EEVblog #525 - Bank Note Acceptor Teardown

[Stonent]

I looked at a US$ 5 and US$ 20 bill under UV light - images attached.

The $5 has no ID strip, but I did see a quasi-random UV image centered on Lincoln's image.  Not sure if this is real or something that has transferred to the bill.

The $20 has the embedded security strip which glows green. However, three of the four corners also fluoresced - but the uneven pattern looks unintentional.

I suspect the glowing corners of the money are glowing for the same reason a crime scene glows. 

[Stonent]

Here are the last 3 $5 bills



So the one that Dave had was the last one from 2006. If the firmware was 2002 it would be looking for the one in them middle that came out around 1996 or so. The one on top is of the same redesign period that the current $1 bill is.

The $1 bill has not really been changed since the 50s or 60s if I recall correctly. Last I heard, it was just not economically feasible to redesign it. When they did the major bill redesign in the late 90s they added color shifting ink. On some bills there was a logo and on some it was the treasury seal.  If tilted it alternates between green, gold and black.

A number of years ago I did see a counterfeit bill that would pass the pen test. Because the person appeared to have printed the the bill on some sort of waxed paper.  Since waxed paper won't absorb the ink, it doesn't turn colors.

The bill failed the feel test pretty badly. Surprisingly this bill came as change to one of my dad's ex girlfriends.  She decided to keep the fake $1 bill as a novelty piece.

The last documentary I saw on how money was made showed they purchased the scraps and trimmings from a blue jeans factory as part of the base of the bill.

[JackOfVA]

I looked at a US$ 5 and US$ 20 bill under UV light - images attached.

The $5 has no ID strip, but I did see a quasi-random UV image centered on Lincoln's image.  Not sure if this is real or something that has transferred to the bill.

The $20 has the embedded security strip which glows green. However, three of the four corners also fluoresced - but the uneven pattern looks unintentional.

I suspect the glowing corners of the money are glowing for the same reason a crime scene glows. 

It seems that cocaine fluoresces green under UV light.  This might imply that this bill has drug residue.  I had 3 $20's in my wallet and only one had the corner fluorescence.

[Stonent]

I looked at a US$ 5 and US$ 20 bill under UV light - images attached.

The $5 has no ID strip, but I did see a quasi-random UV image centered on Lincoln's image.  Not sure if this is real or something that has transferred to the bill.

The $20 has the embedded security strip which glows green. However, three of the four corners also fluoresced - but the uneven pattern looks unintentional.

I suspect the glowing corners of the money are glowing for the same reason a crime scene glows. 

It seems that cocaine fluoresces green under UV light.  This might imply that this bill has drug residue.  I had 3 $20's in my wallet and only one had the corner fluorescence.

I guess only one had been inserted into a dancer's bill validator? 

[ExoUY]

well, that was horrible! But anyways I will post the video, again, sorry for my bad English also I get nervous and start repeating the same words :$
http://youtu.be/rqMqUsUidAk
(it will take a couple of hours to upload.
But to repair the shame of such a bad video, here's a goody I remembered I had on the wall of my lab:


Uploaded with ImageShack.us

It's for a Bvall that uses the same tech as the one Dave disassembled
Hope someone finds it interesting .

[grumpydoc]

I don't understand why someone will willingly pay 3-5 times the retail price for a snack or some soda.

Usually because the vending machine is placed somewhere that your options are limited, ideally where you're more likely to want a drink or snack.

e.g in the gym foyer - you're hot, thirsty and possibly a little hungry after your work-out and the nearest shop is a mile away - what do you do? Walk to the shop or just stump up the inflated price for something from the machine?

[David_AVD]

I don't understand why someone will willingly pay 3-5 times the retail price for a snack or some soda.

Usually because the vending machine is placed somewhere that your options are limited, ideally where you're more likely to want a drink or snack.

e.g in the gym foyer - you're hot, thirsty and possibly a little hungry after your work-out and the nearest shop is a mile away - what do you do? Walk to the shop or just stump up the inflated price for something from the machine?

I saw an online trade article where one of the drink manufacturers (Pepsi or Coke) were going to trial a vending machine that varied the price according to the local conditions.  eg, It discounted the sell price when the weather was bleak and sales were low.

[nitro2k01]

What's the talk about the flash chip needing to be connected to the CPLD to program it? You can program it using 5V only, and commands that can be issued as writes from the CPU. I even think the AM29Fxxx series will let you read the rest of the array while one sector is being erased or programmed, but even if that were not the case, you could easily copy a small programming routine to RAM and run it from there. You can also lock individual sectors through hardware, which then requires 12V programming in order to unlock, so you don't have to worry about locking yourself out in case of a bad in-circuit programming. The IS61C256 is a 32 kB SRAM chip, for what it's worth.

The difference in the sensors on the sides, vs the big one is obviously that ones on the sides are detecting light through the bill whereas the big one is detecting reflected light (thus the white blocking panel.) If you want to determine for certain whether an LED is IR or UV, hold an object (like maybe a bill ) and see if it fluoresces.

When I ripped the IR pass filter out of my camera, I took an IR LED and took some pictures of Swedish bank notes:

20 kr:

Swedish 20 kr note in IR light by nitro2k01 (Gameboy Genius), on Flickr


Swedish 20 kr note in IR light by nitro2k01 (Gameboy Genius), on Flickr


Swedish 20 kr note in IR light by nitro2k01 (Gameboy Genius), on Flickr

In visible light:


100 kr:

Swedish 100 kr note in IR light by nitro2k01 (Gameboy Genius), on Flickr


Swedish 100 kr note in IR light by nitro2k01 (Gameboy Genius), on Flickr
(This is the reflective strip.)

In visible light:


Those are the kinds of tricks you'd be looking at on just about all notes, probably. Something like half of the note is completely white in IR light, so you'd probably need a very specific kind of ink that completely reflects IR while it absorbs red light well.

[Stonent]

Winston Churchill is watching you spend.

(Ok well it looks like him anyway)

[firehopper]

actually, from what I remember.. all US bills use the same paper.. Thats likely to never change.. its a form of rag paper..

@JackOfVA which model of dollars were you using, the ones I tested were from 2009, they tend to move the security strips and mix and match different patterns and UV reactive elements with each model


===-=-=--==--==-=-=--=

Here are some images of how the dollars look under UV light.

My camera does not capture it very well, eg it does not capture some of the variations in the fibers and how they react to the UV light

but each value reacts differently to the light

Here are some examples:

$5   

$20

($20 has the security strip in a different location, and makes use of a few different fiber types which all react differently to the UV light, causing a a few different subtly different shades of of the color of the bill to appear. (couldn't get the camera to capture them) (the $20 also uses reflective ink with some glitter in parts of the image, but under UV light, they are completely non reactive and lose their glitter effect)

$1   

The $1 has a very uniform response to the UV light (they most likely use cheaper paper and ink)

For the bills where verifying the authenticity is important, they use many different fibers and inks which react to different forms of light in specific ways. the formulas are also likely kept secret, thus making it pretty much impossible for someone to make a fake one and have it pass every test.

The down side is that it is also easy to fail the test when it comes to machines. Due to the paper and ink used, they are not the most resilient and thus after a while, they will become  useless in higher end machines, for example a dollar that is beginning to fade may not pass the test that the ATM machines that accept cash deposits may do, but will work just fine in a vending machine.


=-=-=-=-=-=-=-=-

A side note, never buy anything from a vending machine. I don't understand why someone will willingly pay 3-5 times the retail price for a snack or some soda.

 A large case of soda quantities over 100, generally comes out to around 13-14 cents a bottle (16oz or around 473ml) if you get them wholesale, most stores will then charge $1 per 16oz bottle which is already an insane price especially since a 2 liter (2000ml) bottle is also $1. Then you will have vending machines that will charge in many cases, as much as $2.50 for the soda, and you still see people buying stuff from them.

The vending machines at the college I go to, generally have a long line at the vending machines. (I thought you had to have an IQ above 5 to get into college, but apparently they now allow people to push the limits of negative numbers in terms of IQ )

[philpem]

It seems unlikely to me that the note acceptor is using UV. Those photodiodes aren't going to detect UV at all -- the glass lenses will block it quite effectively. Also most UV LEDs I've seen are clear; the smoked plastic is generally reserved for infrared LEDs.

I've had a dig around (Google search: banknote acceptor infrared) and found some interesting stuff:

A patent on an infrared banknote acceptor - http://www.google.co.uk/patents/US6903342  (actually an IR camera, LCD display and IR illuminator intended to help cashiers spot forged banknotes). Interesting because the patent includes an image of a US $100 bill... complete with the IR pattern

A few more images of US and Canadian banknotes under IR illumination: http://www.fraudfighter.com/counterfeit-detection-id-verification/bid/54918/Counterfeit-Detection-Infrared-Scanners-UV-Lights-Multi-Testers

And some information on the Bank of England's testing procedure for "banknote handling machines" -- http://www.bankofengland.co.uk/banknotes/Documents/retailers/framework.pdf
(TL/DR: bill acceptor manufacturers can ask the Bank of England for access to a selection of English banknotes for testing purposes)


I suspect the acceptor Dave has is alternating between red and IR very quickly during the scan and saving both data points for each position. I suspect this is to stop some cheeky so-and-so from making a forged note using red and black ink -- red for the "IR bands" and black for everything else. The real banknote ink probably passes IR but blocks red (and the rest of the visual spectrum). That probably means you have four possibilities:

Reflects red, reflects IR -- "unprinted"
Reflects red, blocks IR -- IR blocking ink (possibly not used)
Blocks red, reflects IR -- IR-transparent black (as seen on US banknotes)
Blocks red, blocks IR -- "black"

Interesting technology.


It looks like this is Money Controls' latest attempt -- http://www.google.co.uk/patents/US8232867 . Perhaps one of their other patents covers Dave's acceptor?


--- EDIT 20130926 0053BST -- It looks like Coinco have taken over the Money Controls MC2600 series -- here's the manual: http://www.coinco.com/index.php/support/downloads/category/14-bill-acceptors?download=48:mc-series-operations-and-service-manual

--- EDIT 0104BST -- Some information on the "Vend Serial" (MDB) protocol -- http://vending.org/images/pdfs/technology/mdb_version_4-2.pdf

--- EDIT 0108BST -- And someone's hooked one up to an Arduino... http://secrettweakerpad.blogspot.co.uk/2012/03/controlling-vending-machine-bill.html

[eliocor]

Dave, are you sure some of the LEDs are UV?
If they were made several years ago, in general their cases were TO18 with glass window and not plastic.
Also transparent plastic like those in the reader is in general opaque to UV light.

The best way to check the wavelenght of your LED is to use your multimeter in the "check diode" position and to measure the forward voltage drop. It should be about:
Wavelenght (nm) / color / Vf

940

Infrared

1.5

880

Infrared

1.7

635

High Eff. Red

2.0

605

Orange

2.2

585

Yellow

2.4

555

Green

2.6

450

Blue

3.0/3.5

360-405

Ultraviolet

3.5

[orion242]

I looked at a US$ 5 and US$ 20 bill under UV light - images attached.

The $5 has no ID strip, but I did see a quasi-random UV image centered on Lincoln's image.  Not sure if this is real or something that has transferred to the bill.

The $20 has the embedded security strip which glows green. However, three of the four corners also fluoresced - but the uneven pattern looks unintentional.

I suspect the glowing corners of the money are glowing for the same reason a crime scene glows. 

It seems that cocaine fluoresces green under UV light.  This might imply that this bill has drug residue.  I had 3 $20's in my wallet and only one had the corner fluorescence.

I guess only one had been inserted into a dancer's bill validator? 

LOL!!!  Dirty money, dirty, dirty money....

[Alexei.Polkhanov]

The best way to check the wavelenght of your LED is to use your multimeter in the "check diode" position and to measure the forward voltage drop. It should be about:
Wavelenght (nm) / color / Vf

940

Infrared

1.5

880

Infrared

1.7

635

High Eff. Red

2.0

605

Orange

2.2

585

Yellow

2.4

555

Green

2.6

450

Blue

3.0/3.5

360-405

Ultraviolet

3.5

Where does this come from? Does not match my experimental data - see images below.

Wikipedia sort of agree with this table. I wonder if I am measuring it right.

[loveman]

If you really want to get the WTF face, try spending a $2 bill... Fifty cent coins are good too. At least the $1 coins were pushed somewhat recently

I've got a $200 Australian coin, try spending that!

Obama thought about minting a trillion dollar coin to bypass debt limit.

[loveman]

The security features on modern bills are mostly aimed at allowing humans eyes to differentiate fake from real currency with simple tools like a UV light and a magnifying loup. http://www.officezilla.com/p-1012683679-royal_sovereign_rcd_2000_portable_4_way_counterfeit_detector.aspx?%22cagpspn=pla%22&gclid=COnMgdrz57kCFcvm7AodSE0A8A

The above device has a magnetic tape head attached to an op-amp for magnetic ink detection. 

The bill accept machine is much harder to fake out than a human.  It compares a pattern of photo diode readings from several tracks.  These are bills have ink illuminated with different colors that will be reflected, florescence, absorbed, or anti-stokes fluoresced.  Ink jet, xeroxed, or printed facillimallies will not react to light the same way as a real deal bill will.   This is hard to spoof.  It might be possible to spoof if you read the light sensor signals and play back the signals with a light source or a special reverse engineered fake bill. 

[loveman]

Old versions of bill accepters used a reference bank note to scan and compare against.  Xeroxed notes used to get through these until some tweaks were made.  These pre-dated cheap memory or cheap LEDs. 
The reference note and the test note were scanned and compared for similarity. 

The new units must have a very large signature repository.  These devices have to be forgiving enough for somewhat crumpled notes.

[guythepie]

The logo on the device is that of Mars Electronics International, a UK based manufacturer. A big player in coin-mechs etc.
The cable looks to be an MDB (Multi Drop Bus), widely used in vending machines. 9 bit, 9600 (if my memory is correct!). By default coin & bill acceptance is disabled until told otherwise by the vending machine. I vaguely recall that the specific note acceptance can be controlled by the vending machine too.
It's been years since I last worked on vending machines though... becoming a bit of a hazy memory these days...

[eliocor]

Where does this come from? Does not match my experimental data - see images below.

It depends on current passing trough the LED.
In general this measure is made with a 20mA current, but it is not so important: you have only to know that the shorter the wavelength, the higher the Forward Voltage.
It will be impossible that an IR LED will get a 3V Forward voltage, so it will be very easy to identify if it is an IR or UV LED.

BTW, when I make LED wavelength measurements, I use this spectrometer (about 0.1nm resolution in my configuration): you'll discover that changing temperature or forward current will shift the wavelength.

This effect is rather dramatic if you put your LED in liquid nitrogen:


in the second video you can see the Forward Voltage changing together with the wavelength!!

[ryanblace]

It appears the layout of the sensors allows the user to insert the bill upside down.  In the code, you just compare the sensors in reverse.

Also, the 45 degree sensor is reading the surface of the bill compared to reading through the bill for the outside sensors. I'm sure that was assumed, but I didn't hear it on the video.

I wouldn't be surprised if it doesn't actually read the signal for the entire bill as it goes by each sensor.  They could then teach it new bills at the factory by basically:
1) Tell it "here comes a good $1"
2) It scans the entire length (or some decent fraction of it)
3) Remembers what it saw.

Then when it is in the field, after reading a bill, it compares what it saw to what it wanted to see:

62% match to a $1 bill (right side up)
89% match to a $5 bill (right side up)
71% match to a $10 bill (right side up)
42% match to a $1 bill (upside down)
30% match to a $5 bill (upside down)
28% match to a $10 bill (upside down)


Must be a $5 bill.

16% match to a $1 bill (right side up)
24% match to a $5 bill (right side up)
36% match to a $10 bill (right side up)
etc

Must be Canadian bill, reject it.

[EEVblog]

Also, the 45 degree sensor is reading the surface of the bill compared to reading through the bill for the outside sensors. I'm sure that was assumed, but I didn't hear it on the video.

Of course. I might have forgotten to say that.
It's obvious the two inner sensors reflect back, but they would also likely go through the note and reflect back due the (non back) surface on the opposite side.

[mikeselectricstuff]

It seems unlikely to me that the note acceptor is using UV. Those photodiodes aren't going to detect UV at all -- the glass lenses will block it quite effectively.

That's only the case for shortwave UV.
It may be looking for fluorescence, so the detectors don't need to be UV sensitive (UV detectors are typically exotic and expensive, usually made bu Hamamatsu)

Also most UV LEDs I've seen are clear; the smoked plastic is generally reserved for infrared LEDs.

Agreed. A quick test for UV is some fluorescent material, e.g. a highlighter pen.
I doubt a video camera would have significant sensitivity at wavelengths below visible violet

[Stonent]

It seems unlikely to me that the note acceptor is using UV. Those photodiodes aren't going to detect UV at all -- the glass lenses will block it quite effectively.

That's only the case for shortwave UV.
It may be looking for fluorescence, so the detectors don't need to be UV sensitive (UV detectors are typically exotic and expensive, usually made bu Hamamatsu)

Also most UV LEDs I've seen are clear; the smoked plastic is generally reserved for infrared LEDs.

Agreed. A quick test for UV is some fluorescent material, e.g. a highlighter pen.
I doubt a video camera would have significant sensitivity at wavelengths below visible violet

Just reminded me of the video you made on separating the wavelengths of light. 

[eliocor]

It may be looking for fluorescence, so the detectors don't need to be UV sensitive (UV detectors are typically exotic and expensive, usually made bu Hamamatsu)

Once I disassembled a rather old French chip card reader (I believe used on ATMs before the advent of Euro):
the technique used for checking the compliance of the card was a VERY small wood lamp and three photodiodes, each one with a very narrow bandwidth optical filter.
I think the card was treated with one or more fluorescent dyes and only with the right combination of emissions it was accepted by the reader.
The reader was very small for its age and surely it wasn't cheap, due to narrowness of the filters.

I admit I have never checked my ATM/credit cards with UV light...

[mikepa]

I guess it must be enabled from the vending machine controller. When the machine is not working the acceptor should not accept notes.

That's exactly what's going on.  this makes sure your money doesn't get swallowed up by a machine that's empty or malfunctioning in some way or another. Otherwise these things are dead reliable electrically.  Their biggest issues as dave noted is that the mechanicals and optics get absolutely filthy. After you push maybe a thousand bills through it you need to just open the thing up clean off the feed belts and sensors and you're "money"

as an aside, these things aren't really about detecting counterfeits, there's really very little of that going on, but rather to make sure the bill is the right denomination.

Switch 6 at 5:10 seems to be enable/disable for host ready.