SparkFun Forums

[johntreacy]

I've have a HTU21D humidity sensor running for a while, that has started reporting humidity in excess of 100%. I first noticed it slightly over 100% a few weeks ago and today I notice its in the 108% range. Has anyone else seen this? Is there anything that can be done, (besides what I'm doing now, reporting the value as min(humidty, 100.0) ?)
Thanks,
John

[Valen]

The datasheet has more information on calibration of the sensor readings. It does even mention that values outside of 0% and 100% are possible due to process variations on page 14. So you'll need another way to provide reference humidity values and compare the scale against it.

[johntreacy]

I looked at the datasheet as suggested and I'm still stuck. Page 14 seems to be all about CRC error detection, not calibration. The items i've found on calibration suggest that the values could be off by a few %, but I'm now seeing values of 120% RH returned. Page 4 to me suggests that typical and maximum tolerance are 3 and 5% respectively.

[Valen]

My apologies, what I read about calibration seems to be on page 15 on the datasheet at Sparkfun's productpage. I must have misread it, or scrolled the counter just past the next page.

I'm sorry to hear that you are getting odd values. The way you make it sound like is that the top side is running away over the course of months. A faulty device is always a possibility of course. But to blame it on that you would need to explain a bit more about how or what you did and provide code examples. Maybe the problem is in there. And get a second way to measure the relative humidity to check the sensor's performance against it. A single device can never provide any trust. Not even it's datasheet. It can lie.

[johntreacy]

I just took an identical sparkfun breakout board and put it in the same location and using the same library, one HTU21D is reporting 78% and the older one is reporting 112%. I'm not doing anything fancy in the code, but using the example function in the sparkfun library.

[Valen]

Try both at the same location in a makeshift climate chamber. Like a closed plastic box with a mug of hot water. See if you can make the readings change in step when changing the humidity and temperature in it. If it doesn't respond to the trend in the same way I would consider it a failure and consider reporting it to Sparkfun customer support. But I don't know how long the warranty holds.

[johntreacy]

It wasn't quite that controlled but I did leave them in the same area outdoors for an afternoon and watched as the humidity decreased and both did decrease, but the first one was always many points higher. If I have time, I'll set up something to log each one and try to characterize the difference a little more specifically (constant, linear, ...?)