The Truth About Voltage Dividers as Power Supplies

Every few weeks, someone posts a schematic in an electronics forum where they've replaced a proper voltage regulator with two resistors. "I need 3.3V from a 5V supply," they say. "Can't I just use a voltage divider?" And then twenty experienced engineers type the same reply, trying not to sound condescending.

The frustrating part? The math looks so clean. You have 5V. You want 3.3V. Slap two resistors together in the right ratio and — boom — you're done. Except you're not. Not even close.

Why the Textbook Example Lies to You

Here's the classic voltage divider formula most people remember from their first circuits course:

Vout = Vin × (R2 / (R1 + R2))

To get 3.3V from 5V, you might pick R1 = 510Ω and R2 = 1kΩ. Let's check: 5V × (1000 / 1510) = 3.31V. Beautiful. Exactly what you need.

Now plug in your microcontroller. That microcontroller draws, let's say, 50mA when active. Watch what happens.

The moment you connect that load, it appears in parallel with R2. Your 50mA load at 3.3V looks like a resistance of 3.3V / 0.05A = 66Ω. Now R2 isn't 1kΩ anymore — it's 1000 in parallel with 66, which works out to about 61.8Ω. Re-run the divider formula:

Vout = 5V × (61.8 / (510 + 61.8)) = 5V × 0.108 = 0.54V

Your "3.3V supply" is now delivering 0.54V. Your microcontroller never boots. This is what engineers mean when they talk about loading effects — and it's the core reason voltage dividers are not power supplies.

The Fundamental Problem: Source Impedance

A proper power supply — whether it's a linear regulator like an LDO or a switching converter — has an extremely low output impedance. When your load demands more current, the regulator adjusts its internal behavior to keep the output voltage steady. That's literally the job of a regulator.

A resistor divider has a source impedance equal to R1 and R2 in parallel (from the perspective of the output). In our example, that's 510 ∥ 1000 = roughly 337Ω. Every amp of load current causes a voltage drop of 337 × I across that source impedance. With 50mA of load, you drop 337 × 0.05 = 16.8V across the source impedance — more than the entire input voltage. The numbers don't lie, they just lie about 3.3V being achievable here.

To make a divider "stiff" enough to supply real current, you'd need to make R1 and R2 very small — say, 5Ω and 10Ω. Now the source impedance is around 3.3Ω, much better. But now your divider is permanently pulling 5V / 15Ω = 333mA even when your load is drawing nothing. That's 1.67W of heat wasted in your resistors, all the time, doing nothing useful. Your "efficient" solution just became a hand warmer.

When Voltage Dividers Actually Work

Here's where I want to push back on the reflexive "never use a divider" crowd, because that's equally wrong. Voltage dividers are genuinely excellent — just not for powering things.

Sensor signal scaling. If you have an analog sensor outputting 0–5V and your ADC only accepts 0–3.3V, a voltage divider is the correct tool. The ADC input draws essentially zero current (usually microamps or less), so loading effects are negligible. This is exactly the use case the formula was designed for.

Setting reference voltages into high-impedance inputs. Comparator reference pins, op-amp non-inverting inputs in voltage follower configurations, bias points for transistor bases (when the base current is small relative to the divider current) — these all work because the impedance looking into those inputs is high enough that they don't load the divider significantly.

The "10x rule" of thumb. A practical guideline: your load impedance should be at least 10× larger than the Thevenin resistance of the divider. If your divider's Thevenin resistance is 1kΩ, the connected load should present at least 10kΩ. Some people push this to 100×. The tighter your voltage regulation needs to be, the more margin you need.

Resistive voltage sensing and feedback networks. Inside a regulator IC, there's often a resistor divider used to feed a sample of the output back to the feedback pin. This works because that feedback pin is essentially a voltage sense input — it draws almost no current. The resistors aren't powering the load; they're just measuring it.

The Zener Diode Partial Solution (and Its Limits)

Some folks try to rescue the divider idea by adding a Zener diode across R2. Set the Zener voltage to your desired output, and it clamps the voltage even as load current changes. This is actually a real circuit — but it's still not suitable for powering most loads.

A Zener regulator works best when the load current is small and relatively constant. The Zener needs to be in conduction at all times to maintain its voltage, which means R1 must be sized to supply both the Zener current and the maximum load current simultaneously. If load current varies widely, the Zener current swings, and the output voltage shifts with it (Zener impedance isn't zero). The efficiency is poor, and thermal management becomes an issue quickly.

For powering anything serious, you're better off with an LDO regulator. The MCP1700, AP2112, or AMS1117 family all cost less than a dollar, require only a couple of capacitors, and provide regulated output with dropout voltages often under 300mV. A Zener regulator is appropriate for very low-current reference applications or when you're in a pinch with limited components.

A Real Scenario Where This Bites People

The most common victim I've seen is the "5V to 3.3V level shifter" built from two resistors to pull down a 5V signal line for a 3.3V GPIO input. People reason: "I just need to drop the voltage, not source current, so a divider is fine, right?"

Almost. The problem is bidirectional GPIO pins. When the 3.3V device tries to drive the line high, it now has to push current through R2 and charge through R1 — and if those resistors are large (to save power), the rise time gets sluggish. At MHz signaling speeds, you get signal integrity problems. At very low frequencies or for one-directional input-only signals, the divider works. For SPI or I2C buses? Use a proper level shifter (74LVC1T45, BSS138 MOSFET, or dedicated I2C level shifting ICs).

How to Calculate a Divider You Can Actually Trust

If you've confirmed your application is genuinely high-impedance, here's how to size the divider properly.

First, determine your load impedance. If it's an ADC input with 1MΩ input impedance, you have plenty of headroom. If it's a 10kΩ pull-up resistor on a GPIO, you need to factor that in.

Set your divider current to at least 10× the maximum expected load current. This keeps the voltage shift from loading effects under about 10%, which is acceptable for most sensing applications.

Choose resistor values with that current target in mind: R_total = Vin / I_divider. Then split them in the ratio that gives you the desired output.

Example: Vin = 12V, want Vout = 3.3V, load is an ADC drawing 100μA max. Target divider current: 1mA (10× safety factor). R_total = 12V / 1mA = 12kΩ. Ratio: 3.3/12 = 0.275, so R2 = 12k × 0.275 = 3.3kΩ, R1 = 12k − 3.3k = 8.7kΩ. Real values: 8.2kΩ + 3.3kΩ or use a 10kΩ potentiometer for trimming. This divider will stay within about 3% accuracy under full ADC load.

The Bottom Line

The voltage divider formula isn't wrong — the formula is perfect. The error is in mistaking a voltage measurement principle for a power delivery mechanism. A divider divides voltage under no-load conditions. The moment you draw current, the voltage moves, and it moves fast.

Real power delivery requires a circuit that can actively regulate — sense the output, compare it to a reference, and adjust a control element to compensate for load changes. That's what regulators do. A passive resistor network simply cannot do this.

Use voltage dividers to sense, scale, and set reference points. Use regulators to power things. Once you internalize that distinction, a whole class of mysterious circuit failures suddenly makes sense — and stops happening.