Very simple overview, I am wanting/hoping to make some kind of monitoring system for our pet feeder (petlibro single automatic hopper) to see when and how much our cat actually eats as she goes in wild swings sometimes. My thoughts, before considering feasibility, was so e kind of small load Cell that monitors the full weight of the hopper + bowl (one unit) and the food within the hopper and the bowl. Basically if we put 1kg of food in the hopper, it could be in the hopper or it could have been dispensed to the bowl and still be uneaten. Tare the measurement with a full hopper and then log when the overall weight drops which would imply the food has not only been dispensed but also eaten, which I would then send back to either our NAS or Google sheets or something.
Have never successfully trie/accomplished any of the above etc, only Arduino project I made was a small solenoid trigger but I've read up enough to be dangerous. A pitfall that I think I see is since the load cells would be constantly under some load (probably like 1.5kg full, 1kg empty or something) forever, they will experience drift/creep and lose their reference point. The ideal operation condition is one where every few weeks the hopper gets loaded with dry food and then Slowly over the course of a few weeks the food is consumed. The weight would decrease slowly but never to zero for more than a minute or two out of a month while reloading.
Questions are as follows:
A1) am I in the right spot to start for the forum?
A) is that assessment correct, load cells would not be ideal in a long term monitoring situation?
B) is there a better/different sensor option out there that I am not thinking of? I can figure these systems.out slowly but total noob at this overall so it is entirely possible I'm starting from the wrong place?
C) would something like solenoids triggering as jackstands to hold up the load and then lower it to the load Cell on a set monitoring interval be a reasonable work around (or simpler options available)?
D) is there actually already a product/system commercially available (and reasonably priced) that could do this monitoring job? Needs to be low profile so I don't make the feeder bowl 1' off the ground? Would be cool to build something here but I'm not married to needing to if there is a better solution.
I assume that you have written the programme by yourself, then it is quite easy to find the error.
There's a trick to figuring out why something isn't working:
Use a logic analyzer to see what happens.
Put Serial.print statements at various places in the code as diagnostic prints to see the values of variables and determine whether they meet your expectations.
So I have not actually written anything yet, so no errors to speak of. I am more asking as a hardware starting point, are the strain gauge load cells suitable for the use case or will there be immediate problems encountered with loss of calibration and drift on the sensor as it slowly creeps due to sustained load.
Consider weighing the whole assembly including the dinner table. Then take your measurements and when you determine the pet is not at the table use those measurements. You will then end up with approximately the amount of food your pet ate.
Constantly loaded load cells are used around the world in long-term monitoring situations (grain hoppers, process vessels in refineries, etc.)
I think load cells would be an ok approach for your problem and that a system to occasionally lower the load onto the cells would be an unnecessary (but fun!) complication for a hobby project.
Just keep something in mind. Cheap off the boat load cells drift. Quality temperature compensated load cells, quality brand name load cells do not drift (much). You get what you pay for in load cells. They have come a long way over the years.
We recently had a thread by a bee keeper who wanted to measure honey production in hives by weighing the hives. A problem was cheap off the boat load cell. It was cheap or inexpensive but was not well suited for the application.
Next consider which type load cell best fits your plan mechanically be it compression or tension (push or pull). You also want to look at what your load cell outputs or load cell sensitivity.
My friend had two cats; one dominant and one submissive. At feeding time the dominant cat would not allow the submissive cat to eat from its bowl, so the submissive cat pawed individual kibbles out of the bowl, onto the mat, the floor, and finally around the corner to eat, leaving most of the food on the mat and floor. The dominant cat died, yet the submissive cat continued this habit of eating. I started feeding the cat directly, letting it know it did not need to sneak food. I prefer to learn behavior, first hand.
Good to know on the cheap vs quality load cells, if a suitable long-term product exists I am happy to pay for the good stuff. Being new to this, any particular vendor/source in mind for the top tier sensor? I see the ones on Amazon for 1-5kg being like $7-20, don't want to pay $500 (hopefully) but more than $8 on Amazon is totally fine.
Data in hoping to collect is if there is any noticeable pattern both short term (hours days) and longer term (weeks). We just have one cat, rescued her in feb and due to a heart condition she can't be spayed. Typically behavior is to eat less when in a heat cycle (which is like every other week for her), but for her it seems to be the opposite. For the first few months she was like a rocket to the feeder every time it went off (5 times a day) more so when In heat, less so when not. Lately, she rarely sprints to the food and will let 3-5 feedings pile up while snacking on it. We do measure (manually) at various intervals but I'd like to see on a continuous basis what she's doing and try to correlate it with any other behavior or where she's at in her heat cycle. Per our vet she's got a clean bill of health so no serious concerns and she never just stops eating. It is possible since we rescued her from one of my construction sites that she just feels totally safe now And not at all worried about food (was never good aggressive, just happy to eat when the hopper went off).
Figured it was a cool project to dip my toes into this stuff and useful to get some more data. If the quality sensors will work (even if they have to be replaced every so often) I'll definitely give it a run