I was wondering if anyone has done any investigation into arduino EMC? I'm planning to use one as the base for a stimulation device for use in an MRI scanner (presents an electrical stimulus to a subject), and it's important that no noise is produced by the device at the frequency the scanner works at (400MHz). I realise that's a long way from 16MHz, but, harmonics will be present and MRI scanners are incredibly sensitive because the signal they detect is so small.
I could put the entire arduino inside a screened can, with shielded RF connectors for all inside-outside lines, but of course this increases cost and complexity (I want to make the whole thing for about £100), particularly because components need to be relatively non-magnetic (so ferrite rings are no-go's, and ferrite inductors used carefully).
Are there any arduino variants which are known to produce less EMI, for example though a better designed PCB?
it's not going inside the actual scanner bore, on the outside, so that isn't a concern.
However, we have used devices such as USB webcams which with adequate shielding operate fine inside the scanner bore, despite the gradients and RF fields, and Siemens scanners come with bluetooth ECG monitors that clip onto fingers.
The issue is that the stimulus is delivered by wires going into the scanner bore, which even with non-magnetic RF chokes or cable traps attached will result in coupling in external noise. Minimising noise emissions and that on the output cables from the arduino is essential.
typically less than 10mA, but as the load varies significantly I'm using a current source op-amp arrangement. I've already built an analogue version of this that uses a 555 timer for waveform generation (3ms pulses at 3Hz), an arduino generating the waveform will be much more flexible (different rates, duty cycles, train lengths etc). I got away with a 5V power supply before, but I think this time I will beef it up to 15V just in case the load is a high resistance.
I've thought about using a photodiode, but it's difficult to calibrate and probably even more difficult to get it to provide a specified current when the load varies so much, and in order to correlate fMRI results and publish you really need to know what the simulus is!. The fibre optic thermometers and pressure sensors cost thousands and use things like interferometry to measure the precise size of a GaAs crystal in the probe which is probably quite difficult to implement with an arduino.. If anyone has made a fibre optic thermometer using an arduino I'd definitely be interested in that though!