Calibrating Humidty

[clippermiami]

For those who are interested in such things, you can calibrate the humidity sensor as follows:

1. in a small bottle/jar cap place a teaspoon of ordinary salt.
2. Put enough water on the salt to make it a paste
3. Place the cap in the bottom of a "Ziplock" plastic bag
4. Suspend the humidity sensor element (preferably just the element, not all the other electronics) at the top of the bag and seal it
5. Wait about 12 hours, the humidity in the bag will reach 75% and your sensor should report accordingly

I did this with the "vanilla" humidity/temperature element, the DH11, and found that it reported within about 2%. I only had the one DH11 so this might not be true of all of them.

[hek]

Thanks for the tip. You should try the same with a DHT22.

[clippermiami]

I have some on order, I think i'm going to standardize on the DH22 for my temp/humisity sensors

[Yveaux]

Most 'high quality' humidity sensors come factory calibrated (e.g. Sensirion's SHTxx).
So I doubt if calibrating them yourself will bring anything...

[clippermiami]

@Yveaux I did it because I wasn't sure about the quality of the mass-market DH11 sensor. I offered the tip in case anyone felt the need to confirm the units they get from China

[clippermiami]

@hek I received my DH22s on Friday and set up another "calibration" test. i replaced the DH11 with a DH22 in the same sensor and did the "plastic bag/salt trick." After 24 hours the results were interesting.

The DH11 registered the benchmark at 73.3, within less than 2%
The DH22 registered the benchmark at 63.2, almost 13% low.

I'm going to swap DH22s and see what happens. If the results are consistent with both then I will add an adjustment factor in the sketch. If they are widely different then ...

[clippermiami]

@clippermiami Further to this I swapped the DH22 with a second unit and replenished the salt/water misture just to make sure I have a fresh start. The second DH11 reads 55% humidity, 19% below the calibration point of 75%. It appears the while they have a much narrower humidity range the DH11s are more accurate "out of the box."

[Yveaux]

Could you repeat the test with 2 sensors exposed to the same humidity?
E.g. use the first DH11 which gave results very close to 75% together with the DH22 which was very off?
Then you also know the repeatability of your measurement.

[hek]

Just got some samples of the SHT21 from Sensirion. Damn it's small.

Just have to figure out how to surface mount it on something.

[clippermiami]

@Yveaux I want to but i do't have the tools to do it right now.

[BulldogLowell]

@clippermiami said:

@Yveaux I want to but i do't have the tools to do it right now.

How do they check out at the high end of the scale (i.e. 99-100% RH in a sealed polybag with a damp towel)?

I tested my Aeons in a temperature/humidity controlled chamber in our lab and got a spread of 10%RH points on two versus known (tested at 72F and 50%RH). I thought that was pretty poor.

I have not run any tests with my DHT22' but I may based on this.

Thanks for the info, very helpful.

Jim

[Yveaux]

@hek How are your eyes ???
I once tried soldering wires to one, but it lost a 'leg' in the process... $15 down the drain...
Better use a small breakout to solder them on!

[clippermiami]

@BulldogLowell The results with the DH22 were surprisingly disappointing, I thought they would have better performance. I tried two of them without any correction factor in the sketch and in my 75% humidity test they never got close. They were however consistent so applying a (rather large) correction factor in the code would bring them into "zero" at 75%.

In my 75% "chamber"
DH22 # 1 reported 55.8
DH22 # 2 reported 55.4

so they are pretty close to each other. A uniform adjustment of say +20 would bring them very close. However in free air the results are different

My two Aeon 4-in-1 report reading of 55% and 43% so they are pretty far apart as is. Tamiami Airport (General Aviation) is about 5 miles SW of us and it reports the still air outside humidity at 55% both yesterday and today --- so allowing for differences in inside (A/C) vs outside humidity I'd guess the actual inside at somewhere around 40-45%

In free air inside:
DH22 # 1 reported 39.0%
DH22 # 2 reported 38.6%

so they may be fairly accurate at normal room humidity levels. But applying a +20 correction factor to make them read correctly in the testbed would trow the free air levels way off.

I've put the DH11 back in the testbed to see what happens.

Of course at the end of the day I just don't have the benchmark tools to do a proper comparison at home so this is all pretty loose.

John

[BulldogLowell]

@clippermiami

wow, I am surprised at the poor results.

repeatable is important, but precision needs accuracy, too

thanks for this; it is making me re-think some of what I'm doing with sensors.

[clippermiami]

@BulldogLowell I think it would be good to have a few more folks to do some tests and report the results. I think I will start from scratch on the data as well just to verify my results

[BulldogLowell]

@clippermiami said:

@BulldogLowell I think it would be good to have a few more folks to do some tests and report the results. I think I will start from scratch on the data as well just to verify my results

I could test a couple too, if you want. I'll get them back in the humidity chamber in our lab.

I am still thinking about the salt test and wondering how the temperature may play a role (since RH is dependent on temperature). I can't seem to get myself to the point where I understand that the salt/water paste will get any (albeit tiny) air atmosphere to 75%RH independent of temperature, unless that is a "generally at room temperature" qualification.

I would like to calculate an offset. I know it is 'correct' at a saturation humidity so I imagine the offset will be some kind of formula off of the results of two or three or more known values.

Actually, that would be a great arduino project in itself... develop an offset curve with 3-4 calibration points.

[Yveaux]

@clippermiami So the 'inside' reading seems to be closer to the expected value than the 'salt chamber' value. This suggests using just an offset for calibration is not sufficient, but at least gain/offset and possibly an interpolation table are required to get some accuracy...

[clippermiami]

@Yveaux The data sheet for the DHT22 (the DHT11 is in Chinese) says that the sensor is fully calibrated and has in internal lookup table to adjust the values across the entire temperature range. S from that one would assume that the output of the device is linear.

I am just finishing up a long (48 hour) test using the salt-water-plastic-bag test jig using the DHT11. Right now it shows 76% humidity and that's after I recompiled the sketch with a a +2% adjustment. In this case +1% adjustment would have put it right on the money at 75%. The 76% value has been stable for the past 24 hours so I am satisfied that the reading is correct and would plan to use a 1% adjustment.

This morning I'm going to take it out of the test jig and let it settle to the ambient room humidity and try to judge where I am at. We shall see. But at least i know the DHT11 can correctly hit the expected test value. I may just settle on the DHT11 for my sensors.

I'm working on two 3V battery operated Minis' with Hum/Temp/Barometric/Light sensors to go outside and i'm gathering parts for 8-9 battery operated Motion/Humidity/Temp/Light Level sensors for indoor use. All will be powered by CR123A 3V Lithium batts.

[Yveaux]

@clippermiami said:

from that one would assume that the output of the device is linear.

From that one would assume the output of the device is calibrated, I would say

[clippermiami]

@Yveaux Good point

[epierre]

Hello,

I have tested the DHT11 for one day with a THGR810 from Oregon (assuming it is pretty calibrated). They were close together.

There are some time the DHT11 moved but not the Oregon. Generally it was 1% below so pretty good.

On Temperature there is more gap, it varies from 2°C+/-1°C below, it is more puzzling here.

I have not yet DHT22 to do this check-up.

Would one day adjustment values sent by the gateway to the sensors