# Roulette Experiments - Ball Decelleration

Hi,

I have concluded experiments on a Huxley Wheel (Mark V) (used wheel). The wheel’s track appeared to be in good condition but it had minor traces of use. The ball is 21mm teflon.

The experiments were conducted using photosensor timing for the ball so the accuracy is at 1ms level.

The findings where a little bit troubling. I attach a graph of the wheel decellarion on 10 consecutive CCW spins. The X axis is the distance travelled (in rounds) from the start of the spin and the Y axis is the time for each round (in ms). As one can see the decellarion is not constant for every spin. There are spins that have for the first rounds almost exactly the same timings and differentiate afterwards.

My question is the following: Is that differences to be expected from variables that change for each spin (eg. moisture from the hands of the croupier, spin given to the ball etc) or there is something wrong with the wheel? I have to note that the person spinning was not a professional dealer.

The wheel is exactly the same as the one in the following video:

But what you see is something else.
Spin has a what we call the knee point, where the ball rapidly changes deceleration, after 1000 ms rotation the next rotation can be 1200
But after 1220 next one may be 1280.

You’re right about the knee point but that’s not what I’m talking about. I’m saying that the ball decellarates in different manner in different spins. So two spins (eg 880 and 884) start by having the same sequence of timings and then spin 880 decellerates faster than spin 884. (the numbers are spin ids not revolution times).

In short, based on your experiments, decellarion parameters are constant over a period of time (eg. 1 hour) or the change for each spin?

here can be other reasons to ball decelerattion changes - for example ball fricttion or some other manier to throw ball.

The X axis is the distance travelled (in rounds) from the start of the spin and the Y axis is the time for each round (in ms)
Must be not distance traveled from start but distance left till ball dropp. And if spins is 7-8 round durattion then can be everything because it is very sensitive to throwing manier.

This is based upon my knowledge when it comes to physics.
First i think you have made the experiment in the wrong way.

There is two sides of the story, i will start with the first side of the coin.

You can not measuring the ball at any static point or reference point.
Because the ball will not be at 845 ms at the same spot each spin and it does not matter what time intervals you use.

Let me explain.

First ball in the beginning of spin is at chaotic state and after around 700 to 800 ms the ball start rolling on the ball track, leaving the chaotic state.
This does not happen at the same spot at the ball track during different spins.

There is a moment on the ball track where the ball is “faster then” and “slower then” a certain time interval.
So during one spin the ball might be faster then and slower then 845 ms at 12 a clock and during next spin the ball might be faster then and slower then 845 sec at 6 a clock.
The exact same time interval divide at different spot on the ball track.

This is one reason you get different time frame from A to B or traveling distance.
One other reason is that if you predict very early, then result is not as stable as if your predict later during spin.

Some methods predict very good at 0.7 ms and other don’t.
I know why and that is the other side of the coin that we can discuss.

I know how to deal with multi drops and what is the correct way using the right reference spot to get acc results measuring the ball.

First of all thanks everybody for taking the time to reply.
Ok, in order to clarify I’ll tell you my understanding of the problem and you tell me your thoughts:

We are studiying the part of the motion until the ball leaves the rotor. Lets get into some math. Lets define:

θ: the angular position of the ball.
ω: the angular velocity of the ball
t: the time

We set θ(t=0) = 0 and ω0 = ω(t=0). Note, that we do not need to set t=0 in the begining of the spin (as lucky_stike correctly said the ball is in a “chaotic” state) We can set it eg. in the second time the ball passes from the reference point. It doesn’t matter.

Now, there should be functions like ω(ω0,t) and θ(ω0,t). These functions will contain a paramer vector P = (p0,p1…pn) of some parameters (weight of ball etc). I have to state that the P vector does not necessarily remain constant over time.

If the P vector remains constant then the future position of the ball depends entiry on the initial velocity ω0.

So 2 questions:

1. Is that modelling of the problem reasonably accurate?

2. Based on your observations the P vector remains constant?

[quote=“lucky_strike, post:6, topic:1009”]First ball in the beginning of spin is at chaotic state and after around 700 to 800 ms the ball start rolling on the ball track, leaving the chaotic state.
This does not happen at the same spot at the ball track during different spins.

There is a moment on the ball track where the ball is “faster then” and “slower then” a certain time interval.
So during one spin the ball might be faster then and slower then 845 ms at 12 a clock and during next spin the ball might be faster then and slower then 845 sec at 6 a clock.
The exact same time interval divide at different spot on the ball track.[/quote]

You are right. But I do not use time intervals. I just measure the duration of the last revolution. I don expect it to be the same each time. But I expect the deccelaration line will have the same shape and the same “parameters”.

Can you explain what is a multi-drop?

I send you private PM
I don’t want to discuss future details in public.