can somebody recommend a stable thermometer?

I tried the DHTs and they were horrible. I just tried a SHT30 and it started turning up wildly wrong temperatures.

I don't need precise, but I do need stable.

Can somebody recommend a known stable thermostat?

A thermostat is a switch !

I like the DS18b20 as a temperature transducer .

The other devices you listed do work , so not sure what problem you have.

It depends what you want to measure the temperature OF. And the range.

I like the BME280's and the BME680's.

The DHT11 might be horrible, but the DHT22 is stable for temperatures.
The SHT30 is a good temperature sensor.
There are also sensors that are accurate to ±0.1°C.

Could it be that your problem is not the temperature sensor, but something else ?
Do you use a breadboard or long wires ?
Can you tell which Arduino board you use and which library.
Please also explain what the problem is. How do the temperatures go wild ? Is it going up and down by 0.5°C ? That can be solved in the sketch with a filter. Could it be that your sketch is wrong ? Can you show your sketch ?

I choose a temperature sensor based on application, things like range and how much error is within acceptable limits. I have a DHT 22 sitting here which has been running for a few months now and is doing fine. Compared to a standard liquid in glass (Mercury) thermometer the DHT 22 is reading about 1.0 or less below nominal at 71.0 degrees F. Not all that bad at all.

The DS18b20 is another good unit so take your pick. As mentioned paying attention to wiring. The wiring is a common cause of problems, especially with breadboards.


I like the BME280's and the BME680's.

I often heard complaints on this forum about the BME280 self-heating significantly, and also the datasheet indicates that the main purpose of the thermometer is to temperature correct the other sensors (pressure, humidity).

There are several gotcha's about measuring temperature.

  1. Self-heating due to dissipation in the sensor. Typically you need to power down certain
    sensors inbetween measurements and not measure too frequently.

  2. Black epoxy packages are quite sensitive to heat radiation from the human body since
    they are not bare metal. The human body radiates around 50W per square metre in a
    normal room, meaning the reading in still air can be warped upto 2C or so if you are working
    over it. Bare metal packages reflect far-IR and are much less sensitive to heat radiation.

  3. Time constant to stabilize reading can be many tens of seconds (except for the paper-thin

It would help to know what you're seeing with the sensors you've tried, and what kind of accuracy and precision you're after. "Horrible" is not much to go on. I suspect that you may have a problem with one of your components, circuit, or coding. I've used the sensors you've mentioned and not had issues ... but then I'm not trying for .01 degree precision either.

I've used LM335s for years with good results. But they self-heat, so you need to power them only when reading.

For digital interface I like Dallas DS18B20 1-wire sensors.

Note that temperature measurement can be surprisingly tricky. The temperature in a room can vary by several degrees depending on where the sensor is placed (high or low, proximity to heat and radiant energy sources, etc). The temperature of the cable can affect your readings if it's not at the same temperature as the sensor junction (e.g. when measuring liquids). As MarkT mentioned, it can take a long time for a sensor to reach equilibrium with the ambient conditions.

This topic was automatically closed 120 days after the last reply. New replies are no longer allowed.