Summary

TLDR:
After real world testing with real mice I have come up with critical improvements to an otherwise very good trap. These parts are drop-in replacements for the original parts that prevent mice escaping the first chamber.
The front door is easier to install, un-grabable from the inside, and closes tighter due to cam action.
The seesaw works in concert with the front door cam, uncomfortable spikes adds ballast weight to the rear making it more sensitive and reduce mice loitering in the first chamber.

Recommend counter-intuitive slicing the seesaw in vertical orientation so that layer lines do not interfere with closure. Mechanisms should operate smoothly at the slightest touch.

2025.10.17 added modified main body and bait key - body is shorter saving material, the extra height doesn't seem to be necessary. Widened and lowered the back door to encourage mice to move from main chamber to holding chamber. Added seesaw v5 with more intense spike pattern.

Excerpt from my blog: https://mem.memoriarius.com/memos/ETWq7SFWx86DXqjYe7EtV2

Using test mice I had discovered that the mice were able to enter the first chamber of the delusion trap but were then able to exit back the way they came. This helped me find that there were several problems with the trap. The main issue was that the first door would stick and fail to close when the mouse entered. There was a problem with the print itself, the main body of the print had a defect in the side wall which caused a ridge, the seesaw mechanism would get caught on the ridge and fail to operate. Rather than re-printing the large main body of the trap I sanded what I could on the side wall, and I modified the seesaw to be narrower to avoid getting caught on the side wall. This helped but there were still issues. Even on occasions when the seesaw worked to close the door the mice were able to grab the door from the inside and pull it down and exit, so the door needed to be un-grabable from the inside. The interface between the front door, front wall and seesaw is an extremely tight tolerance area. I had recently upgraded my printer nozzle from 0.4 to 0.6 for faster printing of larger objects, such as a mousetrap, but this made the prints a bit more coarse. I noticed there were some stray filament strands that needed to be clipped for better functioning of the door but the interface between the door and seesaw was still jamming even after trimming. I struggled to figure out why because both the door and seesaw worked so well by themselves, but not together. Then I wondered if it was where they were touching that caused the problem. I sanded and lubricated the interface between the seesaw and the front door but it was no use. The day was drawing to a close and the test mice were still able to enter and exit the 3d printed delusion.

I set about my task with vigor. I got up early (nothing really unusual about that) and redesigned the front door and seesaw mechanism using blender (computer 3D modeling program). The front door would now fit flush in the entry hole so the mice could not pull it down from inside and exit. I also omitted the full hinge bar at the bottom of the front door so that there would be less opportunity to contact the trap body and stick or jam. Finally I focused on the contact surfaces of the door and seesaw and devised a cam that would translate small seesaw movements into large door movements to close it and keep it closed with minimal mouse contact. Having another test piece ready, I sent it to the printer. As my wife left for work I committed to a goal. I told her "I want to have the mouse trap working by the end of the day." All day I diligently persisted with designing, printing, and testing the trap. Still, I was frustrated by intermittent door sticking issues. Another problem I wanted to tackle was increasing the sensitivity of the seesaw and making it more "uncomfortable" to sit on. The mice were spending too much time in the first chamber (when the door worked) presumably having a nice little nap in the dark. This resistance to push their way into the holding chamber area delayed making way for the next mouse. Toward mid day I decided to counter intuitively print the large flat seesaw component vertically instead of flat on the print bed. This orientation would allow the layer lines to be perpendicular to the door's layers and so they might not catch on each other. Indeed, this seemed to work and after a few more prints and revisions (vertical orientation required supports near the hinge and removing them broke the hinge on two prints rendering them useless) the door no longer would stick. In fact the uncomfortable "spikes" to deter loitering, (no more than a few rows of triangles printed on the seesaw mechanism) served to add significant weight to the back of the seesaw making the trap extremely sensitive - a hair trigger. The mechanism no longer caught or jammed and now operated very smoothly. I was now confident enough in the mechanism to give it a real world test. I installed the trap in an area of high mouse activity and in about two hours had finally caught a mouse.