Vending Machine for Birds

I got busy with other stuff for the last 6 months or so and did not have much time to work on this project. Things will settle down in the summer and I plan to get back to work on the ESP32 version. I've been doing a lot of design work in the background and have a few I I want to try out. ESP32s are very capable uC's and there was (and still is) a lot for me to learn to make effective use of their capabilities.

In the meantime, I've sent the VMFB I was using to Crowbuddy. He is also in NL, has already designed a machine of his own, and plans on working with a local high school on a related project. Crowbuddy has a very robust, installation-type device that is solar powered and has been deployed in several public areas in Europe. I hope we can collaborate later this year.

So it's not over yet :)

I moved the dispense opening close to the deposit opening and modified it so peanuts would not spill out onto the platform. The goal was to force the magpies to walk on the platform instead of just using the rim. This way they would be more likely to sink some accidental depostis.

Unfortunately they are pretty adamant to not walk on the platform if it can be avoided. They would rather touch it with their wings than their feet. Something to work on...

I redesigned the entire interior of the feeder to make it smaller, easier to build, cheaper and somewhat ironically more reliable. I've already had a few problems that I hope I've fixed.

The place I get peanuts changed suppliers or something and the peanuts are a bit bigger now. I had to cut away a bit of the baffle over the dispense opening once I found an entire hopper of peanuts jammed above it thanks to a few finger-sized peanuts that got stuck.

I want to change the design yet again to put the deposit hole in front of where the nuts get dispensed. I think this will increase the likelyhood of accidental deposits and the chance that the birds see the result (getting a nut) so they can reason it out. 

I initially moved the deposit chute away from the dispense opening so dispensed nuts would not find their way into the deposit chute. This happened once when I was still on shelled peanuts and pigeons. The dispense cup had a few peanuts in it and the ones coming out were bouncing off them into the deposit chute causing more peanuts to come out. 

The next design will not dispense nuts onto the platform - the nuts will stay witiin the enclosure and it will have a larger dispense opening (50x100mm) facing forward so birds can see and reach in easily. The deposit opening will be in front of thiis opening at its center.

This change is more about getting results from the birds than any mechanical or material efficiency. I've noticed that while jackdaws walk around on the platform and accidentally deposit sometimes, magpies typically stick ot the rim of the platform and seem to avoid walking on it. Hopefully this change will lure them closer and increase the chance they associate deposit with dispense.

Since I can rotate the dispense opening on my current design I can take a bit of a shortcut and just lower the dispense baffles so the nuts stay within the dispenser.

Then I can rotate the opening closer to the deposit hole. The birds will take a good look inside to make sure they didn't miss anything. Moving dispense and deposit closer together will require the magpies to walk on the platform to get to the nuts, and have a higher chance of accidentally depositing and figuring out the cause and effect.

I've been having more dispenser jams with the new design and the reason turn out to be the increased baffle angle. I made it steeper thinking it would make sure the peanuts would roll off of it onto the vibration platform. The side effect is more lateral force on the peanuts. Peanuts in the shell are relatively light (compared to shelled nuts) and have a rough, irregular exterior - even a little lateral force is enough to jam them up.

So I'll be going from ~30° incline back to a ~15° incline to fix this. Otherwise, everything is working well.

UPDATE

The angle of the baffle is somewhere between 20° and 30°, about the minimum for unshelled peanuts to reliably roll off of it so none get left behind. On the old baffle, there would be a few peanuts sitting on it once the hopper was "empty". Instead of reducing the angle I cut a couple square cm from it and raised it about one cm to leave more clearance for the nuts to pass by.

This seems to have fixed it - no more jams within the dispenser and quick, reliable dispensing of 1-2 nuts per dispense.

UPDATE 2

The angle of the feeder also matters. Since I simplified the mounting system, it is possible for the feeder to be mounted not quite vertical. In this case it was leaning a few degrees away from vertical making it more difficult for peanuts to reach the dispense opening. Adjusting this reduced the dispense time and increased reliability (no jams).

Testing the new design has exposed some issues I want to fix in the short term. Rather than go straight to migrating to another platform I'm going to make the following changes first.

New PCB

• Add a dual channel DAC to replace the two pots and enable automated and remote sensor calibration - the DAC is cheaper than one pot and brings more functionality.
• Improve the header arrangement to make it easier to work with.
• Use connector sockets to group wires instead of having a bunch of separate wires - will make connections more robust and easier to connect. Maybe use locking headers.

Mechanical

• Enlarge the dispense opening - a huge (finger-sized) peanut jammed yesterday and took me 30 mins to unjam remotely
• Check the pitch of the dispense baffle - I think it is too steep now and encourages jams

I also want to look into some simplification.

• Make enough internal space to use a 110mmx50mm 90° fitting instead of a 110x110mm 90° fitting with a 50mm adapter. The sliced-pipe approach to baffles also lets me create spaces within the enclosure to mount electronics. This will reduce size and remove a part from the BOM.
• Try to eliminate the baffle in the dispense pathway. It's mainly for blocking light from entering the enclosure

Previously I was focused on getting the birds to cooperate. Now that I know they will I am getting more focused on improving reliability, simplification and optimizing the BOM.

After a couple more tweaks, I turned it on and left it in timed dispense mode. The magpies and jackdaws remember it as a food source (even with a slightly different enclosure) and are visiting it already.

I extended the dispense pathway tube to allow for a better fit for the mounting bracket and cleaned up the hose clamp mess by removing two of them. - now three hose clamps secure the mounting bracket and two of those also secure the platform. The fourth one below them secures the deposit assembly.

The mounting setup is improved a bit as well. You can mount it on a wall or fence with just one screw. Three adjustable feet provide additional support and levelling.

There are of course a couple more tweaks to the enclosure I want to make, but this is pretty close to final form as far as 110mm PVC goes. So it's open for business. I hope to make some progress with these birds in the coming weeks.

So I made the following improvements to the enclosure and internals.

• The feeder is now 86cm (34in) tall (down from over a meter on the previous version)
• Fewer parts and materials (and shorter wires) required to build
• PIR sensor moved to camera arm
• Dispense path is more difficult to jam
• Deposit pathway is shorter and easier to mount the sensor on
• Better way of placing internal baffles

I've tested it, but have left it turned off for now. I'll first turn it on and let it run on a day I can work from home to monitor it. It took a bit of tweaking to get everything to fit while keeping it simple.

Here's what it looks like now vs what it looked like before.

Now

Before

I tried using one of those pole mounts for lights and cameras, but in reverse. I got this for 11 EUR on Amazon.

It works prety well, but the feeder still needs a secondary point of support below the mount. I used a leftover bit of 110mm PVC pipe for this, but will hopefully think of a better way soon. If you are looking for hose clamps, sorry, I'm using all of them right now.

The enclosure is now basically 4 sections. Here they are, and the parts they're made of.

1. The deposit assembly.

2. The dispense platform

3. The electronics

I was too lazy to disassemble the camera arm - it consists of:

• 110mm to 50mm adapter
• 50mm 90° elbow joint
• ~10cm length of 50mm PVC pipe
• ~3cm 200° arc of 50mm PVC pipe to secure the PIR lens
• 50mm PVC end cap
• small piece of transparency to protect the camera lens

4. The hopper and dispenser

Have just worked out the kinks with the new enclosure and improvements.

• The feeder is now less than 3ft tall
• Fewer parts and materials (and shorter wires) required to build
• PIR sensor moved to camera arm
• Dispense path is more difficult to jam
• Deposit pathway is shorter and easier to mount the sensor on
• Better way of placing internal baffles

It's still using a RaspberryPi Zero W v1.1 to run the show. After field testing these improvements I'll resume working on porting this to an ESP32.

Will include some pics of the enclosure parts and assembly in next log update.

Summer was busier than I expected, and I had no time for this project. Now things are getting back to normal and I'm almost done with the mechanical and enclosure improvements I wanted to make. I expect to have it back on the balcony in a week or so.

On another note, while I was on vacation I was introduced to a site called AllAnimalApp. It has all kinds of resources for animal care, medical attention, rescues, shelters, pet-friendly facilities and more. Very useful if you have questions, are travelling with a pet or have found an animal that needs help.

I think I can stop providing food on the platform after this week. There have been several accidental deposits daily for the last week or so and the birds are starting to add it up.