Topic: The Reliability of the SHT-31 Humidity Sensor & What Psychrometer Should I Buy?

[jgentry]

Tell ya what, this new 31 has my attention. Granted, I've only been between 5 and 37% since installing it, but the dew has been rock solid, so much so I took out my -2% correction factor from the old 31 and now run it at flat 0. Obviously still early, but really like what I'm seeing, at least on the dry end.

Nice! Can you tell any difference the way Davis mounted the sensor?

I had Ryan send me two, so I just compared the old and the other new one. With that, to my eye, absolutely no mounting difference. On the sensor side they look exactly identical, revision number and all. Only noticeable difference is on the back side with the attached white sticker, it's completely different, could just be a stock number for all I know. So if it is different, Davis is keeping it under the cuff.

Gotcha  Let me know how it performs in the long run.

I'm still tickled with it. Even though the monsoon is supposedly right around the corner, kinda wish one of you "wet" guys would give it another try. Not saying this sensor is a "new and improved version", but it's certainly performing better than my last one.
Who knows....

Let me know how the sensor performs when the monsoon hits. I’m assuming that it’s still working good right now.

[CW2274]

Well, I can tell you this, it's been in service 2 weeks now, my dew point spread has been from 12 to 53F, the humidity from 5 to 40%. Still obviously on the dry side of things, but I'm very pleased with what I'm seeing and am still happily using a zero correction factor. This is a big deal as the dew point is showing a much more linear response and hopefully will continue as the humidity rises in the coming weeks. As I said before, it would be nice for one of you more humid folks to pick one up and give it a whirl. If you don't agree with my preliminary findings, send it back, mine's going nowhere...

[CW2274]

Had a little MCS south of the border finally push some BL moisture our way, dew's been in the 50's to low 60's and this new 31 is still rockin' it. There is zero doubt in my mind this is an improvement over my last one. Now, whether that's coincidence or by design, I have no idea, but it works. If I'm lucky, I may see some actual rain today and will continue to hawk it, cause we're gonna dry right the bleep out again after today as the non-soon continues.

[jgentry]

Had a little MCS south of the border finally push some BL moisture our way, dew's been in the 50's to low 60's and this new 31 is still rockin' it. There is zero doubt in my mind this is an improvement over my last one. Now, whether that's coincidence or by design, I have no idea, but it works. If I'm lucky, I may see some actual rain today and will continue to hawk it, cause we're gonna dry right the bleep out again after today as the non-soon continues.

Nice!

[CW2274]

Had a little MCS south of the border finally push some BL moisture our way, dew's been in the 50's to low 60's and this new 31 is still rockin' it. There is zero doubt in my mind this is an improvement over my last one. Now, whether that's coincidence or by design, I have no idea, but it works. If I'm lucky, I may see some actual rain today and will continue to hawk it, cause we're gonna dry right the bleep out again after today as the non-soon continues.

Nice!

I managed a whopping 0.04" of rain on Sunday with the dew in the mid-upper 60's for a while, briefly touched 71, still looking good. Need to see some much better/longer saturation though, I imagine that's what most are waiting to see me report, myself included...

[WheatonRon]

I guess I need to order a new SHT31 (already replaced it once because of unacceptable results) for my VP2 with the 24 hour fan. See attached screenshot from CWOP showing the data from my 3 Davis VP2s. The newest one and the unit that has the 24 hour fan is FW3075. My older units, which have the daytime fan together with the SHT31, are stations CW5020 and FW4350. The ISSs are located within 25 feet or so of each other, same height on the same fence. I verified that the fans in all three VP2s are working—receiving a lot of sun today! Humidity looks reasonable on station FW3075 but not the temperature and three nearby airports confirm these results. How disappointing to say the least.

 [ You are not allowed to view attachments ]

[galfert]

If you are replacing a sensor it would be informative to know how long your sensor lasted.

[ValentineWeather]

I guess I need to order a new SHT31 (already replaced it once because of unacceptable results) for my VP2 with the 24 hour fan. See attached screenshot from CWOP showing the data from my 3 Davis VP2s. The newest one and the unit that has the 24 hour fan is FW3075. My older units, which have the daytime fan together with the SHT31, are stations CW5020 and FW4350. The ISSs are located within 25 feet or so of each other, same height on the same fence. I verified that the fans in all three VP2s are working—receiving a lot of sun today! Humidity looks reasonable on station FW3075 but not the temperature and three nearby airports confirm these results. How disappointing to say the least.

 [ You are not allowed to view attachments ]

Remember humidity is relative to temperature is why the hot station had the lower %. What are the dewpoints running on the 3 stations?
Did you do the -.9F if using an older ISS and the 31? Still doesn't explain the large temperature difference. Almost like something is not working on the 24hr fars. I would swap sensor with one of the Dfars units and see if it still reads that high.

[CW2274]

I guess I need to order a new SHT31 (already replaced it once because of unacceptable results) for my VP2 with the 24 hour fan. See attached screenshot from CWOP showing the data from my 3 Davis VP2s. The newest one and the unit that has the 24 hour fan is FW3075. My older units, which have the daytime fan together with the SHT31, are stations CW5020 and FW4350. The ISSs are located within 25 feet or so of each other, same height on the same fence. I verified that the fans in all three VP2s are working—receiving a lot of sun today! Humidity looks reasonable on station FW3075 but not the temperature and three nearby airports confirm these results. How disappointing to say the least.

 [ You are not allowed to view attachments ]

What are the dewpoints running on the 3 stations?

Yes. The only time knowing the humidity that's worth a damn is 100%.

[CW2274]

I whipped out my handy-dandy calculator http://www.dpcalc.org/ and this is what it is:
94F/54%=75F dew
93/50=72
98/43=72
Now what??

[ValentineWeather]

Humidity looks good so the hot sensor issue I lean toward shield location or fan issue. If this was a replacement sensor for the same reason especially because I've never got a new 31 out of calibration with temperature before. Guess it could happen but 2 in a row? 

[CW2274]

Humidity looks good so the hot sensor issue I lean toward shield location or fan issue.

Hey Ron, don't laugh, but I recall not too long ago that some stock fans were running backwards. Sure yours is sucking from the bottom and not blowing?

[WheatonRon]

Humidity looks good so the hot sensor issue I lean toward shield location or fan issue.

Hey Ron, don't laugh, but I recall not too long ago that some stock fans were running backwards. Sure yours is sucking from the bottom and not blowing?

I remember reading in this Forum that some Davis units were blowing rather than sucking but mine is a sucker.    Dew points were 71.6 on stations CW5020 and FW3075, but was 74.8 on station FW4350 at the time the temperatures shown previously were recorded (4:10 pm CDT). CW2274–our dew point numbers (rounded) agree—but I guess they should as I don’t believe uploads to CWOP include dew points so MesoWest, my data source, used your formula!

I have not used any offsets on my consoles as the console firmware is sufficiently current as well as communicating with up-to-date PCBAs per Davis guidelines.

Maybe I should find another retirement hobby as I am confused but am open to suggestions on what might be going on. Probably the safest answer for my station (FW3075) that has the 24 hour fan model (Davis part no. 6153) is to replace the SHT31-again, the fan and batteries and call it a day. Sending about $100 to Scaled Instruments for parts to “fix” a station that is 14 months old, seems a bit drastic, but it is easier to do this all at once than to cherry pick replacing the parts one-by-one. If I was 100% convinced the new spoon tipper was worth $50, I might replace that part too.

[galfert]

Remember some VP2 consoles need a 0.9 ºF calibration offset. Read the notes on this page:
https://www.scaledinstruments.com/shop/davis-instruments/parts/vantage-pro2-parts/temperature-and-humidity-sensor-parts/davis-7346-070-pro2-digital-temperature-humidity-sensor-sensiron-sht31/

[Tensor]

I made a small test and compared 4 years old SHT21 against 1.5 years old SHT31. Both have been outside for one year, but now they are indoors (they were not outside at the same time period).
Both sensors have been exposed to temperature ranges from -30 °C to +30 °C and very high humidities.

The results of the test are quite clear. Both sensors have excellent temperature measurement precision and accuracy. SHT31 suffers from notable dry bias, which practically renders SHT31 useless for humidity measuring. Despite from being older, SHT21 has only slight wet bias, as Sensirion SHTxx sensors often have. SHT21's humidity readings are pretty close a calibrated traditional hair hygrometer. In the other words, the real humidity level was at the test within 52-54 %.

SHT31:
Humidity: 43.35 %, Temperature: +24.70 °C

SHT21:
Humidity: 55.18 %, Temperature: +24.68 °C

When the SHT31 sensor was last time outdoors, it barely climbed above 90 % even at the most humid days. It followed very well the nearby station' humidity levels in the beginning, but after 6 months it lost completely its accuracy. Setting an empirical calibration offset could be an option, but the readings depend nonlinearly from the real humidity adn temperature. During cold weather, the dry bias seemed to be even worse.

Does anyone have experience from HYT 221? The only piece of information from Arduino forum claims that it's better than SHTs. I am considering buying one or two of HYTs for further tests.

[galfert]

Does anyone have experience from HYT 221? The only piece of information from Arduino forum claims that it's better than SHTs. I am considering buying one or two of HYTs for further tests.

You can't use the HYT 221 because it is I2C protocol. The Davis ISS is not compatible with I2C sensors. The Davis ISS uses LSS (Legacy Sensirion Sensibus) protocol. Which is why you can't use the SHT35/SHT85 since they only come with I2C protocol support. There are different versions of previous SHTxx to respectively support either LSS or I2C. Meaning you can't use just any SHT31 for example.

Interesting that you find that the SHT31 has a dry bias, as some others have reported a wet bias.

[Tensor]

You can't use the HYT 221 because it is I2C protocol. The Davis ISS is not compatible with I2C sensors. The Davis ISS uses LSS (Legacy Sensirion Sensibus) protocol. Which is why you can't use the SHT35/SHT85 since they only come with I2C protocol support. There are different versions of previous SHTxx to respectively support either LSS or I2C. Meaning you can't use just any SHT31 for example.

Interesting that you find that the SHT31 has a dry bias, as some others have reported a wet bias.

I didn't express this in my previous post, but I use Arduino and use it with sensors directly, since I don't have a Davis. This thread is just the only one in the internet, which deals with SHT31 aging and bias problems. I hope my observations are helpful. I will tell how HYT 221 works, if I happen to buy a couple of them.

[jgentry]

You can't use the HYT 221 because it is I2C protocol. The Davis ISS is not compatible with I2C sensors. The Davis ISS uses LSS (Legacy Sensirion Sensibus) protocol. Which is why you can't use the SHT35/SHT85 since they only come with I2C protocol support. There are different versions of previous SHTxx to respectively support either LSS or I2C. Meaning you can't use just any SHT31 for example.

Interesting that you find that the SHT31 has a dry bias, as some others have reported a wet bias.

I didn't express this in my previous post, but I use Arduino and use it with sensors directly, since I don't have a Davis. This thread is just the only one in the internet, which deals with SHT31 aging and bias problems. I hope my observations are helpful. I will tell how HYT 221 works, if I happen to buy a couple of them.

Please get them and let us know. It’s a sensor that I recommended Davis to consider switching too. Would love to see how the HYT would match up with a nearby ASOS or a Psychro-Dyne

[jgentry]

You can't use the HYT 221 because it is I2C protocol. The Davis ISS is not compatible with I2C sensors. The Davis ISS uses LSS (Legacy Sensirion Sensibus) protocol. Which is why you can't use the SHT35/SHT85 since they only come with I2C protocol support. There are different versions of previous SHTxx to respectively support either LSS or I2C. Meaning you can't use just any SHT31 for example.

Interesting that you find that the SHT31 has a dry bias, as some others have reported a wet bias.

I didn't express this in my previous post, but I use Arduino and use it with sensors directly, since I don't have a Davis. This thread is just the only one in the internet, which deals with SHT31 aging and bias problems. I hope my observations are helpful. I will tell how HYT 221 works, if I happen to buy a couple of them.

Please get them and let us know. It’s a sensor that I recommended Davis to consider switching too. Would love to see how the HYT would match up with a nearby ASOS or a Psychro-Dyne

A Kestrel DROP sensor would also be another good standard to compare too

[CW2274]

Nice easterly wave moving through late tonight and well into tomorrow with lots of PWAT, so heavy rainers are likely. This should finally be a good saturation test (I hope) for this new 31. My impression to this point hasn't changed, it's doing a very nice job and what saturation I have had (up to 93% and mid 70's dew), it has recovered just fine.   

[jgentry]

Nice easterly wave moving through late tonight and well into tomorrow with lots of PWAT, so heavy rainers are likely. This should finally be a good saturation test (I hope) for this new 31. My impression to this point hasn't changed, it's doing a very nice job and what saturation I have had (up to 93% and mid 70's dew), it has recovered just fine.

Good deal!  Keep me updated!

[Tensor]

Please get them and let us know. It’s a sensor that I recommended Davis to consider switching too. Would love to see how the HYT would match up with a nearby ASOS or a Psychro-Dyne

I bought a couple HYT 221s and now one of them has been outside for over 2 weeks.
The first observation is related to their mountings. SHT21 is inside a proper humidity probe, so it has good ventilation. HYT 221 is inside a punctured Kinder Surprise Egg. Also, I don't have a FARS, so HYT 221's humidity level may lag behind the truth at calm evenings. At days, the lag doesn't seem to be so large, because the days are rarely calm here.

The second thing is that HYT 221 measures nearly always lower humidity levels than the old SHT21. This is kinda expected, since almost all Sensirion sensors tend to have a small wet bias (but my SHT31 had bad dry bias). Their temperatures are within 0.1-0.3 degrees of Celsius, so both have good precision at temperature measurement.

For the third, dew point of HYT 221 seems to be more correlated with a nearby meteorological institute station (less than 1km away) than SHT21 is. Dewpoint correlation between HYT and the official station is 0.9802, whereas dewpoint correlation SHT and official is 0.9716.

At windy days, dew points of HYT221 are very close to the official station. At wet days, HYT 221 doesn't climb so easily to 99 % as SHT21 does (HYT 221 has climbed up to 99.4% so far, if anyone wonders). I don't think that HYT 221's poor casing can explain the slow climb speed fully, but I can be wrong. Maybe I should remove the Egg completely, because HYT221 has a protective grid against droplets by itself.
Both HYT221 and SHT21 report too high humidities after rain, but that's because my decrepit radiation shield absorbs water and is located near to bushes and grass.
I also think that the "official" station has a slight wet bias at 90%+ humidity levels, because it looks to report 100% even if there are no fog at all, but who knows.

Edit: I still underline, that the distributions of TD differences maybe lie a little bit: there have been many wet days recently, and then TD is almost the same as T. SHT looks to the have edge over HYT at those times, but not during dry weather. For instance: the weather is cloudy, cool and dryish now, and the difference HYT TD - official TD is +0.2 C while SHT TD - official TD is +1.3 C. In conclusion, I have to examine high level humidity measurement capabilities of HYT 221 more and of course its aging over time. SHT31 was also superb accurate at first, but became complete trash within months or so.

[jgentry]

Please get them and let us know. It’s a sensor that I recommended Davis to consider switching too. Would love to see how the HYT would match up with a nearby ASOS or a Psychro-Dyne

I bought a couple HYT 221s and now one of them has been outside for over 2 weeks.
The first observation is related to their mountings. SHT21 is inside a proper humidity probe, so it has good ventilation. HYT 221 is inside a punctured Kinder Surprise Egg. Also, I don't have a FARS, so HYT 221's humidity level may lag behind the truth at calm evenings. At days, the lag doesn't seem to be so large, because the days are rarely calm here.

The second thing is that HYT 221 measures nearly always lower humidity levels than the old SHT21. This is kinda expected, since almost all Sensirion sensors tend to have a small wet bias (but my SHT31 had bad dry bias). Their temperatures are within 0.1-0.3 degrees of Celsius, so both have good precision at temperature measurement.

For the third, dew point of HYT 221 seems to be more correlated with a nearby meteorological institute station (less than 1km away) than SHT21 is. Dewpoint correlation between HYT and the official station is 0.9802, whereas dewpoint correlation SHT and official is 0.9716.

At windy days, dew points of HYT221 are very close to the official station. At wet days, HYT 221 doesn't climb so easily to 99 % as SHT21 does (HYT 221 has climbed up to 99.4% so far, if anyone wonders). I don't think that HYT 221's poor casing can explain the slow climb speed fully, but I can be wrong. Maybe I should remove the Egg completely, because HYT221 has a protective grid against droplets by itself.
Both HYT221 and SHT21 report too high humidities after rain, but that's because my decrepit radiation shield absorbs water and is located near to bushes and grass.
I also think that the "official" station has a slight wet bias at 90%+ humidity levels, because it looks to report 100% even if there are no fog at all, but who knows.

Edit: I still underline, that the distributions of TD differences maybe lie a little bit: there have been many wet days recently, and then TD is almost the same as T. SHT looks to the have edge over HYT at those times, but not during dry weather. For instance: the weather is cloudy, cool and dryish now, and the difference HYT TD - official TD is +0.2 C while SHT TD - official TD is +1.3 C. In conclusion, I have to examine high level humidity measurement capabilities of HYT 221 more and of course its aging over time. SHT31 was also superb accurate at first, but became complete trash within months or so.

Seems to me that your set up might be affecting your results (more specifically the HYT-221 sensor). From what I’ve read on the HYT sensors, it shouldn’t age as quickly as the Sensirion sensors. Please continue on in reporting your results with the two sensors. I think the HYT-221 sensor is the best low cost sensor available for Davis to switch too.

[jgentry]

Please get them and let us know. It’s a sensor that I recommended Davis to consider switching too. Would love to see how the HYT would match up with a nearby ASOS or a Psychro-Dyne

I bought a couple HYT 221s and now one of them has been outside for over 2 weeks.
The first observation is related to their mountings. SHT21 is inside a proper humidity probe, so it has good ventilation. HYT 221 is inside a punctured Kinder Surprise Egg. Also, I don't have a FARS, so HYT 221's humidity level may lag behind the truth at calm evenings. At days, the lag doesn't seem to be so large, because the days are rarely calm here.

The second thing is that HYT 221 measures nearly always lower humidity levels than the old SHT21. This is kinda expected, since almost all Sensirion sensors tend to have a small wet bias (but my SHT31 had bad dry bias). Their temperatures are within 0.1-0.3 degrees of Celsius, so both have good precision at temperature measurement.

For the third, dew point of HYT 221 seems to be more correlated with a nearby meteorological institute station (less than 1km away) than SHT21 is. Dewpoint correlation between HYT and the official station is 0.9802, whereas dewpoint correlation SHT and official is 0.9716.

At windy days, dew points of HYT221 are very close to the official station. At wet days, HYT 221 doesn't climb so easily to 99 % as SHT21 does (HYT 221 has climbed up to 99.4% so far, if anyone wonders). I don't think that HYT 221's poor casing can explain the slow climb speed fully, but I can be wrong. Maybe I should remove the Egg completely, because HYT221 has a protective grid against droplets by itself.
Both HYT221 and SHT21 report too high humidities after rain, but that's because my decrepit radiation shield absorbs water and is located near to bushes and grass.
I also think that the "official" station has a slight wet bias at 90%+ humidity levels, because it looks to report 100% even if there are no fog at all, but who knows.

Edit: I still underline, that the distributions of TD differences maybe lie a little bit: there have been many wet days recently, and then TD is almost the same as T. SHT looks to the have edge over HYT at those times, but not during dry weather. For instance: the weather is cloudy, cool and dryish now, and the difference HYT TD - official TD is +0.2 C while SHT TD - official TD is +1.3 C. In conclusion, I have to examine high level humidity measurement capabilities of HYT 221 more and of course its aging over time. SHT31 was also superb accurate at first, but became complete trash within months or so.

BTW: How exactly close are the Dew Points between the HYT sensors and the official observations?

[CW2274]

Along those lines, I'm still waiting here for some summer storms to wet us up for more than a couple of hours, but it has been futile this year. This next few days could be some nice boundary layer saturation, if not, I'll almost certainly have to wait for the winter. By then, my "new" 31 will be 9ish months old. However, as of now, it's still performing like a champ.