Replace magnetic connections with small, self-aligning, cyclable electromechanical connections

[jonnew]

The magnetic connections are a a clever idea due to their self aligning nature basically infinite use cycles. However, they introduce a major issue due to their proximity to the magnetometer. Alone they introduce a strong local field that must be modeled and subtracted in order to see the weak geomagnetic field. This is not ideal, but also not insurmountable on its own. The real issue is when the device is used in a multi-animal context. The relative motion means that the other's animals' magnetics are not static and thus cannot be distinguished from true motion relative to the geomagnetic field during sensor fusion. This results in spurious orientation readings.

This screen grab shows the fused Quaternion from of a stationary device as I move another optogrid around it:

[danielfrankch]

I was considering this trade-off as well. Maybe we could consider having two variants of the housing unit? One that uses a mechanical connection for multi-animal settings and where the magnetometer reading is important for the experiments, and one that keeps the current design of magnetic attachment to be used in single animal settings, or when magnetometer reading is not important?

The current suite is built around the magnetic attachment for the following benefits:

1. The magnetic force serves as a constant self-correcting force that maintains the integrity of the housing unit pair, so that the electronic inside is "pushed" onto the skull to have consistent optical performance across sessions. If we use a mechanical connection maintained by friction, it would not have a corrective force to correct dislocation due to vigorous head movement. Additionally, if the mechanical connection fatigues over time, it may introduce inconsistency when the device is used beyond rated cycle counts (which I would imagine happens quite often).
2. The magnetic attachment is also for usability. Users just need to bring the cap in proximity, and the magnetic forces pulls it in place. A mechanical connection would require more care from users to get it right. From my experience, this is especially important as the implant is on a mouse's head, who often time never sits still for experimenters to do careful alignment of cap swap. That was why I switched from mechanical connection to magnetic connection in the first place.

I wonder what is your thoughts on the spring-loaded pins (pogo pins) as electrical connector? I looked into alternative connectors like this pair below. But even the smallest profile would require a minimum board to board distance of 7.9mm, because these kind of connectors require a male and female counterpart. The pogo pins on the other hand, I was able to just use the male part, and use a properly-sized gold-padded through-hole on the PCB for the receptacle to achieve a board to board distance of 4.5mm, which makes the device more compact and ergonomic for the mouse to wear on its head.
https://www.mill-max.com/products/interconnect/machined-pin-header/399/00-310000
https://www.mill-max.com/products/interconnect/machined-pin-socket/399/21-300000

So if we were to make a variant that uses a mechanical locking mechanism to secure the house unit pair, I would keep the pogo pins for the above reason.

Regarding the ideas on a mechanical locking mechanism, Jeff brought up that we could use a single-turn screw like a Luer-lock for that. For example the cap can have a structure that latches to the base plate on the back side, and have a single mini "Luer-lock" on the front side to lock it in place. However, I am still looking into if there are suitable parts, especially at this small scale, to implement this.