search

cortex-lab / neuropixels

Information about Neuropixels electrode arrays
82 stars 15 forks source link

Noise #23

Open LibiF opened 6 years ago

LibiF commented 6 years ago

Hey, The data I record suffers from strong noise fluctuations at the nyquist frequency of the probe (15 kHz oscillations). It is also periodic in some sense, and oscillations repeat at 50,100 Hz (so they are "carried"/amplified by electrical noise). I was wondering whether anyone else had experience with this kind of noise/how to solve it?

Here is an example: image

nsteinme commented 6 years ago

Hi - it looks like it might be possible that the data is not being loaded properly, which I'm inferring based on the fact that it shifts in time from channel to channel. Does your params.py have n_channels_dat=385 (if you acquired with spikeGLX, because there are 384 channels from the probe and a 385th from the synchronization information). If you use kilosort, ops.NChanTOT should be 385, and if you load in matlab, also use fread(fid, [385 n], '*int16'). Let me know if that might be the problem?

LibiF commented 6 years ago

Hey Nick, it actually is the problem (or at least one, I am checking to see if all works out)... Should number of active channels (ops.Nchan) be set to 374/375 accordingly?

nsteinme commented 6 years ago

374!

On Sat, 7 Jul 2018, 08:48 LibiF, notifications@github.com wrote:

Hey Nick, it actually is the problem (or at least one, I am checking to see if all works out)... Should number of active channels (ops.Nchan) be set to 374/375 accordingly?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/cortex-lab/neuropixels/issues/23#issuecomment-403193694, or mute the thread https://github.com/notifications/unsubscribe-auth/AHPUPygq2mdcZoN1hwgtmv38G_bfoxQDks5uEFnFgaJpZM4VCy6k .

LibiF commented 6 years ago

Thanks so much!!! I think it finally looks good=]