---
title: "Humidity leakage"
question: "How big can a resistor divider get before humid air moves it?"
url: "https://nearfieldnotes.com/videos/humidity-leakage/"
site: "Near Field Notes"
video: "https://nearfieldnotes.com/media/humidity-leakage/full.mp4"
duration: "2:08"
date: "2026-11-13"
note: "https://nearfieldnotes.com/notes/humidity-leakage/"
---

# Humidity leakage (video)

How big can a resistor divider get before humid air moves it?

2:08, with sound. [Read Note 34](https://nearfieldnotes.com/notes/humidity-leakage/) for the full answer, with figures, conditions and sources.

## Transcript

How big can you make a resistor divider?

At 10 MΩ, one breath makes a 3 V battery read 3.06.

Here's where the leak is, why a bigger part won't fix it, and the divider that can't be fooled.

Take a sensor that runs for years on a coin cell.

Two resistors scale the battery down, so the MCU can read it.

To save power, you make them huge. I had a 10 MΩ divider that moved when I breathed on it.

At 10 MΩ it uses 1% of the cell a year. At 1 MΩ, 12%.

The board itself isn't the leak.

Clean FR-4 in humid air measures over 100 GΩ.

The leak is flux residue.

No-clean flux leaves weak acids around every joint.

Dry, they insulate. In humid air they pull in water and dissolve.

On a test pattern, that's 1,600 MΩ at 60% humidity. 21 at 80. 1.6 at 90.

Breath is warm and saturated. It condenses on the board, so the surface is wetter than any room.

I simulated that film on real 0402 pads.

It acts like a second resistor across each part, about 340 MΩ.

It leaks the same across both, so an equal divider doesn't move at all.

An unequal one is pulled toward the middle. 9 to 1 reads 2% high.

Residue piles up at the joints, so a real board can be much worse.

A bigger part doesn't fix it. An 0201 reads 3% high. A 1206, still 1.3.

The gap gets longer, but the pads get wider.

A slot milled under the part cuts it to about a third.

A bare battery test point beside it more than doubles it.

The fix starts with a smaller divider.

The same film moves 1 MΩ by 0.2%, and 100 kΩ by 0.02%.

But 100 kΩ left on empties the cell in ten months.

So switch it. A FET connects the divider only while the MCU takes a reading.

Or drop it. Many MCUs can measure their own supply against an internal reference.

Some regulators read one resistor at start-up, then never look again.

And clean the board properly, or coat it. A quick wipe with alcohol can make it worse.

The same film is why 32 kHz crystals sometimes won't start.

A big divider is only as good as the board is clean.

---

Source: https://nearfieldnotes.com/videos/humidity-leakage/
