This evening I was truing one of my bicycle wheels. I started out with a truing stand – literally, a metal bolt identifies high spots on the rim by rubbing against them. I managed to get the wheel pretty true. But then I remember that I also have a DTI, accurate to 0.01mm. It’s like a red rag to a perfectionist. After a bit of futzing with that, I could get the wheel true within ~0.1mm and so I stopped there. The first time I hit a bump it’ll blow that precision away! But it got me thinking … how you could automate this (rather laborious) wheel truing task.
After a bit of thinking, I realised you can utilise the fact that the wheel is made of conductive aluminium, and therefore can act as one plate of a capacitor. Sure enough, wikipedia agrees. Unfortunately, the sensors in question are quite pricey – $120 – so I doubt I’ll ever get one.
Could I make a homebrew version? I think it’d be hard. Given how small a wheel rim is, you’d probably only manage a plate 10x10mm. If you got the wheel true within 1mm, your plate distance would be maximum 1mm. So that gives a capacitance of 0.88pF. That sounds pretty small to my ears. Charged up to 5v and draining through a 1M resistor, it’d drop to around half its voltage in a mere 0.88 microseconds.
The typical low-tech way to measure unknown capacitances is to stick it in as part of the timing circuit for a 555 astable and then measure the resulting pulse frequency. Fine in theory, but I don’t own a frequency counter .. and also the frequency would be in the megahertz range.
Microchip have a nice technote describing how to use a PIC microcontroller to measure small capacitances – and when they say small, they mean sub-pF. I’m still digesting that.
I’m tempted to skip all the tricky stuff and just build a larger scale version to test the concept … just to see if I can make the basic method work at all. Whilst I started off thinking about measuring distances, it occurs to me that a large-scale variable-capacitance controller oscillator is … a theremin!