Author Topic: The Reliability of the SHT-31 Humidity Sensor & What Psychrometer Should I Buy?  (Read 107500 times)

0 Members and 2 Guests are viewing this topic.

Online DaleReid

  • Forecaster
  • *****
  • Posts: 2002
    • Weather at Eau Claire, WI
Excellent discussion, some history and some fine real-world information.

Yes, we spend a lot of time reading the specs and hitching our wagon to them.

I know there must be a reason company engineers choose to go down a path, especially if the design and layout and internal programming are already done.

I know that even the expensive (for its time) ID-5001 had major problems with the humidity sensor.  Yet temperature on this, and the now-defunct TWI was usually extremely close to my precious mercury thermometers with gradations fine enough to resolve about 1/2 a degree.

Its interesting to hear the platinum wire sensor wasn't as accurate in those settings (I know we're talking humidity here) but a R M Young device has a platinum wire sensor and is very expensive so one would think that if there weren't accuracy behind it, even the US Government and Universities with deeper pockets would opt for the method that gives best results.

Anyway, I know the precision of some of the accurite home monitors has been exceptional (hard to know accuracy since I don't have any way of testing them in a controlled setting like a salt controlled humidity environment like the calibrators for Vaisala and otherss.  But over all, through humid summer and now fall and entering into heating season indoors with the floor creaking low humidity, two of them are always the same or withing digit bobble of the same.

But I'm not sure what technology they are using.   

Are all the Arduino sensors using the same technology as the SHT uses?
ECWx.info
&
ECWx.info/t/index.php

Offline johnd

  • Forecaster
  • *****
  • Posts: 4827
    • www.weatherstations.co.uk
...As johnd has pointed out numerous times, aside from a few of us who are frustrated by the humidity performance, most users are probably either oblivious or indifferent to the humidity performance, so if it hasn't been improved in 14 years, I can't imagine many resources are being put into improving accuracy.

That's a good summary and I don't really have anything new to add. TBH I doubt that this barely registers as an issue at Davis, who will respond to two main influences - the incidence of reported user issues with RH and then, if this started to become significant, the availability of any reasonably cost-effective alternative sensor.

All I can reiterate is that for our users in the UK (which is of course a relatively cool 8-) but damp climate) it is close to a non-issue. Of the several hundred Davis ISS units we will have sold in recent years, I can scarcely recall a single complaint about the T/H sensor within say 3 years. The sensor certainly does have a finite service life, but most of our users typically get 5-10 years from their sensors before significant issues are reported. NB I'm certainly not trying to argue that the SHT31 is a perfect sensor, but it may be as good as it gets in the VP2 price range. But it's also a factor that RH is not a reading that ranks as a priority for most users, in contrast say to rainfall which can generate a lot of support calls (most of which are down to issues like poor exposure or inappropriate comparisons, but Davis are clearly taking steps to improve the consistency of the rainfall readings).

« Last Edit: December 08, 2019, 03:59:42 PM by johnd »
Prodata Weather Systems
Prodata's FAQ/support site for Davis stations
Includes many details on 6313 Weatherlink console.
UK Davis Premier Dealer - All Davis stations, accessories and spares
Cambridge UK

Sorry, but I don't usually have time to help with individual issues by email unless you are a Prodata customer. Please post your issue in the relevant forum section here & I will comment there if I have anything useful to add.

Offline CW7491

  • Senior Contributor
  • ****
  • Posts: 202


Its interesting to hear the platinum wire sensor wasn't as accurate in those settings (I know we're talking humidity here) but a R M Young device has a platinum wire sensor and is very expensive so one would think that if there weren't accuracy behind it, even the US Government and Universities with deeper pockets would opt for the method that gives best results.


I can't speak to the real world accuracy of the platinum wire sensor on the analog Davis sensor because I never tested it. It may very well have performed better than spec.

I may be completely in left field, but my 31 is about 6 months old and I still think using the SF2 cap from the get go (which I did, never have before) may aid in not letting the sensor get too "wet" and delaying proper drying. Regardless, I'm still happy with it....that's my story and I'm sticking to it.

I don't think you're in left field. I have just had a completely different experience as I've documented here. You do bring up a good additional point though that the board the SHT31 is mounted to and the big filter they use are holdovers from the old analog sensor. The filter had to be that big for the size of the components. I'm guessing they stuck with the board and filter out of convenience, not maximizing performance.

Offline CW7491

  • Senior Contributor
  • ****
  • Posts: 202
The sensor certainly does have a finite service life, but most of our users typically get 5-10 years from their sensors before significant issues are reported.

That is also a good point. For any prospective VP2 purchaser, I don't think service life of the sensor is much of an issue as it was with the SHT11. From what I have seen, they will be reliable for many years, but with the inherent (and frustrating for me anyway!) limitations of the real world performance of the humidity side of their SHT31 mounted sensor.
« Last Edit: December 08, 2019, 04:54:32 PM by CW7491 »

Offline jgentry

  • Forecaster
  • *****
  • Posts: 533
I know that WeatherFlow is using the SHT-31 for their upcoming Tempest station. They did it for cost reasons and they know that they can calibrate the sensor readings if needed. Even though it would raise the cost of the VP2 stations, I do think Davis needs to make a switchover to the HYT-221 sensor. Especially since their target buyers are those in agriculture, WeatherSTEM, colleges/universities, Earth Networks, etc.
Davis Vantage Pro2 & WeatherFlow Tempest. WU: KXALJEMI2, KALJEMIS7, KFLPANAM363 & KALTHORS2. CWOP/APRS: C6353 & E6358

Offline fkapp

  • Forecaster
  • *****
  • Posts: 333
Where can one buy the SF2 cap? what is the part number called?
How are these installed and where on the board do these go? (would it be possible for someone who has done this to post a pic of where to install SF2 Cap on the board?) 
Trying to understand so that can be ready to add this when order my station and appreciate guidance.
Guess going to have to do a mod straight out of the box in the theme of accuracy.
Extra work and not sure where to add but like to do things right.

Thanks for the help,
Frank

Offline CW2274

  • Forecaster
  • *****
  • Posts: 6731
    • Conditions @ CW2274 West Tucson-Painted Hills Ranch
Where can one buy the SF2 cap?
https://www.mouser.com/ProductDetail/Sensirion/SF2?qs=wWC4CIiyLaNCAejbFUK5dw%3D%3D

Since our board does not have the holes required for the four feet, I cut them off and used the two holes available on the board to tightly cinch it down with very thin bare wire while not obstructing the membrane. The gummy substance around the sensor makes for what I believe is a water tight fit.

Offline fkapp

  • Forecaster
  • *****
  • Posts: 333
Thanks. Guess this is a must where I am located with damp foggy snow melt days etc.
Did you notice if this slowed down temperature or humidity response?
I am planning on 24 hour FARS so i guess more reason CAP is a must

One more question, did you have to remove the board from the radiation shield to install, or just take apart radiation shield?

Thanks

Offline CW2274

  • Forecaster
  • *****
  • Posts: 6731
    • Conditions @ CW2274 West Tucson-Painted Hills Ranch
Thanks. Guess this is a must where I am located with damp foggy snow melt days etc.
Did you notice if this slowed down temperature or humidity response?
I am planning on 24 hour FARS so i guess more reason CAP is a must

One more question, did you have to remove the board from the radiation shield to install, or just take apart radiation shield?

Thanks
Whether the cap is a "must" for you or anyone else, I honestly can't say, it merely seemed like a logical step to try when I installed my latest 31.
 
As far as a slow response, no, if anything I'd think it's better than the stock filter.

You'll need to pull the sensor board out of the sensor chamber to expose the sensor itself, I suggest going in from the bottom of the ISS vice the top. You'll need to pull the fan out. It may sound complicated, but once you get it apart, it's pretty much child's play. The VP2 design is a PWS's dream to work on.

Offline fkapp

  • Forecaster
  • *****
  • Posts: 333
Thanks.
So if I understand correctly the stock filter (circled blue) in below image must be removed and the SF2 CAP placed stuck into the black tar like stuff and wire run between the 2 holes circled red in image below?
Any tips on removing? Prying this off brand new board will be nerve wracking.
Guess some thin insulated eletric wire to avoid rust or something is what you used?
[ You are not allowed to view attachments ]

Offline CW2274

  • Forecaster
  • *****
  • Posts: 6731
    • Conditions @ CW2274 West Tucson-Painted Hills Ranch
Thanks.
So if I understand correctly the stock filter (circled blue) in below image must be removed and the SF2 CAP placed stuck into the black tar like stuff and wire run between the 2 holes circled red in image below?
Any tips on removing? Prying this off brand new board will be nerve wracking.
Guess some thin insulated eletric wire to avoid rust or something is what you used?
[ You are not allowed to view attachments ]
The stock cap merely snaps out. The holes are inside the cap next to the sensor, you'll see. As far as rust, never crossed my mind, however I live in one of the most dry climates on the planet. Just check it occasionally after install.
BTW, read about handling the SF2 on the link I sent, it can make it a little challenging to keep ones fingers off the membrane, I suggest buying at least several just in case. Plan on little needle nose pliers for the assist.
« Last Edit: December 08, 2019, 07:11:19 PM by CW2274 »

Offline fkapp

  • Forecaster
  • *****
  • Posts: 333
Thanks again appreciate your guidance and time.
Any Feedback on how the CAP works in more humid climate (how are response times, does this delay the wet bias, based on your experience recommend this extra work/Mod) from those that have tried this would be helpful?

Appreciate it as plan to go VP2 route for my new station, now want to fine tune best can.
Frank

Offline CW2274

  • Forecaster
  • *****
  • Posts: 6731
    • Conditions @ CW2274 West Tucson-Painted Hills Ranch
Thanks.
So if I understand correctly the stock filter (circled blue) in below image must be removed and the SF2 CAP placed stuck into the black tar like stuff and wire run between the 2 holes circled red in image below?
Any tips on removing? Prying this off brand new board will be nerve wracking.
Guess some thin insulated eletric wire to avoid rust or something is what you used?
[ You are not allowed to view attachments ]
The stock cap merely snaps out. The holes are inside the cap next to the sensor, you'll see.
Upon reflection, I'm pretty sure the holes the stock filter snaps into are the holes I used for the wire. Merely trying to avoid confusion.

Offline galfert

  • Global Moderator
  • Forecaster
  • *****
  • Posts: 6822
I'm not an engineer, but I would think from the reading that there are a limited number of ways for a humidity sensor,  of the types that electronic stations use and excluding blonde hair or horse hair gizmos of the 40s, can work.

That being said, why would a sensor be substantially different from others, if the design were nearly the same.  Why would one be more sensitive to high humidity damage than others?

I know that Young and Vaisala have some very long life, extreme conditions, types of humidity sensors.  Are their construction that much more expensive than the glut of low cost things that are like the SHT stuff or the little almost pennies, ones for Arduino etc?

Anyone know the science behind it?

I wrote to Sensirion and asked them exactly what you were wondering. Here is their response.

Quote from: Sensirion Technical Support
Thank you for your inquiry.

There are many things that can influence the accuracy and lifetime of the sensor, especially in outdoor applications.

We used to have a definition of normal operating range with the SHT2x:

 [ You are not allowed to view attachments ]
Quote

Working outside of the normal range you may have to consider additional effects.

One important document is our handling instructions.
http://www.sensirion.com/file/handling_instructions_rht

This document is listing some points to consider, but they're not too relevant for outdoor conditions. The VOC levels are usually very low.

The biggest challenge outdoor are the dew cycles that can lead to corrosion. The electronics need to be protected with conformal coating and the manufacturers have to protect the sensor during that process.

Additionally some regions can be more harmful to the sensor. Airports, heavy traffic areas or proximity to the coast is also known to have adverse effects on sensor lifetime.

I haven't heard many reports about those failures. Do you have more in-depth knowledge if those products applied conformal coating or not?

Are there weather stations equipped with Vaisala sensors that last longer?

Does anyone know what Vaisala uses for temp/hum sensors so that I can respond back to Sensirion?

How about knowledge about what type of conformal coating chemistry Davis uses? (Acrylic, Epoxy, Polyurethane, Silicone, Fluorinated or non Fluorinated - Poly-Para-Xylylene (Parylene), or Amorphous Fluoropolymer)
« Last Edit: December 09, 2019, 12:54:25 PM by galfert »
Ecowitt GW1000 | Meteobridge on Raspberry Pi
WU: KFLWINTE111  |  PWSweather: KFLWINTE111
CWOP: FW3708  |  AWEKAS: 14814
Windy: pws-f075acbe
Weather Underground Issue Tracking
Tele-Pole

Offline Old Tele man

  • Singing in the rain...
  • Forecaster
  • *****
  • Posts: 1365
What? "Normal" range only goes up to 90% RH? Apparently SENSIRION avoids FOG!
• SYS: Davis VP2 Vue/WL-IP & Envoy8X/WL-USB;
• DBX2 & DBX1 Precision Digital Barographs
• CWOP: DW6988 - 2 miles NNE of Cortaro, AZ
• WU - KAZTUCSO202, Countryside

Offline ValentineWeather

  • Forecaster
  • *****
  • Posts: 6364
    • Valentine Nebraska's Real-Time Weather
What? "Normal" range only goes up to 90% RH? Apparently SENSIRION avoids FOG!

Another way of saying our sensors aren't for the serious enthusiast.
Randy

Offline CW7491

  • Senior Contributor
  • ****
  • Posts: 202


The biggest challenge outdoor are the dew cycles that can lead to corrosion. The electronics need to be protected with conformal coating and the manufacturers have to protect the sensor during that process.


I haven't heard many reports about those failures. Do you have more in-depth knowledge if those products applied conformal coating or not?

Are there weather stations equipped with Vaisala sensors that last longer?


Does anyone know what Vaisala uses for temp/hum sensors so that I can respond back to Sensirion?

How about knowledge about what type of conformal coating chemistry Davis uses? (Acrylic, Epoxy, Polyurethane, Silicone, Fluorinated or non Fluorinated - Poly-Para-Xylylene (Parylene), or Amorphous Fluoropolymer)

I find the part he mentions about conformal coating interesting, specifically that the manufacturer must protect the sensor during the process. Those of us unhappy with humidity performance of the SHT31 compared to say Acurite's use of the SHT31 have long suspected that it might be something with the conformal coating in Davis' manufacturing process. I know Acurite does not have a conformal coating on their sensor as it is designed as an indoor sensor. And I get better performance with the SHT75 than the Davis SHT31 and that obviously has no conformal coating. I'm wondering if Davis does not take whatever precautions Sensirion recommends during the process if it may induce an offset and if Sensirion's reconditioning procedure of baking the sensor might help. (Maybe partly why folks in really hot, really dry environments seem so much happier with the performance). Unfortunately, Sensirion would need to know what Davis uses for the conformal coating to answer that question as you ask and I'm not sure Davis will be willing to publicly release that info. I also believe Valentineweather tried the baking procedure without much success.

Here is what Vaisala uses in their humidity sensors ...
https://www.vaisala.com/sites/default/files/documents/HUMICAP-Sensors-Datasheet-B211748EN.pdf

Looks kind of like the old analog sensor, but much more robust ... [ You are not allowed to view attachments ]  [ You are not allowed to view attachments ]
« Last Edit: December 09, 2019, 01:38:59 PM by CW7491 »

Offline fkapp

  • Forecaster
  • *****
  • Posts: 333
 deleted as mis read chart earlier and dont want to create confusion by my misunderstanding.

Thanks guys


Offline CW2274

  • Forecaster
  • *****
  • Posts: 6731
    • Conditions @ CW2274 West Tucson-Painted Hills Ranch
Seeing 90% is normal range cap, have to rethink things...  unreal they would go with 90% as normal cap sensor.
I don't even know what that means. Every properly working 31 will easily get RH's in the upper 90's, my last would reached 99% all the time, it was the below 5% range that caused me to replace it with this current 31. :roll:

Offline galfert

  • Global Moderator
  • Forecaster
  • *****
  • Posts: 6822
:shock:or is it  #-o
Is the above normal range chart for the SHT31?

Wow if yes. or perhaps the Davis custom sensor was wider range, I dont  know anymore guys...

It is the same for the SHT3x series. They just don't call it normal range. Instead the spec sheet shows that it looses accuracy (max tolerance) above 90%.

 [ You are not allowed to view attachments ]
« Last Edit: December 09, 2019, 04:16:45 PM by galfert »
Ecowitt GW1000 | Meteobridge on Raspberry Pi
WU: KFLWINTE111  |  PWSweather: KFLWINTE111
CWOP: FW3708  |  AWEKAS: 14814
Windy: pws-f075acbe
Weather Underground Issue Tracking
Tele-Pole

Offline fkapp

  • Forecaster
  • *****
  • Posts: 333
That makes more sense. above %90 get less accuracy which matches 90% is upper limit of normal variance in above chart.
Thanks for clarifying I mis read/ miss understood pervious chart.


Offline galfert

  • Global Moderator
  • Forecaster
  • *****
  • Posts: 6822
I responded to Sensirion and told them that I had reached out to Davis inquiring about their conformal coating strategy. I have not heard back from Davis regarding this yet, but if I do I will share it and also forward it on to my new contact at Sensirion. I also shared with Sensirion the pictures from the Vaisala station. The purpose being to identify technological differences in something a Vaisala station would use that as a complete station costs thousands of dollars. I then mentioned the frustration Davis users have in that Davis doesn't support I2C sensors and that we wonder what the Davis strategy is now that the LSS type of SHT31 sensor is no longer in production, and to see if and when we might see a Davis I2C type station. I inquired if there was a contract between Davis and Sensirion to keep providing these LSS type sensors past the end of general availability and you can see the response below that basically he didn't know what the Davis strategy is.

Below is the additional response back from Sensirion:

Quote from: Sensirion Technical Support
I do have some additional comments.

The normal range doesn't tell you it's not possible to operate the sensor outside that range. There are just side effects that have to be considered. For instance at high RH levels the RH reading may drift and that's a reversible effect. It's just getting more complicated when operated outside of normal range.

I've heard of people wanting to replace the sensor of Davis stations. And as you state correctly that's not possible because the LSS version isn't available.

I don't know about Davis strategy and why they don't change to the DIS [I2C] version. It might involve a lot of extra work and compatibility issues making it hard to support previous versions after changing the design.

From what I've experienced so far this doesn't seem to be a huge issue with people wanting to replace the sensor themselves. I've only received a few requests so far and haven't been concerned. But I understand that for those affected by a non-working sensor it would be much easier to replace it themselves.

Looking at the Vaisala picture I think it only shows the sensing element. It seems to be porous material and most likely based on capacitive technology as well.

Given the fact that the sensing element is much bigger than the thin material layer on a CMOS chip, an eventual degradation process would take longer to affect the readings.

As it's still capacitive technology it would still be affected by molecules of a similar size as water such as VOCs.

If you receive feedback from Davis I'd be happy to hear about it. But I doubt they'll provide detailed information about their strategy and eventual change to the I2C interface.
« Last Edit: December 10, 2019, 08:29:36 AM by galfert »
Ecowitt GW1000 | Meteobridge on Raspberry Pi
WU: KFLWINTE111  |  PWSweather: KFLWINTE111
CWOP: FW3708  |  AWEKAS: 14814
Windy: pws-f075acbe
Weather Underground Issue Tracking
Tele-Pole

Offline CW7491

  • Senior Contributor
  • ****
  • Posts: 202
I responded to Sensirion and told them that I had reached out to Davis inquiring about their conformal coating strategy. I have not heard back from Davis regarding this yet, but if I do I will share it and also forward it on to my new contact at Sensirion.

Hopefully Davis have read this ...
https://www.sensirion.com/fileadmin/user_upload/customers/sensirion/Dokumente/2_Humidity_Sensors/Sensirion_Humidity_Sensors_Handling_Instructions.pdf

Sensirion note that volatile chemicals in many epoxies and adhesives cause an irreversible offset specifically in humidity readings. Something no reconditioning procedure will fix. They even go on to recommend 3 types of conformal coatings to use with their sensors.

For those with the SF2 or considering it, they also recommend adhesives that can be safely used including Loctite 401
« Last Edit: December 10, 2019, 10:42:59 AM by CW7491 »

Offline CW2274

  • Forecaster
  • *****
  • Posts: 6731
    • Conditions @ CW2274 West Tucson-Painted Hills Ranch
I responded to Sensirion and told them that I had reached out to Davis inquiring about their conformal coating strategy. I have not heard back from Davis regarding this yet, but if I do I will share it and also forward it on to my new contact at Sensirion.
For those with the SF2 or considering it, they also recommend adhesives that can be safely used including Loctite 401
I considered using approved glue but frankly decided it was too difficult to do properly without making a sensor destroying mess (damn thing is soooo small), hence my decision to use wire to seal it down.

Offline CW2274

  • Forecaster
  • *****
  • Posts: 6731
    • Conditions @ CW2274 West Tucson-Painted Hills Ranch
Well, after 6 months, I finally got my wish last night. It went W0X0F about 2am and stayed that way for at least several hours. My 31 would not report above 98% which equates to a temp/dew spread of 0.5F (0.4F is 99%). So, although it didn't reach the magical 100%, it does, as far as I can tell, conform to the stated 2% of accuracy from 1-100%, and in between, as I've been to 1% as well. Am I happy about it not pulling the 100% trigger, obviously not (at least my last 31 would hit 99%), but it does do it's job as specified. As far as recovery, I see no issues.
It is what it is...

Testing done....well at least til I buy another.... #-o