Automatic Dog Door (Part 2a) – Mechanical Problems
In Part 1, I talked about my plan for creating an Automatic Dog Door, or more accurately a Pet Door, since our cats are using it too.
I’d created a basic prototype using a stepper motor, some aluminium extrusion, an insulated board and controlled by a Particle Photon microcontroller. Sadly, It wasn’t as straight forward as I expected to go from this prototype to a working implementation.
Developing the prototype to a now installed and working solution which has passed acceptance test by both human and animal family members was more of a challenge than I expected. The problems were numerous and I’ve documented the most memorable in this series of Part 2 posts:
- Part 2a – Mechanical Problems (this post)
- Part 2b – Electrical Problems
- Part 2c – Software Problems
- Part 3 – Version 1.0 Complete
Problem 1: Attaching slide rails to the motor-driven belt
When I wrote Part 1, I’d reached a bit of an impasse as I hadn’t worked out how I would attach the stepper driven belt to the slide rail carriage. I’d seen some camera slider videos and some 3D printers which had used this approach. Hunting around I found a 3D part for the MGN12H linear slide block to attach to a GT2 belt on Thingiverse. However, I don’t have a 3D printer and quickly learned that online printing services aren’t as cheap as I thought they would be, mainly down to the minimum order charges.
Resolution: Fortunately, following a call for help, a friend John stepped in to help with my 3D printing needs. Not only did he print the parts I needed, but he also altered the way the file had been sliced in order to print a stronger design. Cheers John!
The 3D printed part allowed the slide rail carriage to connect to the belt and now we had a working conversion of rotor motion into linear motion., ready to act as a lifting mechanism.
Problem 2: Mounting motors and pulleys
My chosen 3D printed parts necessitated that the belt ran parallel to the slide rail. Which in turn meant that the stepper motor spindle would need to be facing away from the mounting board and wall. With the 20x20mm aluminium I’d prototyped with, this didn’t give me the space to mount the stepper motor.
Resolution: So I bought some 2040 extrusion for each of the side rails, doubling the distance from the backing board and providing just enough space for the stepper motor I had.
The lower pulley mechanism had to be mounted a close to the floor as possible in order to provide the range of movement necessary for the door to go low enough to cover the existing hole in the wall. I could only find flat pulley brackets, which meant they had to be mounted on to the same face of the extrusion as the slide rails, which impacted how low the cartridges on the slide rail could go. Luckily there was just enough space.
Problem 3: Door lifting mechanism
My initial lifting mechanism consisted of a large L shaped piece of aluminium which was attached to slide rail carriages either side and secured with 4 bolts into each carriage. My plan was to mount a second plate and some micro-switches in order to detect if an obstacle came in contact with the door as it was closing. However, this didn’t work. As the motor lifted from one side only, the L shaped bar would twist/skew slightly. This force, when applied to the non-driven side of the slide rails, caused increased friction which the small stepper couldn’t overcome.
Resolution: I switched to using a single slide rail and small aluminium finger to lift from one side only. This worked with the lightweight foam core board I was testing with, however, further problems arose when using a heavier material.
Problem 4: Lifting heavier materials
Believing I’d solved the lifting mechanism problem but at the expense of needing to find a new way to solve the obstacle detection I carried on and bought the A2 size 5mm thick clear perspex, I planned to use as a more substantial door which would be backed by the foam core board for a bit of added thermal insulation. That’s when the next problem arose. The weight of the perspex was considerably more than the foam core board and the stepper motor couldn’t lift the new door at all.
Resolution: At this point, I wanted to switch to 1/32 stepping, but the initial driver I’d selected only did 1/16. Hoping the smaller micro-steps would provide the additional torque to allow the door to move. I was also questioning the stepper motors reported torque and maximum current figures quoted on banggood.com. As a result, I decided to buy a pair of stepper motors to try a dual-motor drive system and at the same time get a more powerful stepper motor driver with 1/32 stepping for good measure.
The stepper motor kit I went for is quoted at 1.7A and 40Nm and went under the LongRunner brand in the UK (Amazon UK) though look identical to these on Amazon.com.
Upon arrival, I installed just the more powerful TB6600* stepper driver but the original stepper still couldn’t lift the door. Swapping out the banggood.com stepper with the LongRunner 17HS4401 stepper was definitely stronger, able to lift the door but not reliably all the way.
I already had 2 slide rails so switching to driving the lifting mechanism from both sides just required an additional motor mount and pulley hardware. Once arrived and installed the lifting mechanism worked like a dream with the dual stepper, lifting the door smoothly from each side.
* I’m aware that the TB6600 only supports up to 1/16 stepping, and that those offering 1/32 stepping are actually using a TB67S109 driver which has a lower peak current, learn more.
