I'd like to provide an update on the development of the recording system. It's been a long summer, and
I've managed to four essential pieces working:
- Headstage prototype PCB manufactured, assembled, and reflowed in my toaster oven.
- Code has been developed to boot the on-board DSP, control the ADCs and multiplexer,
filter the data (4th order elliptic, passband 400 Hz - 6 Khz),
drive a DAC (for DSP code testing), break the data into packets, and drive the radio. Everything runs entirely in cache.
- USB and parallel port headstage programmer created and verified
- The parallel port is much quicker
than the Activewire-inc. based USB programmer, hence will be shipped with product if reprogramming is a desired feature.
- Headstage radio link functional, protocol development ongoing.
- 2.4-2.5ghz, 2mbps over-the-air rate, +1dbm output -> ~5 meter range, power-saving modes.
- Utilizes a nordic semiconductor radio, not 802.11 (too much power) or Bluetooth (restrictive device licensing, expensive, closed-source / non-linux development)
- Will permit the transmission of up to 5 continuous signals downsampled to 31.25 khz,
or 32 waveform/spike channels @ maximum aggregate spike rate of 180 Hz, each waveform also sampled at 31.25 Khz
- Radio to 100-baseT ethernet bridge developed, tested, functional. Again, protocol development is ongoing.
- Ethernet was chosen because it makes supporting multiple platforms (Linux, Mac, & Windows, and all flavors thereof) much easier than, say, USB.
- Ethernet driver, ARP, UDP, & DHCP code prototyped, functional. Much of this is based on
- Will develop drivers for Zeroconf/Avahi/Bonjour & IPv4 link-local,
to allow customers to use the devices without configuration.
- Presently based on a commercial development board, to be replaced with a small-form custom board.
- Client program prototyped using wxWidgets, a cross-platform GUI library.
Now, back to figuring out why DMA is not working ;)
Tim Hanson, October 9 2007,
Neurorecord makes headstages for extracellular recording of neural activity,
specifically spikes/action potentials. At present the headstages are in
active development; however, the design has the following feature targets:
- 32 channels
- 62.5Khz sampling @ 12 bits resolution (oversample to improve the SNR)
- Aggregate gain of 3,000
- Passband 200-20Khz
- Yes, I know, this precludes measuring LFP, EEG and ECoG; I'll tackle these in a second version.
- 60mw power dissipation (a cautious figure; hopefully I can make this lower)
- 20mm x 17mm size (this does not include a battery)
- Several interface options:
- 7 wire interface, 5 of which can be shared among all the headstages of an animal
- Bluetooth SCO connection (aka headset profile); this is low power but
requires substantial compression of the multichannel waveform data.
- 802.11 b/g (aka wifi). This feature is pending review & further examination;
don't hold your breath, but if I can get it to work it will be awesome!
I will keep you updated as the design progresses. I'm quite motivated to proceed,
because not only do i see a strong market demand for something cheaper, smaller, and simpler
than the products of Plexon,
Neuralynx, but also I really believe that we need
better recording systems to advance science. While there has been much research effort put into
investigating new methods & electronics for recording, too few people are actually selling the
Though we may not achieve the same level of performance as the best research recording systems,
our goal is to have headstages that are practical and available.
Tim Hanson, May 22 2007.
contact me at 919-475-4340 or