The first part of this lock design masterclass covered the basic principles of pin cylinder lock operation and how some forcible attacks can be mitigated.
In part two we will look at the issues of non-forcible attacks, learn how tolerance conflicts are a real problem, that springs can cause sleep pattern disturbance and that you’re damned if you do and damned if you don’t…
There is a little more technical terminology in part 2 than in part 1, but it’s been written with a view to being understandable to the layman with an interest in mechanical design and locks, so if you’re not familiar with some of the things mentioned herein, just remember that springs don’t like being over stressed and that tolerances can bite you on the ass…
So, how else do people try and get past a lock?
Protection against surreptitious attack
The primary form of surreptitious attack is that of ‘picking’ the lock. There are many other methods, but they are beyond both the scope and length of this article.
To pick a lock is to do the following: Firstly, a torsional moment has to be applied to the plug so that pressure is applied to one or more of the drivers (remember that when no key is present in the cylinder, the drivers are the blocking elements that interact with both the plug and the cylinder body).
To do this, a ‘torsion wrench’ is made use of. This is normally a flat strip of bent wire, as seen in Fig.1.
When this pressure is applied, the effect is to ‘bind’ at least one of the drivers between the plug and the body. Then, a separate implement (which can have many names, depending on its specific profile) is used to manipulate the pin/driver/spring upwards until the shear line between the pin and the driver is reached. When this happens, the person picking the lock will feel a very slight advancement of the plug, and will then attempt to find the next pin that is harder to lift than its fellows.
To show this in an exaggerated form, Fig.2 shows a very badly machined plug and cylinder body, where you can clearly see that all the pin chambers have deviated from the Z (vertical) axis to a greater or lesser extent, and that the diameter of these holes is far greater than they should be.
When these are combined together as an assembly, as in Fig.3, you can also see that the deviation is not uniform relative to both plug and body (plug is coloured blue for clarity). This means that if torsion is applied, one driver will certainly ‘bind’ before the others do, and this is a progressive fault that will allow the lock picker to incrementally manipulate pins and open the lock.
To illustrate this further, Fig.4 (showing the pins all in red) attempts to demonstrate an exaggerated view of how the pins would consecutively bind on this particular model, with the arrow pointing to the pin that would bind first if tension were applied to the plug in a clockwise direction.
In order to mitigate these unavoidable tolerance deviations, many different ‘anti-pick’ mechanisms are made use of, but the most common and economically effective form is that of an anti-pick driver. These come in various forms, but all of them are designed to take advantage of imperfections in the machining of the plug and housing components – Fig.5 below shows three typical examples.
Of the three samples in the image, you can see that the one commonality is a reduced diameter mid-section. This allows the lock picker to apply torsion to the plug and achieve several degrees of unhindered movement, but upon applying an upwards force to any of the pins, they will now bind against the cylinder bore. This requires torsion to be reduced so that the pin that is being manipulated may ride over this obstruction.
The desire of the designer is that as this torsion is reduced, one or more of the pins that have already been manipulated to their correct position will now be returned to their ‘home’ position by the spring force acting upon them.
You will see that the three drivers in Fig.5 vary quite significantly. The one on the left is usually known as a ‘castle’ driver and is the most common, owing to the ease of manufacture (remember that the volumes of such a component run into the millions).
The middle driver is usually known as a ‘mushroom’ driver and is very effective if inserted the correct way around. When done so, it has the tendency to incrementally increase the rotation of the plug until it binds very effectively. However, it is susceptible to human error when inserted and this can be counterproductive, as if inserted incorrectly the driver will progressively help the lock picker. It’s also more difficult to manufacture en masse.
The driver on the right is of the serrated type and is rarely used, owing to both machining considerations and a tendency to wear.
So, our sample lock will make use of castle drivers, as in Fig.6 below, but only in limited chambers, otherwise the key holder may apply an unintended torsion when attempting to insert the key, and this would effectively jam the mechanism.
Protection against covert attack
Although these can take many forms, it would take too long to describe them all in detail, so the one that we’ll focus on will be ‘combing’ or ‘overlifting’. This is a method well known in the industry and takes the form of attempting to manipulate all of the pins and drivers out of the plug and into the cylinder body. Fig.7 shows our sample cylinder in a condition where the last pin chamber has had its elements moved to this condition.
If this can be achieved simultaneously with all five chambers, then you can see that this would bypass the lock.
There are various devices made both by locksmiths and criminals to achieve this, and most of them resemble a hair comb that has the prongs cut to match both the pitch of the chambers and height of the plug. As the keyway has a corrugation, sometimes these are made of flexible plastic. In Fig.7, if you look closely you’ll see that the driver in the affected chamber is of a shorter length than the rest, and this is what allowed this condition to occur. As the spring has reached its solid length, if the driver had been longer then part of the pin would have remained in the plug.
In Fig.8, with colour coded pins and drivers, you can see that the first (orange) driver is of a shorter length than the other four silver drivers, and this is because the green pin that it rests on is significantly longer than the other pins. In this image, the last driver has been restored to its correct length.
In general, only two lengths of drivers are required to prevent this form of attack, but that is still undesirable from a manufacturing and assembly perspective, since it is both prone to human error and introduces an additional component into the bill of materials.
Springs
For readers with experience in spring design, you may have already realised that some potential issues are galloping towards us over the horizon. This is a thorny issue for this article, as people who have no interest in springs will find it a bit wearisome, whereas people who do have an interest will find it somewhat lacking in detail.
I think it’s fair to state that what follows will disappoint in equal measure depending on your preference, but I’ve done my best to achieve an equilibrium of disappointment.
In a mass produced product such as this, from a manufacturing standpoint it’s practicable to have only a single spring design of a given length, constant diameter, etc. To do otherwise would be a recipe for disaster. Unfortunately, there are a few elements of the design that are not conducive to any uniform pre-load, stroke or fully loaded length of any given spring per chamber.
To start with, the spring in the first chamber (closest to the keyway entrance) is going to be exercised five times more than the last spring over the lifetime of the lock, as it rides over the different pin cuts in the key as it is inserted and withdrawn. So that would be the worst case starting point.
In order to design a compression spring (or indeed any other type of spring) for an infinite life, many factors need to be considered – material properties, pre-load force, spring constant, index, stroke etc. It’s a long list if done properly.
The main problem is that in this mechanism there is a combination of varying magnitudes of compression that the spring will be subjected to during its life cycle. This is because we have ten different pin lengths and two different driver lengths, so the best we can hope to achieve is a happy medium.
Given that the mass of the pin and driver combination will not exceed 0.6 Grams, a fully loaded force of 0.1 N (approx. 100 grams) per pin should be sufficient to provide an ergonomically acceptable feel when the key is inserted, and also provide sufficient force to overcome any ‘grime’ that may accumulate when the lock has been exposed to the elements for a given period of time without being utilised.
As the length and diameter of the spring are constrained by the chamber dimensions, the three variables that are available to manipulate are the choice of material (this has only limited impact on the available force, particularly for a small compression spring of a thin wire diameter), the wire diameter itself, and the amount of turns used to construct the spring.
Given that the lock will probably be exposed to the elements, the only good choice of economically available material would be phosphor bronze wire (EN 1642) because of its excellent corrosion resistance (although with a relatively low U.T.S.).
So that leaves the wire diameter and the amount of active coils. Obviously a large diameter wire and less coils will equate to a stronger spring, but there’s also the spring index (tooling wear and tolerance issues) and available wire diameters to consider. So for this spring the optimum wire diameter would be 0.2mm, given that the required spring constant is 0.06N/mm, with a stroke of 3mm (in theory).
The maximum pre-loaded length per chamber (for the spring) given the uniform depth of the closing pin and the maximum displacement of any given driver is 9mm. The full extent of the spring compression will be a theoretical 3mm. This requires the spring to compress by 1/3rd of its length and that provides for an acceptable factor of safety of about 1.7 for this particular fatigue loaded spring.
However, as Fig.9 shows, even a spring that supports a zero travel pin and driver arrangement has to be compressed by the highest point of the key, as the key is inserted into the keyway.
This increases the stroke to almost the solid state of the spring, and as readers with any experience of spring design will fully realise, that throws all the calculations out of the window.
To correct this flaw would mean a complete re-design of the spring, and probably also a re-design of the driver and closing pin heights relative to the overall space within the chamber, but consideration still has to be given to preventing over lifting of the pins, as mentioned earlier.
As a general rule, compression springs should never operate over a range that exceeds 80% of their available total compression, but this can be exceeded in the case of barrel or double helical springs, so a refinement on some pin springs is to have more than one closed coil at either end (to maintain an acceptable solid length), but to have a tapered profile, allowing for a greater range of compression (and a higher constant) over the total length.
Tolerances in lock design
Tolerances are little more than a set of contradicting requirements for lock design.
The golden rule of designing to the widest possible tolerance range whilst still achieving a guaranteed interacting set of components does not apply.
High tolerances are required to prevent manipulative attacks, and also to guarantee as many key differs as possible. But low tolerances are also required to allow for environmental considerations such as materiel ingress, thermal expansion and contraction, and even small insects deciding that your lock is a suitable home.
Fig.10 shows one of the inescapable tolerance gaps that come into play in this type of cylinder – a planar surface nested within a cylindrical body. Not good for manipulation attacks.
In the following Fig.11, X and Y axis displacements are shown, and the calculation of these elements becomes quite complex:
A+B is the potential difference in pin / driver dimensions, B+C is the axial differential between chamber hole dimension and pin dimension, D+E is the theoretical axial centerline. E+F is the possible variance in disparate pin component
diameters, and F+G is the tolerance allowance between pin and chamber.
The sum of these component parts is quite complex, and involves not only subtraction from disparate entities, but also integration with discrete entities. For those of you who are familiar with GD&T, my apologies… I’m only trying to show what complications can arise.
And they increase when we consider the axial tolerances…
The Fig.12 shows the potential tolerance stack up between the Z axis elements of the model. I’ll leave it to the readers of the article to figure this one out.
Our U.S.P.? (Unique Selling Point)
Nothing so far has made our lock appreciably any better or worse than anything else on the market. So what can we use to differentiate ourselves?
Any USP should differentiate itself in a way that benefits everyone in the selling chain. In other words, we all score. It sounds easy but it isn’t.
Lock-in is one way to go – sell it, and then place obstacles in the path of the customer that prevent them from jumping ship.
Fig.13, shows the same key but with two small dimples machined into the side of one wall of the blade of the key.
These are hard to reproduce features of the key that interact with ‘blocking’ elements of the lock.
They do not require any springs, and nor do they require any additional shear line calculations; they simply need a space in the key into which they can interact with when the plug is rotated.
From a manufacturing perspective, they add an additional expense as a different set of holes need to be pre-drilled in the plug, however these are not high tolerance and can be made simply through the axial rotation of the plug in the machine.
For the cylinder body, the prospective expense is further reduced, as all that is required is a simple removal of material from the entire length of the internal bore of the body in the form of a semi-circular cavity.
They certainly don’t contribute to the security of the lock in the same way that a locking element does, but they serve a useful purpose.
In Fig.14 below, you can see that features have been added to the model that allow for these blocking elements to interact with the key in a way that enables a specific type of key profile to develop a unique feature that benefits both the seller and the buyer.
In other words, the manufacturer (who is theoretically the only one capable of machining such cuts) can lock in his customer (the locksmith or security retailer, who can have their own specific key ‘profile’), who in turn can lock in his end user, who in turn appreciates the additional security of having their own unique key profile that requires permission prior to unauthorized key duplication.
So, do we have a good lock design? No. Quite a long way from no, actually.
I’d be hard pressed to give it a 2 out of 10.
Modern high security lock mechanisms can easily achieve billions of key combinations per cylinder and attain tolerances that in some cases are measured in microns. But in those instances, you certainly get what you pay for, and our sample lock was designed on a pauper’s budget.
In any event, I hope it’s given you some useful insight into the conflicting design challenges that present themselves in this area of mechanical design, and if you’d like some more insight, feel free to contact me at the below email address.
K.I.S.S.
Text and illustrations © [email protected] 2016
