Improve distribution of stims across phase

[Jimbles]

Distribution of stim phases should be dependent on the number of stims per injection.
Currently it gives either 360 phases or as many as is possible based on frequency. But this sequence could be truncated if the measurement time is too short. This would give an uneven distribution.

Instead the sequences should be set such that as full a distribution as possible is used for a given frequency and injection time.

[JeanRintoul]

Is the reason to get a large distribution over as many phase angles as possible because the output phase of the waveform at time of injection can't be known? Makes sense that you'd want to vary the distribution based length of sequence/frequency though.

[Jimbles]

Hey Jean!
Currently, we just set the Keithley current source going and then when we switch injection pairs we do not reset the phase back to zero - this can easily be done however, as I added the connections on the Arduino shield.

What we are talking about here is the stimulation pulse for evoked responses i.e. the one which gets sent to a whisker stimulator, or a LED for visual stimulation . Kirill found that there are artefacts sometimes dependent upon the phase of the EIT current at the moment the stimulus is applied. So its best to randomise it, so it will decrease with averaging. We do this by looking at the "zero phase marker" of the keithley current source, and waiting a controllable amount of time until stimulating. How it works right now is fine for our normal use case of higher freqs and long measurement times, but we are now trying to do low freqs and short measurement times.

This has also reminded me that I havent replied to your email...

[JeanRintoul]

Thanks for clarifying that you don't reset the phase on the Keithley - that explains the comment. In the 32 tetrapolar MFEIT system I've been working on recently phase is reset to zero(or user defined) for every measurement to limit variation due to phase differences. It's currently in pcbfab and I'm just awaiting it's return and investigating pyEIT integration(python port of EIDORS).

I see, so you send an AC signal to a whisker stimulator, and dependent on the phase you get different effects on the nerve? - makes sense. Also, understood for the shorter measure time for fast neural eit you'll want more precision as everything is at SNR floor or close to it presumably.

Cheers,
Jean