Xiaomi Mijia Bluetooth-compatible Thermometer 2 Electric Humidity Smart Home Wireless Hygrometer LCD Digital Moisture (WHL #91)
Some new gadgets have been sitting here for a while – couldn’t be arsed to write about them last week 
It’s pack of three Xiaomi Mijia Temperature + Humidity Monitor 2 (what a name), product number seems to be LYWSD03MMC. They cost me 15.58€ in December, inclusive of shipping and indeed one CR2032 coin cell per unit. Units from German distributors are more like 10 to 15€ each with negligible volume discounts – and shipping cost on top of that.
Anyway, here’s an explosion view from the unit, likely made by Xiaomi and not the seller:
I’ve been using other cheapo thermometer/hygrometer units for the past, well, probably 10 to 15 years, they’re really inexpensive at sub-2€ each and run on LR44 batteries basically indefinitely. And when I started 3D printing, I ordered a batch with one of each available color for my air-tight filament containers.
Since my prediction of “I’ll only ever use 4 spools at a time” was about as correct as a well-known dude was about not ever needing more than 640k of RAM in a computer, I needed to re-order. And with the recent post about the decomposing Sunlu yellow PLA, which by the was has now fully desintegrated into 10cm pieces of curved spaghetti, I was curious about the accuracy of those things. Temperature, well, that’s easy – but what about humidity? They drop to 10%rH within a very short period of time and just stay there. No humidity payload of “wet” filament will put them over that threshold value, so I can’t really tell if the silica gel actually draws out moisture or if the filament is already “dry”. That would have been an interesting thing to watch on the Sunlu PLA.
So I did some research and found the Xiaomi units. They are Bluetooth capable and apparently can control other devices, but I have no desire to do so (nor do I have compatible devices…I think). But the humidity display is interesting, since unlike the vast majority of meters out there, this one has a “0%一99%rH” measuring range at 1% steps. And they are still relatively cheap, especially when ordered from abroad.
Here’s the units about an hour or two after they’ve arrived and settled. Both temperature and humidity is within one least significant digit across all meters, which is excellent. Doesn’t have to be the true value – but I appreciate the small overall spread. I’d rather measure (slightly) wrong all the time and be consistent across all devices than have one really accurate one and the others drifting about like there’s no tomorrow.
19.5°C, 64%rF
19.5°C, 64%rF
19.4°C, 64%rF
Nice, innit?
Update rate is a couple times per minute, so that’s dope. Once per minute would have been enough, but I guess it doesn’t really eat batteries all that much. The LCD (static view and updates) will cost next to thing nothing, and the controller is optimized to go back to sleep immediately – otherwise they couldn’t be anywhere near their 1y battery claim.
Well, and here’s the things after two entire weeks in an undisturbed 3.4l container that also holds half a spool of black pre-dried PLA besides 130g of freshly baked silica gel:
20.1°C, 16%rH
20.2°C, 16%rH
20.1°C, 16%rH
…and the old cheapo meter says 19.8°C and of course 10%rF
According to the interwebs, silica gel isn’t terrible good at driving out the very last bit of moisture, and that is okay for 3D printing filament. I found some document from Sanner Group, apparently a specialist for desiccants and other packaging needs. It looks like some internal document, since it’s a quick Word and Excel mashup and not really polished to the level that I’d expect from a medical company, but the figures are probably accurate. Units aren’t – likely should be g/m³ (grams of water per cubic meter of air) without the Newtons in there.
At an eyeballed 20°C, the silica gel (“SG”) will not exceed 5% of moisture absorption relative to its own weight at an ambient level of 1g/m³ of H2O. That should be doable in my freshly baked silica gel from the example above, and those 1g/m³ at 20°C equate to 5.8%rH. So with a two week wait time and the containers being reasonably air-tight, I would expect the air inside to drop to those 5-ish percent of relative humidity. And it probably did – but the Xiaomis will not show that. If they’re really capable of going sub-10%rH, this is a) not the way to do it, and b) they’ll not be accurate at all. Maybe I should get a container of zeolite from my former company, that stuff sucks moisture out of the air under any circumstances. You can literally see scales move when that stuff is fully dried, it’s fascinating.
The meters are also VERY slow to drop below 20%rH – sure, that’s a rare occurrence in regular ambient air (outside of deserts or something), so the sensor can be optimized for the usual conditions. But it would be useful for my application, and they are specified down to basically zero humidity – which they will not read and not reach in time. Funnily enough, they got down to 14/15%rF after two days, and slowly crept up ever since. I don’t think however that the silica gel has drawn out so much moisture from the already dry filament that it has lost capacity to dry the contained air, and ambient temperature hasn’t changed all that much, since it’s winter time and the central heating does its job.
Looks like my plan of watching paint dry filament dry has failed, as those hygrometers do not have the precision needed to observe the process, and also not the temporal resolution to do that in a useful manner. So I will not exchange the existing cheapo meters with Xiaomi units since they’re simply not better suited for the intended purpose – and also more expensive. But maybe I can integrate one of them into my iobroker smarthome monitoring system? They got Bluetooth, my Raspberry Pi got Bluetooth, it’s worth a shot.
