- Neuropixels Phase3A probes have 384 recording channels (i.e. can record 384 signals simultaneously), and up to 966 recording sites (depending on the option)
- Recording sites are laid out in a checkerboard pattern, see geometry note below.
- 10 sites per group of 384 are reserved for selection as internal reference sites and cannot be used for recording (whether they are used for referencing or not)
- Recording sites have ~200k-ohm impedances
- Shanks are 70µm wide and 20µm thick (for Option1, shank is 50µm wide)
- Probe weighs 0.3g, headstage weighs 1.1g.
How switching works (for Phase3A and Phase3B)
There are 384 distinct recording channels, and each channel can potentially receive data from one of several physical recording sites on the shank. Each group of 384 sites (or 276 for Option4, see below) connects to the 384 recording channels in order. This means that site 1, site 385, and site 769 share the same recording channel, likewise for 2, 386 and 770, etc. So using the switches, you can choose to record from any one of the two or three sites that share a recording channel. In practice, while various complicated selections are possible, two configurations are convenient and sensible for most applications: 1) Any continuous block of 384 channels may be selected (e.g. 1 to 384, 100 to 483, etc); 2) A double-length, half-density array can be created by selecting odd sites in the range 1-384 and even sites in the range 385-768. This array then spans 7.68mm continuously.
The sites are arranged in a checkerboard pattern.
Each “row”, consisting of just two sites, has spacing 32µm (center-to-center). Alternate rows are offset by 16µm. Each “column”, of which there are four with 240 sites each, has spacing 40µm. So if the x-coordinates are across the face of the shank, and y-coordinates are up the length of the shank, then coordinates of the sites in order are as follows in µm:
xcoords = 43 11 59 27 43 11 59 27 43 11 59 27 ...
ycoords = 0 0 20 20 40 40 60 60 80 80 100 100 ...
For all sites, the nearest other sites are in the rows above and below, and are 22.6µm away, center-to-center.
The center of the first row of sites is 200µm away from the tip of the shank.
Site faces are 12 x 12µm. A 5µm “gutter” exists between the edge of the probe and the edge of the first sites in from the edge.
For Option1 probes, the geometry differs slightly; see figure.
There are four probe “Options”. All four have the same on-probe amplification and digitization, and can all be used interchangeably with the same headstages and recording equipment.
- Option1 probes (no switches, no amps) have 5mm-long shanks, and have no switches (they can only record from the most distal 384 sites) or on-site buffer amplifiers.
- Option2 probes (no switches, yes amps) have 5mm-long shanks, and have no switches
- Option3 probes (yes switches, no amps) have 10mm-long shanks with 960 total sites accessible via switches; they do not have on-site buffer amplifiers.
- Option4 probes (yes switches, yes amps) have 10mm-long shanks with 966 total sites accessible via switches; they have on-site buffer amplifiers. Option4 probes can only record 276 channels at a time.
In practical terms, the probes with buffer amps (2 & 4) have slightly higher RMS noise levels in saline (~10-12µV RMS compared to 6-9) but may have superior rejection of certain types of noise. However, so far we have been unable to document any situation in which the buffer amps actually do reject artifacts better. Probes with buffer amps also have significantly larger light artifacts. Switched probes (3 & 4) do not seem to have any deficits relative to the un-switched, unless the shorter shank of the unswitched probes is more suitable for your experimental situation.
- “array” or “bank” - for probes with switches, this can refer to a group of 384 recording sites which each have unique recording lines.