This is a working fruit machine which has been made almost entirely from K'Nex - the only non-K'Nex components are the labels and two small rubber bands. None of the K'Nex pieces has been modified in any way.
Somewhat smaller than my first version, it uses K'Nex balls rather than large plastic play-pit ones and incorporates various design improvements.
This Instructable does not give instructions for building the machine; rather, it is intended as a vehicle to demonstrate the various K'Nex techniques which have been used and the kind of problems which can be encountered when creating something new.
The machine is simply too complex to be covered by step-by-step instructions, but guidance can be given for potential constructors!
Method of Operation
A game is started by inserting a ball in the slot at the top right of the machine. There is no check that a ball has been inserted, since the player can at any time interfere with the reels to guarantee a win if they really want to cheat (but this might be a future enhancement). If the ball reserve is full, the inserted ball wends its way to a 'cash' box underneath the machine.
The handle is then gently pulled. This winds up the game timer, raises the pay-out cradle (which latches into position), causes any paid-out balls to be replaced from the ball reserve, and then spins the three reels. The handle is then released by the player, and this starts the game timer.
After about 4 seconds, the first reel brake is released, and this is followed at about one-second intervals by the second and third reel brakes.K'Nex Techniques Used
A second or so later, the pay-out cradle is lowered. Each digit on the reels has a slot associated with it, the higher the value of the digit the deeper the slot. The cradle falls to the depth of the shallowest slot. There are three levers at the back of the machine which are lifted by chains, the first releasing two balls, the second releasing two more, and the third four more. A win of two balls causes the first lever to be lifted; a win of four balls causes the second lever to be lifted too, and a win of eight balls lifts all three levers. There is a fourth lever between the eighth and ninth balls which is only ever lifted at the start of each game - this is how paid-out balls are replaced.
Frame and Pay-out Cradle
The frame of the machine needed to be strong - the machine is very heavy and must not collapse when being moved around. Also, the pay-out cradle must not distort, because if its drop is limited only by the left reel, it must fall the same distance as when limited by just the right reel. The way round this was to form cubes where each of the six faces had a diagonal rod and opposite faces had these diagonal rods going in different directions, resulting in a very strong structure.
A timing device was required for the game cycle: the reels have to be spun, the first reel must then stop, then the second reel, and then the third. In my first fruit machine I constructed my own ratchet. It worked, but it was weak and clumsy. Since then I discovered that there was a K'Nex ratchet, and I acquired some. It is a brilliant design: it's easy to see how it works but one wonders how someone thought up something so simple and so clever. Do have a look at one.
A cam was constructed from white connectors and white rods. Each connector was connected to the adjacent one with white rods opposite each other, and these were staggered along the length of the assembly. Have a look at the photos. The result was a rigid cylinder, and this was used not only to wind up the timer but to control the reel brakes as well.
If K'Nex balls are allowed to crash into each other, especially from a height, the two halves will tend to come apart. When a ball overflows the ball reserve, it falls down a chute which is studded with rods so that the ball is constantly diverted as it falls, thus limiting its speed of descent.
Connector and Rod Attachment
If a rod is attached to a connector and is pulled from it with some force, it will snap out. By inserting a green rod either side of the connection, the join will be much stronger, because the flexing of the connector either side of the join is very much reduced. This technique was required where the handle pulled down the reel-spinning lever.
A lever that is one connector thick is very flexible. One of the levers in this machine (the one which is used to spin the reels) needed to be strong and rigid, and was made up to six connectors thick (with blue rods used transversely to clamp them together). The result was a heavy, chunky lever which would never fall to pieces. This technique was not appropriate for the handle owing to the resulting weight.
The pay-out cradle needed to drop slowly, and a series of gears and a flywheel, combined with a K'Nex ratchet, was used to damp the fall of the cradle. The problem was that if the cradle fell too fast, it slipped past the limiting rods on the reels.
Pay-out Cradle Release
The pay-out cradle is heavy owing to its rigid structure. It is released at the very end of the game cycle by the timer, at which time the remaining power is minimal (it is powered by two K'Nex elastic bands).
The problem was how to use a tiny force to release a heavy weight. The technique used was to use a lever which, under its own weight, would fall so that its momentum unlatched the cradle, being reset when the handle was pulled. All that was required was to let this lever fall, and this was achieved by sliding (horizontally) the rod on which it was resting. The friction is minimal and there is enough residual power in the timer to slide the rod.
The key point here is that nearly all the energy for the cradle release is supplied by the player when the handle is pulled, and not by the timer power.
Further details of the techniques will be supplied on request.Randomness
The randomness of the spin is created in two principle ways. Firstly, the game-timer does not start until the handle has been released, at which time the reels have already started spinning. Any delay in releasing the handle will result in a slightly longer spin than if the handle were released sooner. Secondly, if the handle is pulled with more force the reels will spin a little faster, because the reel-spinners will be given more impetus. However, if the handle is pulled too hard, it will fall to pieces! If a child wants to use the machine, some tuition is required...Problems Encountered
There weren't too many problems encountered during the construction of the machine (which started on 6th July - just five weeks ago). These are the ones which spring to mind:
When building a reduced-size version of a machine using K'Nex, there is less scope for achieving required lengths and distances. For example, in a large machine where a rod of length, say, 328mm is required, lengths of 326½mm and 333mm can be achieved (see if you can work out how, bearing in mind that the real lengths of K'Nex rods are 17½, 33, 55, 86, 130 and 192mm, with the connectors requiring 10mm) - probably an acceptable deviation. With a smaller machine, a target length of 100mm might be required, but only 92½mm and 108mm are achievable, a deviation which most likely would not work.
This caused problems, but the only remaining one is that the handle needs to be pulled in two stages. A gentle pull is required as the reel spinners engage and the reels settle into position, followed by a finishing stronger pull. I really wanted the reel spinners to be 2 or 3mm longer, but I was unable to achieve this without compromising their strength. There is scope here for future improvement.
When the balls were replaced after an 8-ball win, the rush of balls occasionally caused a bulge in the leading eight. This was solved by inserting a few yellow rods so that a rising ball would be impeded and rise no further.
Another was that the handle occasionally fell to pieces! The handle has to lift the pay-out cradle, wind up the game timer and launch the reels, all of which requires a certain amount of force. The handle has a square cross-section made from blue and purple connectors connected by white rods, and by reinforcing these with white connectors all the way round (using green rods), the handle was much stronger than before (have a look at the photo). However, if it were yanked hard it would still fall to pieces! The problem is that a stronger handle would be much heavier and this would introduce further problems. The current handle is fit for purpose.
A problem occurred when the pay-out cradle was released. It fell down with too much force even though that was the minimum force required, and this resulted in the vertical slot-testing levers slipping off the limiting rods on the reels. This problem was solved by damping the fall of the cradle by using a gearbox with a flywheel, once again taking advantage of a K'Nex ratchet. You can see this in the photos.Outstanding Problems
As with any recently-developed machine, there are still a few of problems to be resolved, and there will doubtless be other areas for improvement.
The labels, which were fitted today, are a little bit too wide, causing the reel brakes to slightly rub at the sides. This will be a trivial fix, requiring just a pair of scissors.
The pay-out cradle is currently released half a second too early, meaning that if the third reel takes time to settle it can end up between two symbols. This too will be a straightforward fix.
Very occasionally a reel will stop between two symbols. It happens very rarely and no fix is currently apparent, but it is running as a background task in my brain...