Continued work on MIWA/Anker 3800

I had spent quite some time on the Anker 3800 cylinder. This system was originally patented by the Japanese firm MIWA. It contains just 4 pins, each of them having 4 possible depths. But, there’s also 4 sliders that are magnetically operated by magnets in the key.

An Anker 3800 Eurocylinder
The keys

This system was sold in the Netherlands by Dutch firm Ankerslot and is still used in many large, high security setups.

I made a magnetometer to decode the magnets in a key or in a cylinder, made a 3D model of the key to be able to print it and figured out how to get the master key of a system. My talk at Disobey is available online.

At LockCon, Han provided me with a set of five cylinders, all keyed differently but all part of the same masterkey system. No keys were provided.

I started by reading out all the magnets in the sliders. Each slider can have one or two magnets. If there are two, one is for the user key, one for the master key (we’ll ignore submasters for the moment). By knowing the magnets in the cylinder, it is possible to figure out which magnet should be in the key. For master keyed systems, there might be two possible magnets to operate the slider, where one will be in the master key, the other in the user key. With this information, I could determine the magnets in the master key and in the individual user keys (for each slider, there was only one possible magnet that would open all cylinders, so it was clear that that particular magnet should be present in the master key).

The correct bitting is another story. I had no key to start with. It is possible to create 256 keys and try them all to find all possible bittings. I was lucky however to have gotten one half cylinder.

What I did is create a bump key (so four times the deepest cut, which I will call a ‘4’), containing the correct magnets. With the bump key, I could open a cylinder and disassemble it. The half cylinder was my luck, as that can also be re-assembled. For this task, I used a 3D-printed plug follower.

This gave me all the correct bitting positions for this one cylinder (pin 1: 3 or 4, pin 2: 1 or 3, pin 3: 2 or 3, pin 4: 4). Since I did not have the original key, I did not know which depth belongs to the user key and which to the master key. The configuration allows for 2^3=8 possibilities. By creating 8 keys and trying them on the other cylinders, I could find out the master key bitting. With 3D printed keys, it is possible to print, say, a 3/1/2/4 key and if it doesn’t work, file a bit away to get a 4/1/2/4 key. That is exactly what I did and 4/1/2/4 worked on all cylinders, giving me the correct master key.

Next, I wanted to create the user keys. I could have created user keys by only using different magnets and keeping the bitting the same, but that would not give me the keys as they would originally be made for these cylinders. Since I now had a working master key, I could easily probe each pin position for all depths in sequence. I started with a 1/1/2/4 key, trying that on the 4 remaining cylinders and writing down the results, then filing it to a 2/1/2/4, trying again and then 3/1/2/4. With four keys to start with and 4*3=12 filing actions I was able to decode all the bittings.

Decoded cylinders, ‘G’ and ‘R’ are north/south poles (my magnetometer uses green/red light as an indicator)

Here, I have put squares around the magnets and bittings of the master key. If there are other bittings or magnets, they must be in the user key. The first pin of cylinder 2 for instance has possibile depths 3 and 4, and the master key has depth 4, so the user key must have depth 3. Note that in this system (and in this instance), user keys can have a bitting that can be filed down to the master key, as long as the magnets are different. Cylinder 2 shares the middle two magnets with the master key, the outer two are different.

With that information, I knew all the individual keys and printed them.

The user keys only open one cylinder each

And I have the master key.

The master key is golden of course

In fact, with this collection of cylinders, it is now also possible to make submaster keys for certain subsets of cylinders, even if that was not intended originally.

Thanks for reading! -Walter.

Photos CCBY4.0 Walter @ Toool Blackbag

One Response to “Continued work on MIWA/Anker 3800”

  1. Han Fey says:

    I am impressed by this technical high quality deep going presentation / investigation from Walter.

    Han Fey
    Senior High Security Lock Analyst AA