Like many things on this site this story is quite crazy: it's about the building of a door latch.
Summer and winter: the door in question leads from the kitchen onto the terrace. In summer I want to keep the heat out and in winter I want to keep the heat in. But most of us forget to close the door, therefore it has been equipped with an automatic closing device:
Great, but then… Yes, at times you want to keep the door open, e.g. while clearing the table outside after dinner on the terrace. That means during that time the door has to be kept open somehow. A latch?
Because in another page I make a point against Creationists who think the world was designed. I have come to understand that design itself is a form of quick evolution of ideas inside the brain. The door latch is simple enough to serve as an example to demonstrate that devices are not appearing fully finished in the designer's head.
This page does not give you all the stages I went through, but I hope it's fun to read.
Believe it or not, the first version of the latch was a simple cardboard tube.
This was, shall we say, less than convenient:
A lot better was something like this:
The whole point of this page is: how do you come up with an idea like that?
The simple latch above resembles the finger of someone holding the door open, so it should not be too surprising where that idea came from.
But version 2 has obvious disadvantages:
We attack each disadvantage of version 2 in turn.
If we want to hold it in position when the door is closed, we can put a simple pin under the finger.
It can then also no longer fall against the wall. But we still need to lift it up to hook it over the door rim. It would be better if the door could push it up. Ratchet-type mechanisms we have seen in many places including in ballpoint pens, so we're borrowing from others here, not really designing from nothing.
All we need to do is to modify the profile a little. We also move the positions of the pivot and the pin to get better transmission angles, and we round things off a bit:
Now when you open the door the door frame will push the latch's finger up, it will fall into place and lock the door in the open position. It's not obvious what slope is best, it depends on the materials, the place ther is, the wear we afford on the door frame and so on.
But even if it now latches, I still need to reach up to unhook it. A piece of string through two eyelets helps:
That in itself required a few quick trial-and-error positions inside my mind. We are progressing, but note how often we are comparing ideas and picking what works, discarding other solutions. And the dimensions are now wrong: we need to adjust several lengths. It's also obvious that the little block attached to the wall is not quite right.
A very bad problem remains: what if our distracted visitor tries to close the door? The hook should "give". Usually springs are used for that. The arm needs to have two parts: one attached to the block on the wall, another one that hooks over the door frame but can move out of the way.
The red dots in the side view are pins: some are for pivoting, others for restraining movement and two are for attaching the spring.
This part took most time to do and resulted in quite a few drawings that ended up in the wastebasket plus a few bits of wood that were sacrificed (they ended up lighting fires in the fireplace). It would take several more pages to describe in detail how I got to this stage and what happens to the device when the door is forced shut. That's beyond the scope of this argument about design, but the point is made.
The main engineering problem with thflexible finger solution is to attach the spring at the right points. That is quite difficult to get right: the spring needs to be strong enough to keep the door open against the pull of the door closer, but not so strong that it damages the mechanism when the door gets closed forcefully.
The final prototype that actually worked well looked like this:
It's rough... but it does the job.
In evolutionary terms we now have a beast that is "adapted" to its environment: it will not be killed by the first absent-minded visitor (or more likely myself). It works even without having to pull the string.
But it makes an awful noise when you force the door closed: the spring slams the finger back against its stop.
To make the latch work nicely we would need a mechanism that offers no resistance in one direction (when the door pulls the finger out) but that prevents the finger from smashing into its retaining pin on the way back. There is one such system inside the door closer: it is a piston that moves in oil, with a valve. When you pull to open the door the valve opens and lets the oil pass from one side of the cylinder to the other easily, but when you let go of the door the valve closes and the oil has to pass through a little hole in the piston, which prevents the door from slamming shut.
In principle I should mount a miniature version of that damper on the finger so that it gives easily one way but does not slam the other way. I have no access to the necessary machine tools to make such a device. A rachet would do the job, and I have an idea for a double ratchet system whereby the damping is achieved by air friction.
If I ever build it, I'll add the development of it to this page.
