resistor at

(V_{in(\max)} - V_{Z(\min)})^2/R_S

. Expect the output to move by a tenth to a few tenths of a volt, and expect the voltage to read low if you run the zener far below I_ZT.

FAQ

How do you calculate the resistor for a zener diode?

Use the worst case: $R_S = (V_{in(\min)} - V_{Z(\max)})/(I_{Z(\min)} + I_{L(\max)})$ . Then check the opposite corner, maximum input with no load, for zener current and power. For 10.8 to 13.2 V in, a 1N4733A 5.1 V zener, 0 to 20 mA load and 10 mA minimum zener current, R_S comes out at 181 Ω, so 180 Ω.

What is the minimum current for a zener diode?

There is no single number. Datasheets give a knee test current I_ZK, often 1 mA or 0.25 mA, where the impedance is still hundreds of ohms, and a test current I_ZT where the nominal voltage is guaranteed, 49 mA for a 1N4733A. A practical minimum is a few milliamps for small zeners and 5 to 10 mA or more for 1 W parts, and the voltage will sit below the printed value at low current.

When does a zener diode dissipate the most power?

With the load disconnected and the input at its maximum. In a shunt regulator the resistor current is fixed by the input voltage, so every milliamp the load stops drawing goes through the zener instead. Size the zener's power rating for that no-load case, not for normal operation.

Why does my zener give a lower voltage than its rating?

The rated voltage applies only at the datasheet test current, which is 49 mA for a 1N4733A 5.1 V zener. Run it at a few milliamps and it sits below 5.1 V, because the zener's dynamic impedance is high near the knee. Tolerance of ±5% and temperature add to the spread.

What wattage resistor do I need for a zener regulator?

The resistor dissipates $(V_{in} - V_Z)^2/R_S$ whenever the zener is regulating, regardless of load. In the 12 V to 5.1 V example with 180 Ω that is 0.39 W at 13.2 V in, so use a 1 W part. If the output can be shorted, the resistor sees the full input: $13.2^2/180 = 0.97\ \mathrm{W}$ .

Sources

Vishay Semiconductors, 1N4728A to 1N4761A zener diodes datasheet, document 85816 rev. 2.7 (V_Z, I_ZT, I_ZK, Z_ZT typical, Z_ZK maximum, ±5% tolerance, 1300 mW, R_thJA 110 K/W): vishay.com/docs/85816/1n4728a.pdf
onsemi, BZX79C2V4 to BZX79C18 zener diodes datasheet, rev. 5 (V_Z range, Z_Z maximum and temperature coefficient at 5 mA, 500 mW): onsemi.com BZX79C2V4/D
onsemi, "TVS/Zener Theory and Design Considerations" handbook, HBD854/D (zener and avalanche mechanisms, impedance between I_ZK and I_ZT on log-log axes, low impedance near 7 V, knee noise, shunt regulator design equations 5 to 10): onsemi.com/pub/Collateral/HBD854-D.PDF
Texas Instruments, TL431 / TL432 precision programmable reference datasheet (|Z_KA| 0.2 Ω typical, 0.5 Ω maximum; I_min 1 mA maximum): ti.com/lit/ds/symlink/tl431.pdf
Kunkune, Zener Diode Resistor Calculator (rounds the result down to the nearest E24 value): kunkune.co.uk zener calculator
Calculate.co.nz, Zener Diode Resistor Calculator (typical single-point calculator: 12 V, 5 V, 20 mA example): calculate.co.nz/zener-diode-calculator.php

Method note: Figures 3 and 4 are direct evaluations of the shunt regulator equations with the datasheet tolerance limits. Figure 2 and the "pessimistic" regulation numbers use our own log-log interpolation between datasheet impedance points; they are estimates, not guaranteed values.

Header image: NASA (image jsc2025e015683), regraded for this article.