There are a fair few attempts at “unpickable” locks out there, and overall I don’t think I have a great deal to add to the range of pre-existing methods. But a couple of years ago I thought I might have a go at my own design anyway.

I’ll point out up front that there’s not a lot of practical use in making locks harder to pick than they already are. It’s only people who need to hide the fact that they were there who are going to faff about with picks when they could just break something to get in. That limits the market somewhat.

A pin tumbler lock is a plug with a slot cut along it for the key to enter and displace pins which extend into it from outside. These pins normally prevent the plug from turning unless they’re all moved to a position where breaks in the pins align with the shear line where the plug fits in the body of the lock.

The basic idea of lock-picking is to apply light torque to the plug and then poke around with those pins until they get stuck because you’ve managed to wiggle one up to that shear line and the plug has moved just enough to catch it.

A common mitigation is pins with weird shapes that will catch in all the wrong places, but which don’t cause any trouble when they’re lifted directly to the right place with the right key.

Taking this a step further, one can design a lock which captures some kind of imprint of the profile of the key. In effect catching anywhere you set it, and then allowing the cylinder to turn until it’s impossible to change the pin any further, and only then beginning to bind or carry on depending on if the pins are all set correctly.

And that’s where I went.

My design has two plugs. The first, where the key goes, is a skeleton-key kind of arrangement. Each pin is split at every step so that it will always turn. The second is an inversion of the usual binding test, but it shares the pin stack with the first plug.

The trick is to allow the first plug to turn until it has captured a specific set of pin segments (“master pins”), so they can’t be adjusted any longer, before beginning to turn the second plug.

There are ergonomics problems, here, in how you choose the coupling. You need both cylinders to return to the home position at the same time, to free the pins so the key can come out. And you can’t just have the top plug turn slower because then you end up turning the bottom plug further than usual, and that would be hard to use.

So I decided to couple them with non-circular (or at least non-concentric) gears. At the home position, where the cylinders are aligned, rapid motion on the first plug maps to slow motion on the second plug, and then this begins to accelerate to make up the difference once the first plug is sufficiently off-axis that it can’t be picked any longer.

Here’s a set of gears which achieves that (and also avoids the screw holes for mounting it on a door): rear view

Where the bottom plug’s gear is close to the centre (marked by the hole) it requires the most travel to effect rotation of the outer gear, and this corresponds to where the top plug has the opposite arrangement, doubling the effect. By the time the bottom plug has turned 180 degrees the relationship is reversed and both plugs will synchronise again. This corresponds to the point where one would normally not need to turn the key any further.

I’m not sure how effective this would be against bumping. It changes the dynamics somewhat, but not necessarily in a way that would prevent it. I tried bumping a prototype and it caused the thin master pins to turn around and jam the lock, and I had to give things a jolly good wiggle to shake it all back into place, so maybe that’s sufficient.

In my design the pins and springs are standard parts, and the plugs are nearly standard parts but modified. I filled the top one with resin and re-drilled the holes, and then stuck some gears around the back. But gears with a radius larger than the plug fails to leave any clearance for the screws, so that’ll have to be re-designed a bit. I had been waiting for the release of Gearify 2.0 in order to be able to design those, but since it was released I haven’t had the free time to actually get started with it.

So far I haven’t been able to fabricate anything durable enough to send to an expert for their feedback. I’m sure the concept can work but my first implementation will likely have flaws which result in embarrassing outcomes.

Here’s a prototype in action:

Also, I’m putting left-handed digital calipers on my Christmas list. Just so everyone knows.