The SHT was telling me the room was 20°C. The Fluke and DHT were saying 18°C. The room did not feel like 20°C. I know which ones I trust.
Also the relative humidity reported by the SHT is way off, because it is temperature compensating from the wrong temperature. And I know this as my dehumidifiers cut out at 55%, and the DTH agrees.
I don’t know if I got a good batch of DHTs or what but I’m sticking with them. The only downside is that they are a little slower to react than the SHTs.
BTW, both sensor were well isolated from the ESP to prevent heating by it (they were on leads outside the ESP enclosure).
Interesting…
Do you have originals or chinese “fakes”?
I have 5 sht31 and 5 Bme280 (chinese) and all bme’s show pretty much same temp, also all sht’s show same temp, but all sht’s are a good degree lower than bme’s. If readings would be random, ok, i understand, but since all same ones are equal, i don’t know who to trust. That’s why i intend to buy an original to be able to compare and calibrate.
Sensors, bought in aliexpress, banggood (even ebay at some sellers) are more or less all “fakes”. Even if they are originals, they are definitely chips which didn’t pass internal quality control (and therefore they weren’t factory calibrated) - therefore wrong reading. You must take into account that original in adafruit costs around 15 USD (SHT or BME), while these sensors at aliexpress cost in the range of 1-2USD per piece… they just can’t be originals at that price.
If you buy SHT31 at adafruit it will definitely show correct temperature, as said: SHT has ±0,3 degrees guaranteed temp error, humiditiy somewhat more, but still pretty good.
Oh, well…
now i’m baffled… if all that is true then there’s no point of wasting 15 USD for sensor which also “lies”…
The point is which one to trust? I’d just like to “calibrate” my existing sensors, so that they will show at least kind-of correct temperature… i can’t sink them in boiling water, neither in freezing one…And, also, temperature of boiling water is dependant on altitude,so another error to compensate…
i guess i’ll just connect all my sensors, record readings and then draw a line in the middle…
I could easily calibrate the display of the temperature sensor however this temperature is also used internally to calculate the relative humidity and so this will still be wrong. I have no desire to fiddle around with salt solutions to calibrate the RH display (and I suspect it will be a nonlinear equation).
FWIW, my AM2301 and BME280 attached to the same ESP8266 are always within 0.1 to 0.3 degrees of each other. My SHT’s have arrived a month or so ago but I haven’t been arsed to play with them yet…
Thanks for info! I just ordered TMP117 and i’ll be able to check/calibrate my other sensors! That’s what i’ve been looking for… i also ordered oringal BME280. I’ll see it’s accuracy, compared to “fakes”…
I looked and read regarding this “accuracy”… it turned out that many sensors have pretty high accuracy, but quite lower PRECISION, which basically means that they ARE accurate along specified range, but they can LIE quite a bit through while range. What i mean is: sensor can be accurate, say, 1%, but it can deviate from correct temperature for, say, 2 degrees and this deviation is then “accurate” within that 1%…
What this means is only that they are not factory calibrated, so their reference can be off… (TMP117 is, BTW, each and every one calibrated, supposely…).
I’ve got myself TMP117 “reference” sensor. So i compared it’s temperature with all my (chinese) sensors.
I also bought one original BME280 and compared pressure and humidity with it (let’s say that original one is accurate). Interesting thing is that ORIGINAL BME280 (marked as “1”) shows quite lower temp than reference…
Here are my results. Surprisingly all my chinese BMP280 shows pretty accurate temperature, while all 3 SHT31 are practically “spot on”…
Numbers show offset: -1,1 means that sensor shows 1,1 deg lower than reference.
Me again…
i’d like to hear your opinion on this one:
like i said in previous post i bought TMP117 and one original BME280. After a while it turned out that original BME280 doesn’t have only 1.1 degrees lower temp reading (like i posted above) but whooping 2.5 degrees. It seem that i didn’t wait enough when i “calibrated” it first time…
Both sensors are definitely original adafruit ones (from Mouser via my local dealer).
Is this normal? I mean, now i begin to wonder - who to trust - which one is actually accruate: TMP117 or BME280 (or neither…)? TMP should be inside 0.1 degrees… right?
I mean: i have no problem setting offset to all my sensors (that’s why i bought precision TMP117 for), but, as said: which one is “the one”?
You both wrote in this topic that you weredoing some further tests. Could you share your results?
I’m looking at what cheap sensor I can buy which gives decent results.
So far I have tried one HTU21D and two AHT10 sensors all from aliexpress.
For one AHT10 I have decent results, the other AHT10 I can’t get to work.
The HTU21D is around 4 degrees celsius off and in terms of humidity 30%, so unuseable.
I was thinking of buying the SHT31-D but read that @tom_l has bad results with them, and I couldn’t find much about the SHT40.
Haha, and that whilethe guy doing the massive comparisons says the DHT are poor.
I was also considering this one which combines the AHT20 with the BMP280. Thay way I can average the results of both.
That’s reassuring.