Fix register mapping for Neuropixels 2.0 public probes #190
Conversation
jonnew
commented
Aug 3, 2024
jonnew
commented
- Neuropixels 2.0 has a completely different register map compared to 2.0-beta. This is reflected in the changes in this commit.
- Common static helper functions for 1.0, 2.0, and 2.0-beta were moved to BitHelper.cs and NeuropixelV2.cs
- Neuropixels 2.0 has a completely different register map compared to 2.0-beta. This is reflected in the changes in this commit. - Common static helper functions for 1.0, 2.0, and 2.0-beta were moved to BitHelper.cs and NeuropixelV2.cs
- It was left connected to the last probe configured which caused a huge amount of cross talk if i2c bus was used during acquisition
- Gain corrections were not used in NeuropixelsV1 - Gain corrections for NeuropixelsV2 were changed form fixed point multiplication to floating point since we are no longer doing this on an FPGA
|
Its hard to know with 100 percent certainty, due the lack of proper Neuropixels documentation in general and unavailability of their API source code (even though its supposed to be open source given the terms of IMECs contract) that everything is functional but course tests with Jakob reveal that we are definitely shifting electrode positions etc. @jvoigts when you program your probes with the package I sent you, do you see text appearing in the console window regarding failure of SR checks? |
|
Side note: I changed the SR check reporting level from raising an exception to printing to the Error stream. The reason is because on multishank probes the users should be able to use a probe even if one shank is broken. |
|
I think this is totally reasonable and appropriate, especially given the current state of the integration, we can update this once we get certainty. |
|
Yeah to be honest there are a lot of systematic issues I would like to go over that relate to shared state between devices that use passthrough and the way that gain corrections are handled here, which I think is pretty bad: declare as nullable double, use if sn detected and cast to double just feels gross. |
|
I'll have time to debug this some more next week. I dont think I'm seeing any SR check errors but I have not verified that the channel map is behaving as expected. |
|
@jvoigts When you use the library with np2.0 probes, please let me know so I can verify the SR programming is working for the whole probe |
