Humidity sensors other than HIH-4000?


New Member
I've been running a 1-wire weather system for years built out of pieces and parts - some Hobbyboard, some homebrew.  I've always measured humidity along with everything else but never tried to verify how accurate it was until recently when I noticed that my outside sensor was consistently -- and significantly -- different from a NOAA station located very close by. It seemed highly compressed, meaning that when they were reporting 99% I was seeing low 80's; when they reported 30% I was seeing mid forties.
So I took two sensor boards (both Hobbyboards) and mounted them side by side indoors and tracked them for awhile. They differed significantly in output. I was using the same voltage to raw humidity (linear) conversion parameters, and the same temperature compensation factors for both boards. They also differed significantly in their internal temperature measurements, even though the two boards were screwed 1-inch apart with standoffs. One reported temperatures consistently about 2-3 degrees C higher than the other.
Reviewing the Honeywell literature (application notes, published spec sheets) I found that there were some differences (and inconsistencies) in their temperature compensation recommendations. But the linear voltage to raw rh calculation was the same - for both the older HIH-3610 sensors and the newer HIH-4000 sensors.
Then I tried to independently calibrate the sensors using known constant humidity environments, which can be created using various saturated salt pastes (I'm not going to hazard a link here at this point). I picked four salts spanning relative humidities of 11 to 88 percent. It was an educational PIA that provided me with some very good raw data which I used to optimize my own set of custom calibration parameters for each of my three humidity sensors. And now the two inside sensors pretty much agree with each another (and the outside one with NOAA).
But -- and this is the reason for this post -- I am concerned for two reasons. 1) why are all of these sensors so far off spec? Honeywell claims 3.5% accuracy with less than 0.4% drift per year. These parts are way off that. I'm using a hub with boosted +14V so I don't have power problems. The sensors are seeing about 5.05 volts each.
2) how long will my calibrations hold up? If these parts are drifting that much, they may be off again in six months. I don't want to have to go through this calibration process again.
So I'm wondering if there are other higher quality sensors out there people are using. I did a little bit of a searching but haven't seen anything that really stands out.
I have tried the HIH sensor as well as many different brands of thermostats, and 8 CAO Tags and they all wander all over place. I have digital hygrometers that vary 20 percent against each other.

I don't think you will find a stable unit.

Home Depot has calibation kits for 33 and 70? Percent using chemical moisture stabilisers .

Sent using Tapatalk
Also when you stand in front of a stat on the wall and adjust the calibration, expect the registered humidity to change by about 5 to 8 percent from your breath over the 30 seconds it takes.

Sent using Tapatalk
Here utilize the combo sensors from HAI and Omnistat, 1-wire original sensors from Midon and HB, "el cheapo" arduino style sensors and Dallas Weather station.
All are close in readings but not the same.
Sensitivity is a good thing in most cases - certainly for weather station applications. I'm not standing in front of the sensor breathing on it and I want it to respond to changes in it's environment without a lot of lag. But sensitivity is orthogonal to the qualities of accuracy and drift. These are published specifications by an industrial manufacturer whose big customers expect them to perform as advertised. I'm trying to understand if they make sense. I have three parts bought over a span of maybe six years. The oldest of these is probably about 10 years old plus whatever time it sat on a shelf after being manufactured. The don't have a date stamp evident anywhere, so that's an unknown.
They say it's supposed to be accurate to +/-3.5% absolutely with a drift of 0.2% per year. So lets say 10 years, that's 2% drift on top of the tolerance. So I should be looking at a maximum inaccuracy in the range of +/-5.5%. I'm pretty sure I'm seeing differences way outside that.
What sensors are used in the HAI and Omnistat products? And, has anyone worked with the capacitive sensor HS1101LF. Looks like an implementation would require external power to charge it fully, and a frequency bridge, but I have the power. It looks like it might be a little more accurate than the Honeywell and possibly more stable. Curious if anyone has worked with these.
It looks like the Home Depot kit consists of two salt pastes: they are obviously using Magnesium Chloride (33%) and table salt (75%). Current price for that is $61 USD. I bought three salts from a science supply outfit online and rock salt at the hardware store, all for under $20. That gave me four calibration points (11%,33%,75%,88%). There is a description of the procedure at for anyone that is interested. If I do it again I'm going to add Potassium Carbonate, which will add another point at about 48%.