What a reading costs
The sensor is one of the most frugal parts in the whole TinkerBlock line — under two microamps averaged over a reading a second. The power light beside it draws about a hundred and sixty times that, and on a battery project it is the only number on the board that matters.
The sensor's side of it
Sensirion's numbers, and they are remarkable:
| Current | |
|---|---|
| Idle between single-shot readings | 0.2 µA |
| While measuring | 600 µA |
| Averaged, one low-repeatability reading a second | 1.7 µA |
| Idle in periodic mode | 45 µA |
A measurement is expensive and short. At high repeatability it is 12.5 ms of 600 µA, which averages out at about 7.7 µA if you take one a second; at low repeatability it is 2.5 ms, which is the 1.7 µA in the third row. Between measurements the part genuinely stops.
The last row is worth noticing. Periodic mode costs 45 µA just to sit there, because the sensor is running its own timer so it can measure on schedule without being asked — more than two hundred times the 0.2 µA it idles at when you drive it yourself. For anything battery-powered, single shot on your own schedule is the mode you want.
The board's side of it
R4 is 5.1 kΩ and LED1 is a red 0603 with no part number on the bill of materials, so its forward voltage is an estimate — somewhere near 1.9 V at these currents. That gives about
(3.3 V - 1.9 V) / 5100 Ω = 0.27 mA = 270 µAat 3.3 V, and roughly 610 µA at 5 V.
The estimate hardly matters. Any plausible forward voltage between 1.6 and 2.2 V lands the answer between about 215 and 330 µA, and every number in that range is more than a hundred times the sensor's own average. The light is the board's power consumption. The sensor is a rounding error on it.
What that does to a battery
Take a 2000 mAh cell — a pair of AAs, or a modest lithium pack — and one reading a second.
- As shipped: about 272 µA, so roughly 300 days.
- With LED1 or R4 removed: about 1.7 µA, so a figure in the decades, which in practice means the cell's own self-discharge is what ends the project rather than your circuit.
That is the whole article. The difference between a sensor node that needs a new battery once a year and one that outlasts the battery is one 0603 component, and it is not the sensor.
When it is worth doing
Almost never, and then absolutely.
On USB, or on anything with a mains supply behind it, 270 µA is invisible and the light is useful — it is the fastest way to know VCC arrived, which is the first thing to check when nothing answers. Leave it.
On a coin cell, an AA pack or a solar node, take it off before you build the enclosure. There is no software way to do it later, and it is the single largest thing you can change about this board.
One thing to remember if you go further and switch the whole board from a GPIO pin: the pull-up resistors lose their supply along with everything else, so while the board is off both bus lines float. If anything else shares that bus, give it its own pull-ups, or switch only the boards that are alone on theirs.
When it does not work
No. LED1 and R4 are wired straight across VCC and GND with nothing in between — no transistor, no GPIO, nothing the sensor controls. The only way to switch it off is to remove one of the two components, or to switch the whole board's supply from a spare pin.
Either works and R4 is easier: a 0603 resistor lifts cleanly with a fine iron and a little flux, and it is the one whose absence is obviously deliberate if you come back to the board in a year. Take LED1 off instead if you would rather keep the resistor pad free to bridge later.
Because it is not idle in the same way. In single-shot mode the sensor really is doing nothing between commands — 0.2 µA. In periodic mode it is running its own timer so it can measure on schedule without being asked, and that timer costs about 45 µA whether you fetch the data or not.
Five times more per reading — 12.5 ms of measuring instead of 2.5 — and at one reading a second that is the difference between about 1.7 µA and about 7.7 µA averaged. Both are dwarfed by anything else on a real project, including the power light, so it is not a choice worth agonising over until you have removed the LED.
Usually yes, and check the numbers first. The whole board including the light is well under a milliamp at 3.3 V, which is inside what an ESP32 or an AVR pin can source. Give the sensor a millisecond to power up before talking to it, and remember that the pull-ups lose their supply too — so the bus lines will float while the board is off.
Five ways this board fails, three of them producing exactly the same silence, told apart by what the two bus lines are doing.
When the scan finds nothing →Edit this page — content/books/sht31/what-a-reading-costs.mdx
Questions about this product
See what other owners have asked, and read their solutions.
SHT31 Temperature and Humidity Sensor
Loading discussions…
Discuss this article
Ask about this page. The answer stays here, on the page it belongs to, for whoever hits the same wall next.