Author Topic: Ephys over coax  (Read 2914 times)

clay

  • Newbie
  • *
  • Karma: +0/-0
    • View Profile
Ephys over coax
« on: October 27, 2016, 10:51:39 PM »
I'm thinking of doing some LFP recording while imaging with the miniscope.
To do this, one could either make a simple unit-gain preamp and run a few wires along with the coax cables, but I think it was mentioned in the miniscope workshop that it might be possible to send an additional signal in spare bandwidth on the same coax as the image data.
So I'm basically brainstorming about how this might work, and feel free to jump in if you have ideas.

Components:
1.) unit gain preamp (SMD opamp, or part of dual/quad)
2.) signal amplification (single-supply opamp, with fixed gain?)
3.) and basic filtering? (another channel of quad opamp, e.g.)
3.) AD chip (SMD package)

This would run into the CMOS board MUX, which would synchronize data read/write with the AD chip.

Main questions:
- what is data rate of coax transmission?
- thus how fast could I sample the LFP?
- how to coordinate reading the amplified LFP on the AD chip with MUX chip
- I think the AD chip might need an oscillator- is there one on the CMOS board I can use for this?
- how to DEMUX the LFP on the DAQ board, and take it into acquisition software

Does this make sense? Does anyone else have any ideas about this?

Daniel Aharoni

  • Administrator
  • Full Member
  • *****
  • Karma: +15/-0
    • View Profile
Re: Ephys over coax
« Reply #1 on: October 28, 2016, 01:06:38 AM »
Hey Clay,
One of the projects I have been working on recently is exactly this. While one could make a standalone system to run in parallel with the Miniscope it would be much nicer, as you mentioned, to incorporate it into the current electronics and coax cabling, which is very doable.

I can imagine 2 different approaches to ephys with the Miniscope:
  • Low channel count (1 to 8) and slow sampling (<2KSps) for EEG/LFP.
  • Higher channel count (32+) and faster sampling (30KSps+) for extracellular recordings + LFP

On the current CMOS sensor PCB there are 4 open channels on the serializer to send high speed digital data over the coax cable. The trick is to match the digital output of the ephys electronics with the sampling clock driver the serializer. This sampling clock is generated by the CMOS sensor itself and runs at 26MHz. For low channel, slow sampling ephys setups you could just connect the digitial output to one of the open serializer channels and over sample the digitial signal, reconstructing it once it arrives to the DAQ PCB or computer. For higher count, faster sampling it is not as straightforward but luckily the 32 channel Intan chip runs at almost the exact clock speed as the CMOS sensor. By slowing down the CMOS sensor slightly and using an MCU one could hook it up to the serializer.

To answer your questions:
- what is data rate of coax transmission?
The serialize is currently sampling 12 digital channels at 26MHz. 8 of those channels are being used for imaging data and the other 4 are currently open.
- thus how fast could I sample the LFP?
This is depends on how you want to sync the ephys data to the serializer. Either you would have the ADC output at the sampling rate of the CMOS sensor or much slower.
- how to coordinate reading the amplified LFP on the AD chip with MUX chip
The coax cable carries 4 GPO channels as well as I2C. These max out at ~100KHz so they may be too slow to communicate directly with the ADC. One solution would be to add a small MCU into the circuit and have it communicate over I2C to the DAQ PCB and a much faster communication to the ADC.
- I think the AD chip might need an oscillator- is there one on the CMOS board I can use for this?
There is a 26MHz oscillator on the CMOS PCB.
- how to DEMUX the LFP on the DAQ board, and take it into acquisition software[/li][/list]
There are a couple options here. At slower data rates you could send the ephys data over an open USB endpoint. At higher data rates you could encode the ephys data into the imaging data stream but this would require some tweaks to the UVC implementation.

If you will be at SfN we should touch base and discuss this further.

clay

  • Newbie
  • *
  • Karma: +0/-0
    • View Profile
Re: Ephys over coax
« Reply #2 on: November 01, 2016, 08:05:00 PM »
Yeah, thanks Dan- I figured you guys were working on this. I'm not going to SfN this year but I will think about these problems and communicate over the message board. Not sure if I want to go to the trouble of an Intan chip since I just want LFP for now, but it's cool that the clock speed is similar...