Sensirion SHT11 experiments

awasis

I am experimenting with the SHT11 and am wondering if anyone has done any experiments to validate the sensor's accuracy and reliability in regards to data. Also what other have found about its reaction time to sudden temperature changes.·
·
I have·sealed the sensor in a jar and am currently doing experiments by heating or cooling it to see what type of accuracy and reaction time I get.· I have a thermometer in the jar with the sensor.·· I am using the program from the Sensirion pdf manual.· My objective is to see if I can use this sensor as a leak detection device for an ROV.·
·
1) I have found with this particular sensor that·temperature is off by about 1.5 C from the thermometer.· Manual specs say 0.5 C but Parallax website says, at room temp, 1 C.·
·
2) I have found that at ambient temperature in my house, when starting the sensor, it takes about 10 minutes·for·the temperature to stabilize.·
·
3) The RH reading does not follow the temperature changes sometimes.· I placed it in the freezer; sometimes as the temp goes down the RH is still going up and takes a few minutes for it to "correct".
·
4) I have also had the temp reading steady and the RH, in an instant, jumps up or down for no reason.
··
I would like to hear from anyone that had used this sensor about their experiences.
·
Thanks,
Bill

allanlane5

If you put a jar of room-temperature air in a freezer, I would guess the "relative humidity" WOULD go up as the temperature drops, at least until the moisture starts to condense out of the air.

Cold air can't hold as much moisture as warm air. Therefore, as you cool the air (with initially the same amount of moisture) the amount of moisture the cooler air can hold drops, thus raising the "Relative" humidity being measured.

I would guess the opposite would be true as well. If you fill the jar with cold air, then let it return to room temperature, the RH should go down.

Now, the RH jumping, that I don't have an answer for -- I don't know how it's being measured.

Tracy Allen

I've collected quite a bit of data on the SHT75, and when done carefully find that the accuracy is within their specification of +/- 0.5 degC at room temperature. I do the comparison referenced to an NIST traceable ASTM mercury in glass thermometer, with calibration transferred to a an LM34CAH mounted on a fine cable (low heat transfer). And testing done either in moving air or in a shielded tube (wrapped in aluminum foil). I guess my first question with regard to your tests would be, first, can you trust the reference thermometer both respect to its calibration and with respect to heat conduction via the stem, and radiation through the glass?

I'm not sure what is going on with the humidity. I can say I've found it to be pretty stable, and when it shifts I can usually trace it to an external cause. There is indeed a time lag in response between the temperature and the humidity, so when either is changing rapidly the results and dewpoint calculation can be off until it re-equilibrates.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

PatM

I have four SHTs deployed right now. Two in a portable HVAC datalogger and two in duct mounted display units. The portable datalogger is four years old now and has spent a lot of time in my trunk over winters and summers (Canadian ones). I just hauled it out last week to check the dehumidification system in a wave pool complex because we had three different humidity sensors two of which read 56% while the other read 40%. I didn't have the psychrometer handy so I used the datalogger and·both of the sensors on it·agreed with the 40% sensor. Problem was, the two 56%'ers were industrial sensors while the 40% one was a $40 "weather station" like you would buy at your local bargain store. I tracked down the psychrometer and when I checked the bargain weather station and my datalogger were both within 1% of the psychrometer!

I've had the duct mounted sensor in a different swimming pool's return air system for about two years now and its still bang on accurate. This is in one of the worst environments for humidity sensors -· very wet and highly corrosive. The old style horse-hair sensors typically last less than a year which is why I decided to try the SHT. Hell, at one one year old we had dampers corroding so badly they would seize up and Honeywell actuators burning out, despite nearly constant cleaning and greasing (wrong unit for the job).

The accuracy is pretty good, typically withing 1 degree F but even better is that the error is linear. If you're 1F out at 32F you'll still be about 1F out at 120F. So all I do is bag them up and use an ice-water mix bath then adjust the reading in software to 32F/0C.

About the only warning I have is that it is probably a good idea to put a resistor on the data pin. I'm not absolutely sure about this but I had problems with my circuit resetting itself for no apparent reason when I first breadboarded the project. I surmised that the switching from input to output on the UC pin was sometimes resulting in my UC being output high while the SHT was output low - no resistance and it dropped the voltage down to where the uC reset. I put a 120 ohm resistor in there and the problem went away. Again, I never verified this - but it seemed a low risk fix and appears to work.
·

awasis

What really concerns me is the jump and drop of the RH reading. I performed another test this morning. I let the sensor run and stabilize and then placed the whole system (jar) in the freezer. I was charting the data and everything looked great. The temp was dropping and the RH was rising, then out of nowhere the RH drops immediately from 45 % to 35 %!!!!! But the temperature continued to chart its orderly decline – no jumps or blips. Does anybody use the sensor in an application that sees large temperature swings in a short period of time? The exact same thing happens when heating the system. The ROV that I want to install this sensor in can and will experience large temperature swings due to heat generated by the thrusters and using the ROV in near freezing water. I had a test where it behaved very well for a half hour - I took the temperature up then back down and everything was OK. Then the RH dropped after the system (jar) was cooling off and reaching room temperature again. Not only did the RH drop, but after it dropped the RH started to track the temperature movement for another 10 minuets until I terminated the test! A moment before the drop, the RH was moving inversely to the temperature, which is how it should move. Do I have a bad sensor or are rapid temperature changes an Achilles heal for this sensor?

Tracy Allen

It is possible that the walls or lid of the jar hit the dew point temperature, and water condensed out of the air onto the inside surface of the jar, leading to the sudden drop of bulk humidity from 45% to 35%. Are you logging this data in a manner where you could post a graph of what happened?

I'm just saying that I'm inclined to look for thermodynamic reasons first. Nonetheless, it is true that the humidity measurement of the SHT11 deteriorates rapidly as the temperature drops below 10 Celsius as can be seen in the graphs in the data sheet for humidity and dew point accuracy.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

terry_bear

For what it's worth:
Are the symptoms repeatable? Do they happen under the same conditions eacd time? Have you tried mtg the sensor so that it faces downward in the jar?

Regards,

Terry

awasis

This is really weird, but I think I may have found some problems with this sensor. When COOLING in the freezer for about 10 minutes, the RH dropped at 12.0 C dropped from 46.7% to 33.7 %. After the RH dropped the graph was flat for a few minuets but then assumed a normal inverse relationship relative to the temperature that was continuing to drop. I then took the jar out of the freezer. After warming at room temp, right at 12.1 C, the RH shoot back up from 33.8 % to 46.7 % RH!!!!!!! If you looked at the graph, except for the RH right after the drop (which was flat for a few minutes) - if you shifted the whole middle section of the graph for the RH up, it would "match". In other words the graph of the RH overall looks consistent except that the middle portion is shifted down.

Additionally I have now two experiments that show the same behavior with first warming and then letting the jar cool off at room temp.

C RH%
Start 15.5 43.1
Selected temp data point - warming 29.6 32.8
Selected temp data point - cooling 29.6 39.7

As you can see the RH on warming phase at 29.6 C is 32.8 % but on the cooling phase at 29.6 C the RH is 39.7 %!!!! At one point, after cooling the jar, the RH actually exceeds the original RH, which since it is in a closed system, is not possible. After exceeding the original RH, and the slope of the temp and RH start to decrease, the RH shifts down a few % of RH and then decreases from that point as the temp DECREASES, which is not correct. Dew point @ 15.5 C is 3 C so I am no where near that and do not see any condensation. On the cooling phase, after the RH dropped, I let it run for another hour while returning to room temp; I stopped it at 18 C. From the RH drop to when I ended the experiment, the RH dropped as the temp dropped. I am starting to have real reservations about the practical dependability of this sensor. It seems that rapid temp fluctuations are no problem for the temp sensor but the something happened to the RH.