Hall Effect sensor

C

cadcamboss

New Member
Hi all,
I have a student who is after making a coin recognition system for an electronic money box. Heard that a hall effect sensor can be used to recognise the field effect of different coins passing by it. Has anyone done any work in this area and can point me in the right direction? Schematics and link to Pic Logicator/ basic commands etc
 
Is it anything to do with your "opto slot interface" from a year ago .... you never replied boo hoo... we felt very neglected ;)

My 15 quid coin counting robot certainly uses magnetic sensing and was very good until it broke after 2 months.
It only detects Chinese coins now.

How much have you Googled (only posh people Alta Vista)?
i.e. have you researched this task?

I must say i've never seen it mentioned on this Forum, though you have searched obviously.
As to whether your average cheap Hall effect sensor will be up to it I don't know as there are so many types.

I take it you want code as well?
 
Thanks for getting back so soon. Sorry not replied to opto slot thread..please accept my appologies!! Googled it but come to a blank re actual use of sensor..Waiting for a mate to bring in a cheap money box to have a look. At moment have looked at weight sensor using QTC material, peizo transducer but not accurate enough to distinguish between coins. Also looking at timer to time different coins dropping between two optical sensors..the good old opto stot option again!!! :-) Hopefully the different weight of coin will give a time difference big enough to distinguish between different coin denominations
 
Thanks for the link..looks very comprehensive...will print off and have a read.. Looks like I will need to set up a simple interface circuit and have a play at putting coins across it.
 
cadcamboss said:
Also looking at timer to time different coins dropping between two optical sensors..the good old opto stot option again!!! :-) Hopefully the different weight of coin will give a time difference big enough to distinguish between different coin denominations
Click to expand...

What's the rate of accelleration of a falling object? ANY object?
So, what would be the time difference between a big coin and a little coin?

clue:- g ~= 10m/s/s
Tip: you will need to have some very clever mechanics as well.

Have you considered using a system similar to the common metal detector?
Have a browse around for circuits and find one which gives an output proportional to size.

Don't think (I may be wrong) that a hall sensor will work on most coins.
 
BeanieBots said:
What's the rate of accelleration of a falling object? ANY object?
So, what would be the time difference between a big coin and a little coin?

clue:- g ~= 10m/s/s
Tip: you will need to have some very clever mechanics as well.

Have you considered using a system similar to the common metal detector?
Have a browse around for circuits and find one which gives an output proportional to size.

Don't think (I may be wrong) that a hall sensor will work on most coins.
Click to expand...
Thanks..will need an accurate timer me thinks. Was thinking of two optical sensors to trigger a timer on/off using a PIC and then display cycles on screen. Will then need to record time and use this as a reference?
 
cadcamboss said:
Thanks..will need an accurate timer me thinks. Was thinking of two optical sensors to trigger a timer on/off using a PIC and then display cycles on screen. Will then need to record time and use this as a reference?
Click to expand...
There are simple equations for falling objects that allow you to calculate the velocity and distance traveled, as well as the time taken to achieve a given velocity or distance.

These equations are based on the fact that the force of gravity for objects relatively close to Earth equals the mass of the object times the acceleration due to gravity (F = mg). The acceleration due to gravity (g) is constant for all objects up to altitudes beyond many space satellites. From this simple equation, it is determined that the velocity, distance and time are independent of the mass of the object, provided air resistance is negligible.

Time to fall distance
The equation to determine the time it takes to fall a given distance is:

t = √(2x/g)

Example
If you dropped a weight from a height of 64 feet, you could calculate how long it takes to hit the ground.

t = √(2x/g)

t = √[2 * (64 ft) / ( 32 ft/s2)] = √(4 s2) = 2 seconds

Does it follow then that mass will not effect time to fall a given distance?
 
Well spotted and aren't you glad I didn't simply tell you?
ALL coins will take the SAME time to fall.

If you made them roll, then the one with the lowest rotational inertia (NOT mass) would be the quickest.

I've made sensors for very accurately testing the diameter of discs before. The method used was to have an optical sensor with a large-ish detection area and fairly wide angle light source. The idea is that the amount of shadow cast on the sensor is more for a larger object. This could be measured with an anlogue input. The tricky part is knowing when to take a reading as the 'object' passes. Maybe a slotted-opto could be used as a trigger. Just a thought.
 
Well, my Robot Bank didn't trouble Newton too much for sensing... looks like others haven't troubled him much either ;)

Whilst it worked, the Robot Bank worked almost 100% on UK currency sensing.
And I can't imagine anything really leading-edge for a £15 Chinese product.

UK 'copper' became copper-plated-steel after 19..., oh I've forgotten, well you can Google if you want.

Maybe a combination of weight (mass) and magnetics?
The Robot Bank instructions did explain it but my pet Goldfish ate them.

Suggest you buy a Robot Bank "Science Museum" (yeah right) and strip it. You may as well, because the mechanial arms will drop off shortly.
 
Very easy to spot the 'signature' by eye, but one of that hardest things to do in computing. I'd say impossible with a PICAXE maybe with a dsPIC.
Done similar using a microphone for early warning of bearing failure but that used a neural net which could compare 1 million samples to each stored waveform every second.

If you have such a sensor, try using a 'detector' circuit to give an overall waveshape rather than the raw signal. Then see if you can still 'notice' the difference on a DSO trace. Might be possible if the shapes are still very different and can be captured.
 
Dippy said:
I can't imagine anything really leading-edge for a £15 Chinese product
Click to expand...
I agree. So WHAT is in it?
Surely you had it stripped down within a few hours of ownership just to find out?
Must be able to determine the type of sensor even if not the 'blob' on the board.
 
Allright already...
I'll go and take it apart.

Or I could just flog it CCB 'for his student' for a tenner.
It MUST be worth a tenner to keep one step ahead of the kiddies eh :)
 
he he he..a tenner of THEIR money..i'm not THAT worried. Just annoying that the chinese can do it for such a cheap price..everywhere has coin ops..but cant find a definitive answer on how its done!! Got some tests up my sleeve. will keep you posted on progress
 
From one of the "How Do They Do It?" programs:

Coins are sorted on:
size (diameter for US coins, UK might also include width)
weight
magnetic properties (or the lack thereof)

John
 
Sadly, the dismantling of Robot Bank was brought to a halt by the owner.
I may have another go when they aren't looking.

Summary as far as I can see:
1. Coin is pressed through a slot which has a damped strongly sprung lever.
I would guess the combination of strong-spring,damping and curved edges forces odd shaped coins (like UK 20p and 50p) around to get a reasonable 'diameter' measurement.

2. It then passes into a box (thats the bit where i had to stop) which, according to the instructions does the magnetic stuff.

3. Coin falls into hopper - which I noticed was empty when handed to me :(

I have never tried sticking washers through but its sensing of valid coins was very good.
 
20ps and 50ps have odd number of sides, so the diameter is always constant. Roll one along a surface, and you'll see the distance to the top is always constant.

A
 
I did it for my GCSE project.

The coin rolls down a slope, and blocks light gates. The 5p just blocks the 5p gate, the 1p blocks the 5 and 1, the 20p blocks the 1,5 and 20 etc.

Each light gate consisted of an LED and an LDR. To improve the sensitivity, fibre optic cable (1mm diameter), was used to connect the sensor block to the LDRs.

It works 100%.

A
 
Simple. It is activated via a PIR sensor.

People trying to defeat the system are cold blooded. PIR sensors see hot things. Therefore the PIR sensor won't see such people, thus the machine won't turn on. Although metal discs would fool it, it has timeouts on the sensors to stop people sticking bits of paper in.

And in case you did manage to put steel discs in... it gives wooden discs out (to represent coins). I'd much rather have steel discs than wooden discs!

A
 
Last edited:
You mean then it wouldn't let me stick £20 notes in?

You missed a good income stream there then Andrew.

The only answer is physical measurements of the 'coin' and then by using a small collimated thermal neutron source plus a good gamma-ray semiconductor detector and spectrum analyser. By detecting the capture gammas and analysing the spectrum you could determine the ingredients in the 'coin' material .

Ideally, duff coins get a few extra seconds in the neutron beam (just to 'warm' them up) and then spat directly into the offenders trouser pockets.
 
Actually, would it be possible to detect coins based on radiation absorbtion? This might get around anyone using any wooden etc substitutes. It would still need some other help, like the light sensor network stated above.

Question is propably hypothetical, as the equipment for measuring the radiation + the source for the radiation would cost some big $$$
 
The way that a mate and i did it for a end of year project was purely on the OD of the coin, Finely engineered slots will allow certain coins to drop down. This however does not take wood, washers out of the equation however a hall effect switch on the front end to enable the fact that the right type of object has entered the sorter could work.
 
Most currency is alleged to be contaminated with illegal substances.. just cram a sniffer dog in there to detect steel discs.. as they shouldn't be normally coated with anything illegal.

Is anyone going to suggest an 18M2 Touch sensor solution? (or have I missed the boat again?)

Would it be possible to use the coin as a variable pad and use (fixed in place) 'something' on the far side of the coin instead of a finger? or just use the dog's paw.. no point in having a dog and barking yourself.. who's wagging who's tail here anyway.
 
You must log in or register to reply here.
Back
Top