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

[openvista]

Well... it happened. Looks like my good 31 just went bad on humidity.

So that was what? Like 2 and a half weeks old? This is the new 31 in the 7714 right?

No this was the newer (9 months old) 31 in my FARS that was still accurate on humidity as of the day before. The 31 in my passive shield (16 months old) has run wet since I first started testing it months ago.

[openvista]

UPDATE: Looks like the FARS 31 has returned to calibration! Only took 48 hours 

We're presently at ~50% humidity. I guess my sensor isn't hosed (yet) and it still returns to spec after it's been below 80% humidity for some time (~9 hours).

[dendrite]

Has anyone ever tried artificially warming a 2nd SHT sensor to always keep the RH below 75% and then calculate the dewpoint from that?
 
Something similar to the warmed probe from Vaisala.

https://www.vaisala.com/sites/default/files/documents/CEN-TIA-Warmed-Probe-Application-Note-B211246EN-A_0.pdf

[CW2274]

Welp, so much for that. I got another new 31 today from Ryan that was properly packaged, ie, no bubble wrap. As is, my 8 month old 31 in use has a 2% humidity subtraction for the wet bias, and does a very nice job there, for now. The new (properly packaged 31, have you), is even wetter at ~1.3F higher dew and ~2% higher humidity than the old 31 in use. I brought the ISS inside for testing to stabilize the data and found that the old, drier 31 tracts right in line (with the 2% subtraction) with SHT11 in my Vue, dew and humidity. Point being, the Vue's SHT11 looks spot on with no correction. Davis obviously assembles both, why does the console not have the wet bias?
This QC blows, and frankly is unacceptable.

[jgentry]

Welp, so much for that. I got another new 31 today from Ryan that was properly packaged, ie, no bubble wrap. As is, my 8 month old 31 in use has a 2% humidity subtraction for the wet bias, and does a very nice job there, for now. The new (properly packaged 31, have you), is even wetter at ~1.3F higher dew and ~2% higher humidity than the old 31 in use. I brought the ISS inside for testing to stabilize the data and found that the old, drier 31 tracts right in line (with the 2% subtraction) with SHT11 in my Vue, dew and humidity. Point being, the Vue's SHT11 looks spot on with no correction. Davis obviously assembles both, why does the console not have the wet bias?
This QC blows, and frankly is unacceptable. 

It probably all boils down to how Davis mounts the 31 chip into the PCB. They should’ve done at least what RW does and others by making a temp/hum probe. But who knows. All I can say is that, as of now, the best thing you can do is to switch to the SHT-75. I’m in the process of doing just that for my two field stations and for one of the fruit growers that I know here in Chilton County, AL.

[ValentineWeather]

Welp, so much for that. I got another new 31 today from Ryan that was properly packaged, ie, no bubble wrap. As is, my 8 month old 31 in use has a 2% humidity subtraction for the wet bias, and does a very nice job there, for now. The new (properly packaged 31, have you), is even wetter at ~1.3F higher dew and ~2% higher humidity than the old 31 in use. I brought the ISS inside for testing to stabilize the data and found that the old, drier 31 tracts right in line (with the 2% subtraction) with SHT11 in my Vue, dew and humidity. Point being, the Vue's SHT11 looks spot on with no correction. Davis obviously assembles both, why does the console not have the wet bias?
This QC blows, and frankly is unacceptable.

I got one also the other day and for a new sensor it's not good.  Struggles to reach 98% and has wet bias.
I think these are old sensors maybe couple years now open air stored at Davis prior to shipping out to retailer so don't blame SI at all.
Seems some of best sensors reported are people seeing 100% coming with new stations until they go bad.

[ValentineWeather]

Welp, so much for that. I got another new 31 today from Ryan that was properly packaged, ie, no bubble wrap. As is, my 8 month old 31 in use has a 2% humidity subtraction for the wet bias, and does a very nice job there, for now. The new (properly packaged 31, have you), is even wetter at ~1.3F higher dew and ~2% higher humidity than the old 31 in use. I brought the ISS inside for testing to stabilize the data and found that the old, drier 31 tracts right in line (with the 2% subtraction) with SHT11 in my Vue, dew and humidity. Point being, the Vue's SHT11 looks spot on with no correction. Davis obviously assembles both, why does the console not have the wet bias?
This QC blows, and frankly is unacceptable. 

It probably all boils down to how Davis mounts the 31 chip into the PCB. They should’ve done at least what RW does and others by making a temp/hum probe. But who knows. All I can say is that, as of now,

Agree 100%. My acurite sht31 doesn't have the same issues.

[CW2274]

Welp, so much for that. I got another new 31 today from Ryan that was properly packaged, ie, no bubble wrap. As is, my 8 month old 31 in use has a 2% humidity subtraction for the wet bias, and does a very nice job there, for now. The new (properly packaged 31, have you), is even wetter at ~1.3F higher dew and ~2% higher humidity than the old 31 in use. I brought the ISS inside for testing to stabilize the data and found that the old, drier 31 tracts right in line (with the 2% subtraction) with SHT11 in my Vue, dew and humidity. Point being, the Vue's SHT11 looks spot on with no correction. Davis obviously assembles both, why does the console not have the wet bias?
This QC blows, and frankly is unacceptable.

don't blame SI at all.

Oh, absolutely not! I even confirmed with Ryan that I could return this one if it sucks, read above , and I will. He's even RMA-ing all his bubble wrapped stuff back to Davis, although it doesn't seem to matter....I'll see if he'll let me buy three or so and see if I get lucky with one and return the others.
Maybe if we swamp Davis with 31 returns they'll get the picture a little more in focus.

[JudinNorman]

Welp, so much for that. I got another new 31 today from Ryan that was properly packaged, ie, no bubble wrap. As is, my 8 month old 31 in use has a 2% humidity subtraction for the wet bias, and does a very nice job there, for now. The new (properly packaged 31, have you), is even wetter at ~1.3F higher dew and ~2% higher humidity than the old 31 in use. I brought the ISS inside for testing to stabilize the data and found that the old, drier 31 tracts right in line (with the 2% subtraction) with SHT11 in my Vue, dew and humidity. Point being, the Vue's SHT11 looks spot on with no correction. Davis obviously assembles both, why does the console not have the wet bias?
This QC blows, and frankly is unacceptable.

don't blame SI at all.

Oh, absolutely not! I even confirmed with Ryan that I could return this one if it sucks, read above , and I will. He's even RMA-ing all his bubble wrapped stuff back to Davis, although it doesn't seem to matter....I'll see if he'll let me buy three or so and see if I get lucky with one and return the others.
Maybe if we swamp Davis with 31 returns they'll get the picture a little more in focus.

Has anyone had any correspondence with Davis on this issue?

I'm pretty bummed out my 10+ year monitor 2 dewpoints track local weather service station while my just over a year old sht31 has gone out of calibration with dewpoints a few degrees higher usually.

[ValentineWeather]

I'm running a 75 right now (Thanks Jerry) and do see dp running lower around 2 degrees  vs SHT31's . This removes any wet bias. Problem is when I did the damp rag test for 2 hours max was only 97%.
I'm testing inside the PRO meteo shield along with a 31 and plan on doing this for few days and see what I get in the morning against the SHT31 in same shield. Also want to watch temperature. They run almost identical on temps so far in the 80-90° range.

Currently the 75 is very close to airport 1.3 mi away on DP 63°  https://mesowest.utah.edu/cgi-bin/droman/meso_base_dyn.cgi?stn=KVTN&unit=0&timetype=LOCAL/

 [ You are not allowed to view attachments ]

[CW2274]

Welp, so much for that. I got another new 31 today from Ryan that was properly packaged, ie, no bubble wrap. As is, my 8 month old 31 in use has a 2% humidity subtraction for the wet bias, and does a very nice job there, for now. The new (properly packaged 31, have you), is even wetter at ~1.3F higher dew and ~2% higher humidity than the old 31 in use. I brought the ISS inside for testing to stabilize the data and found that the old, drier 31 tracts right in line (with the 2% subtraction) with SHT11 in my Vue, dew and humidity. Point being, the Vue's SHT11 looks spot on with no correction. Davis obviously assembles both, why does the console not have the wet bias?
This QC blows, and frankly is unacceptable.

don't blame SI at all.

Oh, absolutely not! I even confirmed with Ryan that I could return this one if it sucks, read above , and I will. He's even RMA-ing all his bubble wrapped stuff back to Davis, although it doesn't seem to matter....I'll see if he'll let me buy three or so and see if I get lucky with one and return the others.
Maybe if we swamp Davis with 31 returns they'll get the picture a little more in focus.

Has anyone had any correspondence with Davis on this issue?

Oh, they're aware....

[WheatonRon]

...
Has anyone had any correspondence with Davis on this issue?

I'm pretty bummed out my 10+ year monitor 2 dewpoints track local weather service station while my just over a year old sht31 has gone out of calibration with dewpoints a few degrees higher usually.

If you go back to earlier postings in this thread, yes, Davis is aware of the issues associated with the SHT31, but to the best of my knowledge, no one has received any feedback since it was brought to Davis’ attention.

Along these lines, I bought an additional VP2 station from Ryan at SI about three months ago, and recently its 31 sensor has become unreliable too—even the normally reliable temperature is whacky. I now have three VP2s with the SHT31 about 30 feet apart so I can easily compare. And, this morning none of my stations reported 100% humidity, but all of my windows were dripping wet and nearby reliable stations showed 100% humidity!

[dendrite]

I'm running a 75 right now (Thanks Jerry) and do see dp running lower around 2 degrees  vs SHT31's . This removes any wet bias. Problem is when I did the damp rag test for 2 hours max was only 97%.
I'm testing inside the PRO meteo shield along with a 31 and plan on doing this for few days and see what I get in the morning against the SHT31 in same shield. Also want to watch temperature. They run almost identical on temps so far in the 80-90° range.

Currently the 75 is very close to airport 1.3 mi away on DP 63°  https://mesowest.utah.edu/cgi-bin/droman/meso_base_dyn.cgi?stn=KVTN&unit=0&timetype=LOCAL/

Yeah...give it a little time. I've routinely been hitting 98-99% every night with the brief touching of 100% some days. Have you also seen a big difference in the temp response compared to the 31...especially during the day with better mixing?

[ValentineWeather]

Haven't noticed yet but been busy. I did see difference in temps I'll need to check out. May post graph later.

[ValentineWeather]

Red box is when the 75 went in service. Both sensors in same Pro shield.
 [ You are not allowed to view attachments ]
 [ You are not allowed to view attachments ]

[galfert]

What software is that you are using to make those nice comparisons graphs?

[dendrite]

Randy...do you have all of those sensors online? I know your E3958 one.

[jgentry]

I'm running a 75 right now (Thanks Jerry) and do see dp running lower around 2 degrees  vs SHT31's . This removes any wet bias. Problem is when I did the damp rag test for 2 hours max was only 97%.
I'm testing inside the PRO meteo shield along with a 31 and plan on doing this for few days and see what I get in the morning against the SHT31 in same shield. Also want to watch temperature. They run almost identical on temps so far in the 80-90° range.

Currently the 75 is very close to airport 1.3 mi away on DP 63°  https://mesowest.utah.edu/cgi-bin/droman/meso_base_dyn.cgi?stn=KVTN&unit=0&timetype=LOCAL/

 [ You are not allowed to view attachments ]

So can I view your 75 data online on your website?

[ValentineWeather]

Main sensor E3958 and website click on globe for live data is now the SHT75 with no offset for humidity like I had for the 31.
The other sensors are not online. I'm using envoy8x and the software it comes with WDTU which is capable of exporting to excel for graphs.

5 min update to cwop here:
https://mesowest.utah.edu/cgi-bin/droman/meso_base_dyn.cgi?stn=E3958&unit=0&timetype=LOCAL

If you want to compare dp with airport here:

https://mesowest.utah.edu/cgi-bin/droman/meso_base_dyn.cgi?stn=KVTN&unit=0&timetype=LOCAL/

[jgentry]

Main sensor E3958 and website click on globe for live data is now the SHT75 with no offset for humidity like I had for the 31.
The other sensors are not online. I'm using envoy8x and the software it comes with WDTU which is capable of exporting to excel for graphs.

5 min update to cwop here:
https://mesowest.utah.edu/cgi-bin/droman/meso_base_dyn.cgi?stn=E3958&unit=0&timetype=LOCAL

If you want to compare dp with airport here:

https://mesowest.utah.edu/cgi-bin/droman/meso_base_dyn.cgi?stn=KVTN&unit=0&timetype=LOCAL/

Thanks!

[dendrite]

Looks like you're running the usual 2F cooler than the ASOS, but you've been consistently 2F cooler on dewpoint as well. You were fairly well mixed overnight though. I'm interested to see how the daytime dewpoint track.

I noticed your MADIS data is consistently staggered 1hr from the analysis. What's up with that? It looks like MADIS is lagging 1hr behind your station. Is this some kind of CDT vs CST issue? It looks like your graphs would be fine if you set your clocks back an hour.

https://weather.gladstonefamily.net/site/E3958

[ValentineWeather]

Looks like you're running the usual 2F cooler than the ASOS, but you've been consistently 2F cooler on dewpoint as well. You were fairly well mixed overnight though. I'm interested to see how the daytime dewpoint track.

I noticed your MADIS data is consistently staggered 1hr from the analysis. What's up with that? It looks like MADIS is lagging 1hr behind your station. Is this some kind of CDT vs CST issue? It looks like your graphs would be fine if you set your clocks back an hour.

https://weather.gladstonefamily.net/site/E3958

Madis is most likely a time zone issue. We are right on the switch over to Mountain.
This might not be the best days ahead to track dp. Later this week back into 90's I think.
Everything is 60° currently. 
 [ You are not allowed to view attachments ]

[ValentineWeather]

Looks like you're running the usual 2F cooler than the ASOS, but you've been consistently 2F cooler on dewpoint as well. You were fairly well mixed overnight though. I'm interested to see how the daytime dewpoint track.


https://weather.gladstonefamily.net/site/E3958

I noticed today my Main online live Console still had the -.9 left from using the 31 and I'm pretty sure I took it out last night. So yes the live online was .9 low until about 3:10 pm today. Can't believe I forgot it.
Not sure what's up I remember specifically taking it out at console maybe it didn't save is all I can think of. Graphs it was out for sure on the envoy8x. I do have 2 different consoles and the envoy8x anytime I do something with sensors so it gets confusing at times with this offset BS.

[jgentry]

Been keeping track of your DP readings Randy. Been impressed so far. Seems like it’s neck to neck with the ASOS. Sometimes it’s lower.

[ValentineWeather]

Been keeping track of your DP readings Randy. Been impressed so far. Seems like it’s neck to neck with the ASOS. Sometimes it’s lower.

 [ You are not allowed to view attachments ]