While tuning DC-DC converters on a few boards recently, the most annoying part wasn't picking the controller—it was the two output divider resistors: anyone can write the formula Vout = Vref × (1 + R1/R2), but when it comes to resistor values you can actually buy, I used to go back and forth against the E96 table. Then I found an online DC-DC feedback resistor selection tool: enter Vref and the target voltage, and it directly lists the standard resistor combinations with the smallest error. I wrote down the usage and a few real sessions as notes for later.
Tool URL: https://www.dd-cat.cn/tools/dcdc-feedback
Three inputs, and the output is combinations you can actually buy
There are three inputs: the feedback reference Vref (given in the datasheet; common values are 0.6 / 0.75 / 0.8 / 1.0 / 1.21 V), the target output voltage Vout, and the resistor series (E3 to E192).
What it does: first it calculates the ideal ratio ratio = (Vout − Vref) / Vref, then iterates over every R2 in the selected series, picks the R1 closest to the ideal ratio for each R2, computes the actual output voltage and error, and finally returns the top six sorted by error from lowest to highest.
It outputs the following:
- Recommended R1 (top) / R2 (bottom): the pair with the smallest error, with the actual output voltage shown;
- Alternatives 1–5: other combinations with similar error, each row showing actual Vout and error;
- Divider current: Vout / (R1+R2), with a rule-of-thumb reference of 50µA–1mA;
- Recommended combination error: the percentage error for that pair.
Next to it is a Buck feedback divider diagram with the calculated values marked on R1 and R2, Vref marked on the FB pin, and the actual voltage marked on the output.
One real use: trade-offs for a 3.3V output
The default parameters are the most common case: Vref = 0.8V, Vout = 3.3V, E96 series. The ideal ratio is (3.3 − 0.8)/0.8 = 3.125, and the tool recommends 357 kΩ / 115 kΩ, with an actual output of about 3.2835V and 0.50% error.
But there's a catch worth noting: this pair draws only 7 µA of divider current, below the tool's 50µA–1mA rule-of-thumb range. The divider branch resistance is too high, making the FB node impedance high and more sensitive to noise and leakage current. Fortunately, the "Alternatives" in the output list other magnitudes of the same ratio: 35.7 kΩ / 11.5 kΩ (about 70µA divider current) and 3.57 kΩ / 1.15 kΩ (about 0.70mA), both with the same 0.50% error.
I usually choose this way: in a normal noise environment where light-load power matters, use the hundreds-of-kΩ range; on a board with heavy switching noise and long traces, move down to the tens-of-kΩ range and trade a little more quiescent loss for noise immunity. The tool puts "error" and "divider current" in the same table, so the trade-off is easy to see.
An example where E24 actually gives zero error
Still using 3.3V / 0.8V, switch the series to E24 (±5%) and the recommended pair becomes 750 kΩ / 240 kΩ, with an actual output of exactly 3.3000V and 0.000% error. The reason is that 240k × 3.125 = 750k divides evenly, and 7.5 and 2.4 are both E24 base values—so the classic 7.5 kΩ / 2.4 kΩ (or 75k / 24k) set hits 3.3V exactly.
This shows that "higher series precision" and "the ratio happens to align" are two different things: E96 has denser base values, but the 3.125 ratio simply doesn't land on an E96 integer ratio, while E24 does. The tool computes both series, so you can use whichever is more accurate.
Three other scenarios I used
1.2V core voltage (Vref = 0.8V)
The ideal ratio is 0.5, meaning R1 : R2 = 1 : 2. The tool gives 232 kΩ / 464 kΩ, with an actual output of 1.2000V and zero error. This integer ratio hits many E96 combinations, so just pick a magnitude with a suitable divider current.
1.8V (Vref = 0.6V)
The ideal ratio is 2.0, meaning R1 : R2 = 2 : 1. The recommendation is 464 kΩ / 232 kΩ, giving 1.8000V with zero error. For chips with a 0.6V Vref (common on many low-voltage Bucks), enter 0.6.
5V (Vref = 0.8V)
The ideal ratio is 5.25, and the recommendation is 147 kΩ / 28 kΩ, giving 5.0000V with zero error and about 28.6µA of divider current. If noise immunity matters more, the alternatives include the 10.5k / 2k magnitude (about 0.4mA divider current).
Common combinations quick reference
| Vref | Target Vout | Ideal R1/R2 | Recommended pair (example) | Actual Vout |
|---|---|---|---|---|
| 0.8V | 3.3V | 3.125 | 357 kΩ / 115 kΩ (E96) | 3.2835V (0.50%) |
| 0.8V | 3.3V | 3.125 | 7.5 kΩ / 2.4 kΩ (E24) | 3.3000V (0.00%) |
| 0.8V | 1.2V | 0.5 | 232 kΩ / 464 kΩ | 1.2000V (0.00%) |
| 0.6V | 1.8V | 2.0 | 464 kΩ / 232 kΩ | 1.8000V (0.00%) |
| 0.8V | 5.0V | 5.25 | 147 kΩ / 28 kΩ | 5.0000V (0.00%) |
A few notes
- Don't only look at error when checking divider current: the tool sorts by error, and the lowest-error pair may draw only a few µA. A rule of thumb is 50µA–1mA to balance power and noise immunity; choose the hundreds-of-kΩ range when Buck light-load efficiency is sensitive.
- R1 goes to the output, R2 to ground: R1 is the upper resistor and R2 the lower one, with FB taken from the middle. Swap them and the divider relationship is reversed.
- Choose the series by tolerance: E96 is ±1%, E192 is ±0.5%, and E24 is ±5%. When tight error is required, or the best combination still exceeds 1% error, switch to E192/E96 or fine-tune the target voltage to absorb the error.
- Keep the feedback trace short: keep the FB trace as short as possible and away from switching nodes and the inductor, and add the FB filter capacitor per the datasheet. No matter how accurate the resistor values are, layout that couples switching noise in will defeat them.
- Use the datasheet Vref: 0.6 / 0.75 / 0.8 / 1.0 / 1.21 V are all common. Enter the wrong reference and every combination that follows will be wrong.
Conclusion
The hard part of feedback resistors isn't the formula—it's that the calculated ideal values can't be bought. This tool does all three steps at once: enumerating standard resistor values, calculating error, and comparing divider current. I basically take the chip's Vref and the desired Vout, enter them, pick a pair, double-check the divider current, and move on. The link is bookmarked alongside my divider and RC filter tools.