Optical gate for mechanical lock

[eric434]

Here's an idea: Replace the fence in a lock (like a safe lock) with an optical gate. (LED+phototransistor) Super simple, and now it's impossible to get feedback, which means it's impossible to manipulate.

Thoughts?

[Raymond]

Brilliant idea. Now you just have to get the electromechanical part that translates and connects the optical sensor to the physical fence and then pulls the bolt in invented. Maybe you are on the start of something grand.

[eric434]

That's easy, for the safe lock case -- have e.g a little motor-controlled nub project in from the back plate of the lock case to hold the nose off the gate area. If the combination is right, retract the nub and the nose can now drop in, allowing the lock to retract the bolt normally.

(might also work with a latch mechanism on the top of the lock case to hold the nose high, etc.)

[Wizer]

As Raymond said: Brilliant!
Alltho it wouldn´t be as easy to fit in, you could use the same consept for rotating disc tumbler locks too. Like Abloy.

[eric434]

Yes, though I'd perhaps use a laser diode/beamsplitter/photodiode bouncing down the length of the lock and back off a retroreflective patch -- you'd have to "see" through a much smaller hole.

Two additional thoughts on the original idea, though.

- Obviously you have to defend against X-raying the lock. The traditional defenses involve using plastic wheels, lead shields, or (something I came up with off the top of my head) a cleverly placed film that will get exposed if someone x-rays the safe.

However, plastic wheels or "detector" film aren't an ideal defense against radiographic attack. It's theoretically possible for someone to develop a sensitive enough X-ray detector to image the difference in density between the solid part of a plastic wheel and the gate. And it's also possible for someone to develop an imager that uses radiation at low enough levels or at wavelengths that won't expose the film. It's even possible someone might someday develop a holographic ultrasound imager or something that uses a wholly different kind of radiation.

Therefore, to provide real security we'd need to do something that's not possible with a mechanical lock -- fill the gate area of the wheels with a transparent material that matches the properties of the opaque material around it as closely as possible.

Now reading the wheels would mean either using something like an MRI (whose action I will assume would be blocked by the massive safe, even if you could make one that worked on non-organic materials) to measure the wheel composition, or optically like the lock does.

So, is there any way to read the wheel pack optically? Unfortunately, there theoretically could be.

- The Falle pin-tumbler lock decoder teaches us that it's sometimes possible to decode locks by sending probes through spaces the designer thought would be impassable. In the Falle case, that means a fine wire shim to read the length of the bottom pin.

For this lock therefore, we have to consider the possibility that someone might develop an optical probe capable of being inserted from outside the safe into the lock mechanism. Maybe something as simple as a nanometer-thin optical fibre attached to memory wire (sort of a version of the Falle memory wire safe lock decoder), maybe something more sophisticated with nanotechnology.

Eliminating the possibility of a probe therefore means sealing the lock off inside the safe, with no spaces connecting the inside of the lock to the outside world. Assuming the safe is built to high-security standards and therefore has a laser-cut / precision-machined door that leaves no gap between door and body (I believe at least one standard specifies less than .001", correct me if I'm wrong), the only possible connection is the spindle hole.

The spindle hole as a vulnerability we can eliminate by taking a page from another high security lock's book, the Evva MCS, and devising a magnetic coupling that puts an impenetrable brass wall between the outside world and the secrets of the locking mechanism. In this case, it would be a pretty simple clutch-type thing, with two rotors containing oppositely-polarized magnets that influence each other despite a nonmagnetic (brass or...?) barrier between them.

[mh]

Hmm, don't you think there are cheaper, more reliable and more secure solutions than keeping the traditional mechanical interfaces?
Like - for the safe lock - measuring the angle electronically and hiding the secret in an electronic part?

Cheers
mh

[eric434]

Cheaper and more reliable? Certainly. But electronics are funny things -- they emit signals you don't want them to emit, they can get backdoored, and they can be hacked. Even the history of encryption in the last century is one of cryptosystems getting broken by clever mathematicians and increasingly clever computers. Neither, therefore, are things which I would consider components of a theoretically perfect lock.

Compared to electronics and crypto, mechanisms are much better understood. So it's easier to spot the potential holes and engineer them out.

It's also much easier to machine together a simple optomechanical device in your basement than it is to make a high-security electronic lock!

[eric434]

It's important also to note that this is something you can fit to an existing lock, which is much easier than making a new one from scratch.

[Daggers]

to defeat the optical sensor, all you need is a key with teeth that move up and down kinda like pins in a lock when you brush a pick underneath them. That would cycle through many many key combinations for the sensor to pick up and eventually you would get close enough to the actual key to open the lock. To defeat that attack, there could be a delay when you stick the key in so you can't cycle through the key combinations as fast. But that would not be as user friendly and frustrate people in a hurry.

[FarmerFreak]

Daggers, he is talking about safe dial locks. So keys and raking don't really apply.

eric, if you really want to make this lock a truly secure lock you may want to avoid making it out of existing safe locks. I'll PM you a link to a rather ingenious tool, the tool isn't really a secret, but may be against the forum rules to discuss here. Anyways, if you are worried about sophisticated tools, your competition is going to be the Kaba Mas X-09 and the S&G 2740. And those locks are purely electronic, possibly because of the potential x-ray type attacks.

FYI, the S&G 2740 doesn't have anything electronic on the outside of the safe. It just looks and feels similar to a regular dial, except without the feeling of different wheels picking up and a drop zone.

[Raymond]

As you make the gate detector much more sensitive and accurate, remember that this also requires the person dialing it to be more careful and exact. The S&G 2740 requires the correct number to be dialed within one-half number on either side. Careless failures will make the user hate the lock and blame the lock for their own inaccuracy while dialing. I think we all know someone who cannot dial accurately and often requires several attempts to get a safe open.

The Kaba X09s have their own issues but when the user dials one he only has to stop exactly ON the required number in the electronic display.

I can make an old Sentry almost impossible to manipulate but it can still be dialed open relatively quickly by trial and error or sequence progression due to sloppy construction and wide gates.

I think you are trying to mix old technology with new technology but holding on to all the faults of the old technology. Entering numbers onto a sight restricted keypad is about as secure and fast as can be found.

Scramble keypads prevent someone from watching and recording the finger location and movements while pushing a normal keypad. Scrambles also prevent detecting the fingerprint residue to determine the combo. I have seen and used scramble number keypads with the red lit numbers. I hate them because of the time it takes to locate the next number. They are also difficult to see in many light conditions.

What I would like to see would be a liquid crystal display that scrambles. It could even display 20 numbers at a time of which you would have your choice of which correct number to push. That kind of keypad could also be programmed with letters just to increase the choices.

Just some rambling thoughts.

[eric434]

Interesting points, guys. Thanks!

Regarding the original idea, one strength I should point out is that it's truly invisible. If you leave out the magnetic coupling/brass wall (which would be proof against the tool you sent me, FarmerFreak), the safe would appear to be an ordinary safe with a Group 2 or Group 1R lock on it even to a safe technician, assuming you use my idea of a short-travel nub on the back cover to keep the nose from dropping in.

Say the lock is protecting some critical engineering plans for your new iPad accessory. The competition hires a crooked safe tech and an out-of-work former Egyptian spy to get him into your office and photograph the plans. The safe tech (looking at photos of the safe the Egyptian shot the night before) figures it'll be a two hour manipulation, tops. Only he shows up at your office and ends up sitting there until the janitors come in, wondering why his contact point graphs are all flat as the Sahara -- it's as if your lock happens to be made to within obscenely perfect tolerances, because he can't see a single gate anywhere.

The only possible difference might be a slight change in the width of the contact points, if the nub doesn't hold the nose at the precise height where it would be if the fence was riding on the wheels. (And even that can be eliminated, by e.g filing down the nose or building it up a bit with some solder)

[Daggers]

But let's say that someone can't open the safe and gets angry. They could make it inoperable by frying it with microwaves (not the kind in your kitchen, the ones that the one in your kitchen makes lol).

[eric434]

Then you get there the next day, find it won't open, drill it, and figure out you've got a major problem with attackers getting through your other security measures. So you plug the holes and now they can't get in anywhere.

Or, because it's a really really simple design with no microcontrollers, they set the microwave way past "defrost," point it at the safe, and the lock still works because the electronics are so simple. (simple enough you might be able to protect them against this sort of thing with cleverly placed inductors/capacitors/diodes and the like)

Or or, because they can't tell it's an electronic lock -- it's undetectable from the outside, remember -- they get frustrated and vent the only way they know how against a mechanical device, by kicking it. They kick it so hard they break their foot which gets infected and kills them like happened to Jack Daniel. (http://en.wikipedia.org/wiki/Jack_Daniel)

[Evan]

Interesting points, guys. Thanks!

Regarding the original idea, one strength I should point out is that it's truly invisible. If you leave out the magnetic coupling/brass wall (which would be proof against the tool you sent me, FarmerFreak), the safe would appear to be an ordinary safe with a Group 2 or Group 1R lock on it even to a safe technician, assuming you use my idea of a short-travel nub on the back cover to keep the nose from dropping in.

Say the lock is protecting some critical engineering plans for your new iPad accessory. The competition hires a crooked safe tech and an out-of-work former Egyptian spy to get him into your office and photograph the plans. The safe tech (looking at photos of the safe the Egyptian shot the night before) figures it'll be a two hour manipulation, tops. Only he shows up at your office and ends up sitting there until the janitors come in, wondering why his contact point graphs are all flat as the Sahara -- it's as if your lock happens to be made to within obscenely perfect tolerances, because he can't see a single gate anywhere.

The only possible difference might be a slight change in the width of the contact points, if the nub doesn't hold the nose at the precise height where it would be if the fence was riding on the wheels. (And even that can be eliminated, by e.g filing down the nose or building it up a bit with some solder)

So what you are saying is that you will build your optical-mechanical device, have its power supply concealed inside the container and hope that the user/owner will replace the batteries according to the required interval ?

Or were you going to rig up some sort of inductive generator mechanism and require the user to dial x-number of turns and stop before beginning to dial the combination sequence to charge the lock up...

The dead giveaway would be the emergency power application terminals on the exterior of the container that your lock is more than meets the eye -- unless you want to have to have the container drilled and manipulated open by a safe expert to replace the dead AA or 9V batteries...

~~ Evan