Idea for a new Show

It might be easier to stick it on the digit read out. I see some little white tick lines on the 1/10 dial, they are distinct and evenly spaced. You might have better luck placing your sensor on that. But here's an alternative idea; I wouldn't be surprised if you can get a measurable variance if you stick a photo diode right up against the glass where the white tick lines are for the 1/10 dial and hit them with a cheap laser pointer.

Here's an example circuit for a photodiode from LetsMakeRobots:

http://letsmakerobots.com/files/userpics/u1533/Photodetector.jpg

if you search for photo detector circuit you will find many that either use op amps or transistors.

That would give you a signal that would go higher and lower as a line passed under it, but it likely wouldn't be logic level high and low, unless you're just lucky. It may be for example 3.1-3.7V in one state and 1.6-2.2V in the other. From there you could either use an analog input and logically separate out these states, or use a comparator IC against a voltage divider potentiometer to chop your analog singal into a digital readable high low.

Like this (copy pasted, not mine)

http://4.bp.blogspot.com/-cvKWtb9BmUc/Tvk2p1s813I/AAAAAAAAAXA/OSBWjQrjcc4/s1600/voltage+divider+with+comparator+circuit.…

but of course the two resistors are the pins of the potentiometer.

Hope this idea helps

I would look into using the Pi camera and looking at the digits if you already have a Pi.  There might already be things out there for reading digits from video or pics.

I was going to say the same thing. A webcam or pi camera module with a machine vision library might be the easiest way to get consistent output while dealing with the access limitations.

I've seen projects with camera's before...

But the 'fun'/'challange' of this project would be; hardware needed; a 'old mouse'

Also just counting the frequency of the spinning disc is much easyer to implement than OCR-ing the digits on the meter.

Your Pi would also be working quite hard to process all the images.

I've seen projects with camera's before...

But the 'fun'/'challange' of this project would be; hardware needed; a 'old mouse'

Also just counting the frequency of the spinning disc is much easyer to implement than OCR-ing the digits on the meter.

Your Pi would also be working quite hard to process all the images.