Here is my recent project, a sensor for measuring the sugar level in fermenting apple cider. My friend uses it to monitor his fermentation at his small cidery.
These devices float in beer and measure the specific gravity which is correlated to the angle that the (offset weighted) device floats at. Sugary, higher density liquid means more buoyancy, so the devices float more horizontally and out of the liquid. Alcoholic, lower density liquid means the devices float more vertically in the fermenting cider.
The iSpindle uses WiFi, the Tilt uses Bluetooth, which only barely work, if at all, in large steel tanks.
So I wanted to try the beefiest sensor radio that I could think of, Lora, and try something that could be prototyped easily.
The two prototypes at the cidery
This is the inside of a purple sensor: a RadioBridge LoraWAN dev kit, stainless steel weight, and HDPE spacers to position the weight and sensor appropriately. I had to cut this sensor open to replace the battery. (I tried a screw-top pipe but it frequently leaked.) The LoraWAN module was misconfigured and was sending too man high spreading factor messages that take longer to send and use more battery life.
Here is the example web app that shows the current and recent readings using the The Things Network LoraWAN stack. (TTN)
With this Lora-powered design, the user only needs one gateway (TTN Indoor Gateway - $89) for the whole cidery, instead of multiple bluetooth receivers mounted close to each tank, as the other products require. (Or even an in-tank antennae)
Here is an example of the calibration curve from sensor angle to specific gravity.
What I learned from this project:
- Making a repeatable mechanical product requires jigs and practice!
- Calibrating a non-linear response take work and you don’t want to do it for each sensor.
- It’s hard to make a pro-looking waterproof, foodsafe enclosure
- It was a good call to use a dev kit to test the product idea
After this test I’m not that keen to make this into.a product for sale. First, there are IP issues, as the Tilt seems to have a key patent. Also, relying on a cloud service like TTN seems like an unnecessary dependency on the global internet when you just want a sensor reading to get out of a tank of cider.