Resistor divider and E-series solver
Searches every standard pair for the ratio you actually want, then shows what tolerance does to it.
Best pair
—
—
Runners-up
| R1 | R2 | Output | Error | Current |
|---|
Ranked by error, then by lower divider current where errors are close. A slightly worse ratio with half the standing current is usually the better part.
The physics
The divider
V_out = V_in · R2 / (R1 + R2)
Z_source = R1 ∥ R2 = R1·R2 / (R1 + R2)Two resistors, one ratio. The part that is not obvious is that the ratio and the impedance are independent choices: scaling both resistors by the same factor keeps the output identical while changing current, noise and settling behaviour completely. That is the freedom this tool searches.
Why searching beats rounding
Suppose you want 0.275. Fix R1 at 10 kΩ and the ideal R2 is 3.793 kΩ; the nearest E24 value is 3.9 kΩ, giving 0.2806 — an error of 2%. Search both, and E24 offers pairs landing inside 0.2%. The ratio is a two-dimensional search, and treating it as one-dimensional throws away most of the available precision.
Tolerance is not symmetric with the ratio
V_max = V_in · R2(1+t) / (R1(1−t) + R2(1+t))
V_min = V_in · R2(1−t) / (R1(1+t) + R2(1−t))Worst case is R1 low with R2 high, and the reverse. The resulting band is close to but not exactly twice the component tolerance, and it is always wider than people expect. For a regulator feedback divider this band sits directly on top of the reference tolerance, and together they set the real output accuracy.
What tolerance does not cover
- Temperature coefficient. Two 100 ppm/°C resistors that drift in opposite directions over a 60 °C swing add another 1.2% of ratio error. Matched pairs and networks exist for this reason.
- Self-heating shifts a resistor's value while it is running, which shows up as a slow settle after power-on.
- Input bias current from whatever the tap drives flows through the parallel impedance and offsets the reading. At 1 MΩ source impedance even a few nanoamps is millivolts.
- Board leakage across a dirty or humid surface parallels your divider with an unknown resistor. Above about 1 MΩ this stops being theoretical.
Where this stops being accurate
- Assumes the tap is unloaded. Any real load parallels R2 and shifts the ratio.
- Tempco, ageing and self-heating are not modelled — they often exceed the initial tolerance.
- E-series values are the standard decade sets; not every value is stocked by every distributor in every tolerance and package.
- No AC behaviour. Parasitic capacitance at the tap matters for fast signals and for switching-regulator feedback.
Design guidance, not a manufacturing instruction. Verify every result against your fabricator's stack-up and the applicable standard before release. Closed-form models are approximations; the fab's field solver and process window are the authority.
Questions
What people ask about this
Why not just calculate the ideal value and round?
Because rounding one resistor to the nearest standard value throws away the freedom you have in the other. Searching both simultaneously routinely finds a pair an order of magnitude closer to target than rounding does — the ratio is what matters, not either value alone. That search is the entire point of this tool.
What does the tolerance band mean here?
The exact-value error assumes perfect resistors. Real ones vary, and two 1% resistors can push a ratio by considerably more than 1% in the worst case. The band shown is the true worst-case output across both tolerances. If the band is wider than your requirement, no amount of clever value selection helps — you need tighter parts, or a different topology.
How do I pick the overall impedance?
Two competing pressures. Low values waste current continuously, which matters enormously in a battery device — a 10 kΩ divider across 12 V burns 1.2 mA forever. High values make the node vulnerable to input bias current, leakage and noise pickup, and slow down anything with capacitance on it. For a feedback divider follow the regulator datasheet. For an ADC input, keep source impedance inside what the sample-and-hold can settle.
Should I add a capacitor across the top resistor?
On a switching-regulator feedback divider, often yes — a feedforward capacitor adds phase lead and improves transient response. On an ADC input, put the capacitor from the tap to ground instead, to give the sampling capacitor a low-impedance source. They look similar on a schematic and do opposite jobs.
Precision analog that has to hold over temperature?
Divider ratios drift with tempco mismatch, self-heating and layout. We design the signal chain and prove the error budget.