Humididty meters

I think using something like this would have been much easier:

I don't understand the issue?

If you want to measure dynamic properties of air such as temperature and humidity, you get inaccurate results if you do this close to a surface (such as a PCB). This is because you get a layer of dead air against a surface which doesn't move, so it may be stale and not have the same properties as the bulk of the air which is what you intended to measure. Sensors therefore need to be spaced away from expanses of circuit board to give accurate results. There can be other reasons for doing this too, such as to avoid conducting heat generated by other components to the sensor, and FR4 which most PCBs are made of does absorb and release some moisture, so this would also contaminate the dead air space which will be relevant to humidity sensing.

If you look at the design of the older SHT75 sensor, this gets around these issues by being available in what they call a pin format, which means the sensor is easily positioned about an inch away from the board, and the pin board it comes mounted on is no wider than the sensor. Sensirion did say at the time the SHT75 was discontinued that they'd look at making some of their other sensors available in pin format, but I haven't looked to see if they did.

Andrew

I have a big elaborate one of these I saved from a 1970's computer room. It has a tall perforated tube coming out of the top with horses hair stretched in it (like a violin bow). The case under this with a glass front has a circular paper graph and a clockwork drive which will rotate it once a day or once a week, and a pen plots the humidity based on the varying length of the horse hair. Another pen does same for temperature.

Never used it, but I couldn't bare the thought of it going in a skip, which was where it was destined to go. At the time, I did establish that the graphing paper disks were still available, but that was decades ago. Nowadays, I could print them of course.

Andrew

LOL. The hot wire air mass sensors in some F1 car engines were a disaster in full wet conditions.

The pitot tubes in the Air France Airbus that crashed into the Atlantic, didn't work when covered in ice.

I wonder how well it would work if you created an open ended tubular enclosure of some sort where the sensor could be on its PB near the bottom, then the pi a bit above it at the top of the tube. The small amount of heat generated from the pi would then rise and create a gentle airflow up through the tube via the chimney effect.

IME those plastic strip & switch things work very badly in practice in dehumidifiers, they just weren't stable, repeatable. Prone to sticking too. In short, don't. The hysteresis was vastly in excess of what's useful, so they got put in the output airflow to effectively reduce hysteresis. The result is a hot mess.

What a lovely bit of equipment!

I have somewhere a checklist for those horsehair hygrometer: don't blow on them, spit contains salt that throws them off. Rinse them in distilled water to clean. New horsehair is degreased with acetone(?), and sometimes rolled to flatten it

-- decrease response time. And they must regularly be run up to 100% for them to be accurate ("wet towel" over the sensor, using distilled water ISTR). I think thee's also an effect where they go up to 110% and then drop back to 100% after some time.

Also: the paper is easy. I have a temp/air pressure/humidity graphing device here, writes on a cylinder, clockwork switchable 24 h or 7 days. I have the paper -- the pens are fiber-tipped, and dry. A new set of three is well over

I'm going to try to recharge the fiber pens with stamp pad ink, one of these days...

Thomas Prufer

Yes -- they were put in because they are cheap, and use no power. Barely serviceable... but measuring humidity is tricky, and an accuracy of 3% is pretty darn good. Compare that to even an analog voltmeter, where 3% would be a crap barely-consumer-grade voltage estimator.

Thomas Prufer

They use no power, but waste tons.

They did maybe 10% on a good day. On a bad day they never turned on or off.

Really useful info which I didn't know, thanks. This one has what you describe as metal nib pens which take a drop of ink. Do you know what ink they used?

Andrew

I have them on 3 EBAC-look-a-like dehumidifiers. I've never checked their accuracy, but they seem to operate somewhat believably. The one used most is in the airing cupboard with the stat set to 30%. When a load of wet washing is put in, it runs until it's dry and then stops. Probably fires up occasionally afterwards, but gets switched off when the washing is taken out.

The control board on one of them died and it now has a raspberry pi doing the job with one of the SHT75 humidity sensors doing the sensing, and hysteresis implemented in the software, but also a 3 minute delay on restarting the compressor so it doesn't try to restart until the dynamic pressure in the refrigeration system has dissipated. Also, the defrost cycle now works using my software (DS18B20 on the evaporator), which it never did on any of the three units with the original control board, and this means the dehumidifier works at lower ambient temperatures too, without the evaporator just turning into a large block of solid ice like they did as supplied.

Andrew

Funnily I was thinking along the same lines, but thought that perhaps Andrew had already though of some cunning plan to force air over the sensor from some sort of chimney effect.

I was also thinking that if the sensor was at an elevated temperature for one reason or another, that this could be compensated by knowing the actual temperature of interest, as well as the sensor temperature. Indeed that way it could be operated in it's 'accurate' range.

If its those nibs that have a little cone shaped bottle on top ISTR any writing ink worked. Otherwise drawing ink for special pens (Rotring) can be obtained at most art shops Or here:

"barograph ink"

First hit is UK-based, 6.50 pounds for 18 months/10ml:

Thomas Prufer

Ah, yes, but they waste *your* power and the manufacturer saved a tiny bit and made that much more profit, plus the standby loss of the dehumidifier is zero, so that ticks a box!

(I'm not saying it makes sense: it doesn't.)

Thomas Prufer

For 6.50 you can get enough dye to add to non-drying oil to keep you in nondrying ink for several lifetimes.

Well, that jinxed one of them - operating for 10 years outdoors, and died two days after I posted that!

The temperature is still reporting correctly, but the humidity has gone completely off the scale. I might try baking it in an oven, which is a procedure for recovering the humidity sensor if it drifts or gets water in it, but this failure doesn't look like that.

Andrew