K'Nex Bouncing-Ball Amusement Machine

Background

This machine was made during the COVID-19 lockdown (in England) from March to June 2020.

As a collector of old-penny slot machines for many years, I decided to make a machine from the construction toy K'Nex which was very loosely based on the 1963 Bryans Bullion roulette-type slot machine. This machine had five slots labelled 2, 3, 4, 8 and 12 (or 10 on some machines after decimalisation in 1971). When a penny was inserted in at least one of these slots a handle could be turned. This made a pointer spin round which could land in one of 35 positions labelled 0, 2, 3, 4, 8 or 12 (or 10). If the player had bet on the winning number, that number of pennies was paid out.

With this machine, coins have been replaced by K'Nex balls, the spinning pointer has been replaced by a bouncing rubber ball which can land in one of 14 columns (which are labelled 0, 2, 4, 6 and ?), and there are four slots labelled 2, 4, 6 and ?. That last one (?) represents a random number from 0 to 19. As with the Bullion machine, if the player has bet on the winning number, that number of balls is paid out. Apart from a K'Nex 12v motor which is used in a chain-lift for the rubber ball, the machine is completely mechanical.

The size of the machine was dictated by the number of columns and the betting/winning mechanism. Too few columns would mean an uninteresting game, but too many would lead to a huge machine in which the duration of each game would be unacceptably long. With 14 columns, each game lasts about 25 seconds.

The machine is 147cm wide, 216cm high and 56cm deep (58" x 85" x 22") and sits on a concrete floor. Having moved it into its final position by sliding it along the carpet, it will now have to remain there because moving it in the opposite direction would be against the nap of the carpet, and the machine would probably collapse in the attempt.

It wasn't possible to use a K'Nex ball as the bouncing one because it wouldn't have been heavy enough to trigger the pay-outs; indeed, there were difficulties anyway as described later. Apart from the labels, this solid-rubber ball is the only non-K'Nex component in the machine. It weighs 49g and is 42mm in diameter, twice the weight of a hollow plastic K'Nex one (which has a diameter of 45mm).

Game Sequence

From when the rubber ball is at the bottom of the chain-lift, the following actions take place:

1) The ball is lifted as soon as two prongs on the chain-lift reach it (the average wait is 2 secs.) [Step 14, 4th photo]

2) After 4 seconds, any previous bets are reset and the 'Bet Now' flag is set. [Step 17]

3) For each number backed, a yellow indicator is set above the slot's label and, internally, a lever is set which diverts the path of the rubber ball to prime the pay-out for that bet [Step 11, 1st photo; Step 4, photos 3 & 4]

4) After a further 8 seconds, the 'Don't Bet' flag is set [Step 18]

5) After a further 2 seconds the ball is released so that it can run down the chute to the top of the pinfield—any bets made after this is considered cheating! [Step 15]

6) 1½ seconds later the ball reaches pinfield [Step 8]

7) The ball bounces down the pinfield in a chaotic manner and ends up in one of the 14 columns [Step 8]

8) The ball runs into the pay-out mechanism which corresponds to the number of the column [Step 6]

• If that internal lever is set—see 3) above—the ball is diverted so that the pay-out is triggered
• Otherwise, or after the pay-out, or after landing in a 0 column, the ball returns to the bottom of the chain-lift.

This Instructable

First of all, a quick word about the lengths of K'Nex rods: the official lengths of K'Nex rods are inaccurate. The lengths given in this instructable are the correct ones; for example, red rods are 130mm long, not 128mm, and grey rods are 192mm long, not 190mm.

The machine is too complicated to give step-by-step instructions for it. Instead, various aspects of the construction have been covered so that a creative K'Nex user (or maybe a wood- or metal-worker)? could be inspired to build something similar. It is not a suitable build for non-creative instruction-followers. Further information is available just by asking.

The steps which follow are broadly in the order in which the machine was built.

1) Build a minimal base—it can be extended later

2) Make the pay-out reserves

3) Make the pay-out triggers

4) Make the win-priming mechanism

5) Make the pay-out gates

6) Make the ball channels to go under the pinfield

7) Extend the base

8) Make the pinfield and install it over ball channels

9) Make the column labels

10) Make the channels which direct the rubber ball from the bottom of the pinfield to the appropriate ball channel

11) Make the slots and the indicators above them

12) Make the slots' ball channels which convey the betting balls to the pay-out mechanism levers

13) Make a channel which diverts betting balls to the pay-out reserves to top them up

14) Build the chain-lift

15) Make the ball channel which carries the rubber ball from the top of the chain lift to the pinfield

16) Make the 'Bet Now' and 'Don't Bet' flag mechanism

17) Enable the chain-lift to reset any previous bets and set the 'Bet Now' flag

18) Enable the chain-lift to set the 'Don't Bet' flag

19) Make the 'cash'-box

20) Make the pay-out channel and tray for the wins

21) Make a ball tray for the player's convenience

Oodles of K'Nex and a 49g ball with a diameter of 42mm. And then some more K'Nex. And then a bag containing 200 or so hours of patience.

At least 100 K'Nex balls are required.

I wanted the machine to look machine-like from the front and so many dark-green, grey, black and metallic-blue parts were purchased from eBay sellers. From the side and back I used traditionally-coloured pieces.

The machine was going to be very heavy and so a very strong base was required. It is based on cubes which have 130mm green rods as sides, with 192mm grey rods as diagonals. By having diagonals in different directions on opposite faces, great strength and rigidity can be achieved. Note the orientations of the corner connectors.

The base was provisionally made just high enough to be able to work comfortably on the pay-out reserves which were to be placed on top of it, the idea being that it would be extended and made higher as the build progressed.

The pay-out sizes were originally going to be 2, 4, 6 and 8.

There were going to be four reserves, each in a gentle slope (1 in 6) and each capable of dispensing two balls by letting them roll out.

A win of 2 would release two balls from the first pay-out reserve; a win of 4 would release two balls from each of the first two reserves, three reserves would be needed to release six balls, and four for a win of eight.

The method was to use two prongs in each reserve to create a barrier between the second and third balls, and then get the rubber ball to raise the barrier and then lift a gate so that only the first two balls were released. For a win of four, the first reserve's lever would lift the second reserve's lever too, and for a win of six, the second reserve's lever would lift the third reserve's one.

The problem was that getting the fourth lever to lift the other three levers was just too unreliable. The problem was caused by lag: when one lever lifts another, the is a slight delay before the movement is passed on, owing, amongst other things, to distortion in the structure. Getting one lever to end up moving another three just didn't work consistently.

I decided, therefore, to have just three win sizes, because there was no problem with that, but I wasn't happy—a machine needs a decent-sized win to make it attractive to play.

I then had a brainwave! What if the 8-win reserve was simply emptied if the rubber ball landed in the appropriate column? Since each reserve could hold a maximum of 19 balls (this was dictated by the width of the pinfield), this could be a form of jackpot.

The betting balls were going to be used to top up the reserves: the 2-win one first, then, when that was full, overflow into the the second reserve, then the third, and then the fourth, and when that was full, overflow into the cash-box.

After a series of wins, the fourth reserve could have only a few balls in it, and so I decided to make the randomness a feature, i.e. turning a failure into a success—just by removing those two prongs!

Even when the reserve was full, the way it emptied in practice was variable; sometimes all 19 balls would be won, and at other times anything from a few all the way up to 14, 15 or 16 or so. If the ? column was entered twice in succession it is likely that no balls would be released the second time—that is why the pot-luck amount is stated as 0 to 19.

The gradient of the payout reserves had to be just right—if it were too steep, the balls would run down so fast that the pay-out gate would be shocked open and many of the balls would run out; on the other hand, if the slope were too gentle, the balls would take too long to roll out before the gate closed.

In practice, the 2-win and 4-win reserves emptied every now and then because of the shock of the top-up balls hitting the gate. It was not possible to relax the slopes any further because the winning balls would take too long to roll down, and so a damper was installed in each channel to slow down (but not stop) the balls as they rolled in. These are shown in the second and third photos above.

It took a long time to get this right.

The image above is a photo of the left-hand end (from the rear of the machine) of one of the pay-out levers. It is shown schematically in the PDF below.

The idea was that, if a bet was successful, the rubber ball would land on the end of the pay-out lever (there are four of them, and each runs under the whole length of the reserve) and tilt it so that the lever rises at the far end to release two balls.

The immediate problem was that the lever did not rise far enough to release any balls because the rubber ball was not heavy enough. Some extra leverage was required (annotated in the above picture). The black trigger rod is joined on its right to a yellow rod via a white connector with a pivot at the right-hand end of the yellow rod. This rod pushes down on the pay-out lever 41mm to the left of the pivot. The total length of the black-plus-yellow rod and the red connector on the right is 192mm + 20mm + 86mm + 10mm = 308mm. Since the yellow rod pushes the pay-out lever 41mm from the pivot on the right, the mechanical advantage is (308 - 41) / 41 = 6.5.

The next problem was that the pay-out gate did not stay open long enough to let the balls escape. The solution was to lengthen the run where the rubber ball rolled off the black pay-out lever at the left-hand end (looking from the rear). This was done by adding an orange two-way connector to the end of the black trigger-rod. This not only improved the mechanical advantage to (337 - 41) / 41 = 7.2, it made the rubber ball hold down the black rod just a little bit longer.

When a bet is made, the plastic K'Nex ball falls into a cradle which becomes unbalanced, causing the ball to roll off. As the cradle turns, it primes the pay-out for the bet. The ball subsequently finds its way to the pay-out reserves to top them up.

The cradle can be seen in the first and second photos. In the machine, the cradles are made in pairs.

Ordinarily, when the rubber ball runs down the ball channel it is directed to the bottom of the chain-lift. After a win, it must fall onto the pay-out trigger first.

The third and fourth photos above show the rubber ball gets diverted if a bet has been made. Only part of the mechanism is shown. They were made in pairs, hence the red rods.

The third photo shows the mechanism in the unprimed state. The rubber ball will slide down the slope and end up at the bottom of the chain-lift.

The fourth photo shows the state after a bet has been made. The falling rubber ball will now land on the pay-out trigger which would be underneath it, and from there go to the bottom of the chain lift.

The fifth photo show the mechanism in the actual machine, but is it so deeply embedded that it is difficult to see what is there!

The PDF below shows the primed and unprimed states side-by-side.

The pay-out gate is the bit at the right-hand end of the pay-out lever (looking from the rear) that rises so that balls can roll out into the pay-out tray. Before this happens, two prongs (actually, yellow rods) rise between the second and third balls so that only two balls roll out (this is schematically shown in the PDF below).

The gate is finely balanced so that only a little force is needed to raise it—that rubber ball has to be able to raise three of these for a win of six balls.

For a pot-luck win, the two yellow rods which rise between the second and third balls are absent. When the pay-out gate is raised, the balls tumble out in a fairly haphazard manner, sometimes emptying the whole trough and sometimes releasing only a few, or anything in between.

These are the five channels which sit underneath the pinfield.

When the rubber ball lands in a column, it has to be directed to one of the channels so that any win can be handled. Looking from the left-hand side of the machine, the channels correspond to column values of 2, 4, 6, ? and 0 respectively.

The 2- and 4-column channels and the 6- and ?-channels have been combined, as shown in the fourth photo above. There are 30 connectors along each length. The gradient is about 1 in 9.

Now that the size of the machine's footprint is known, the base can be extended and made the required height. If this were done later, the machine would be too heavy and cumbersome to handle.

OK, there's no such word as 'pinfield', but it adequately describes what it is!

The first two photos show the pinfield from the front and side.

The layout is illustrated in the first PDF below. Eight 'blockers' (shown as \ or /) have been added to change the column distribution so that the chances are evened out a bit. The percentage of times each column is visited appears in the second PDF and is based on a few thousand trials (yes—really!). There will inevitably be a sampling error in these figures. The blockers can be seen in the third photo.

Also in the second PDF are some interesting statistics relating to the combinations of bets. If more than one bet is made per game, the pot-luck reserve gets topped up more quickly than if there were only a single bet, because more balls are being used to top up the reserves; however, the standard deviation of the expected return will be relatively low, making the chance of a profit after a couple of dozen games quite small.

If a pot-luck bet wins, it would be sensible to avoid that bet for a few games because the reserve will be low, but how long to wait will depend on how many bets have been made. Interesting, isn't it?

The column labels have been designed so that when a ball lands in a column, the label swings for a second or so. This has been done to allow a player who was not looking at the time to glance back at the columns and have a chance to see where the ball landed.

But what win sizes should the columns have? I tried to have no columns with a zero, but whatever blockers were used the return was over 100%. I wrote a wee program to simulate the behaviour of the machine before making a final decision and values were chosen which would give a return of around 65% to 85% depending on the nature of the bets made (the pre-decimal version of the Bryans Bullion machine had a return of 24/35 = 68.6%).

These are the channels which carry the rubber ball from the bottom of each column to the appropriate ball channel. The columns labelled 0 don't have a pinfield channel—the ball drops straight into the ball channel.

When a ball is inserted, a block of black four 4-way connectors is raised, making a yellow 'bet acknowledged' flag appear at the top.

Four 130mm rods are used to join the black and yellow connectors, and the unit slides upwards inside a frame.when a ball is inserted.

This part of the machine is very compact—it had to fit underneath the slots!

A betting ball has to end up, via the bet's cradle, falling on the appropriate trigger for the value of the win, and this is achieved via a convoluted system of mini-channels.

Each betting ball, after having triggered a 'ball inserted' flag for the bet, ends up travelling to the 2-win pay-out reserve. If there is room for it, it tops it up, otherwise it tries the 4-win reserve. Then in turn, it tries the 6-win and ?-win reserves, and if all of them are full it falls into the cash-box.

The chain-lift has seven ball-lifters and about 150 links. They have all been oiled because K'Nex links are prone to wear.

The design of the ball-lifter is shown in the first two photos above. It is slightly different from the normal K'Nex recommended one in that there is a white rod on the orange connector, not a green one—this is because it is lifting a rubber ball, not a (larger and lighter) K'Nex one. Also, there is a blue clip on the end of each orange connector which lifts the ball, not a one-way connector.

When the ball reaches the top of the lift, the white rod pushes the ball off so that it runs down the channel to the pinfield. The orange connector which holds the white rod has to be fitted very tightly on the link because of the weight of the rubber ball. After a bit of use, the ends of an orange two-way connector diverge very slightly and do not grip an inserted rod as tightly as a new one, and so ones in very good condition have to be used if glue is to be avoided.

There is a jockey wheel and there are a couple of brackets as well—these provide tension in the chain to reduce vibrations which otherwise would occur. The vibrations happen because there is a very slight jolt as the links pass over the driving and driven red gears and there is some kind of resonance going on.

It takes about 16 seconds for the rubber ball to be lifted from the bottom and released at the top.

Sections of this channel are shown above.

Note that there are connectors both below and above the slope that the ball runs down. The channel was first made with only the connectors above, but owing to the nature of the K'Nex pieces, there was a slight bow to it. The channel now is dead straight.

The player is allowed to make a bet only after any bets for the previous game have been cleared and the ball has not entered the pinfield. This has not been enforced but relies instead on a flag which reads 'Bet Now' or 'Don't Bet'. What's the point of enforcing it given that the player could grab the rubber ball and plonk it in a winning column?!

The mechanism is large and clunky, but it works!

The photo above has been annotated to show what happens.

This uses a similar method to that for setting the 'Bet Now' flag. The flag is set near the top of the chain-lift, just before the rubber ball is sent down the channel to the top of the pinfield.

When bets are mode (using K'Nex balls—the rubber ball is never touched), the balls are used to top up the pay-out reserves. The 2-win one is topped up first, because every win takes two balls from the reserve. The 4-win one comes next, then the 6-win, and finally the pot-luck one. When all the reserves are full, the balls are sent to the cash-box ('ball-box' doesn't sound quite right!).

This has a lid on it so that the player can't see how fast their supply of balls is being gobbled up. The inside of it is shown in the third photo.

When won balls roll from the pay-out reserves, they need a channel to convey them to the pay-out tray. The channel needs to be large enough for balls not to jam in it when more than one reserve is releasing balls.

The first photo above shows the end of the pay-out chute from the back-left of the machine. The second photo shows bits of the edge of it from the left-hand side.

The second to seventh photos show various aspects of the pay-out tray. It needs to be large enough to hold a few dozen balls and the front of it needs to be high enough to prevent balls overflowing on to the floor. It has been covered so that the end of the delivery chute is concealed.

The player's supply of balls can be placed in this removable tray, as well as any winnings. It makes it easy to see how many balls are won from a given supply. If winning balls are placed in this tray whenever it is empty, the player can see how quickly they shrink in number!