At Starfish, we value collaboration and sharing. We’re constantly evaluating how we can best contribute to the world, be it through our own efforts or by enabling others to innovate in meaningful ways. While we develop technologies initially for our internal goals, we expect these technologies may also open doors in areas and ways we haven’t imagined. We are therefore excited to share info about our upcoming chip for miniaturized ultra-low power electrophysiology, and want to hear from you if and how this could be useful to you.
Ultra-low power, miniature electrophysiological electronics: our upcoming custom chip
Distributed neural interfaces hold great promise for future therapies. Existing approaches to interfacing with the brain predominantly focus on interacting with a single brain region (for example, deep brain stimulation for Parkinson’s disease). However, there is increasing evidence that a number of neurological disorders involve circuit-level dysfunction, in which the interactions between brain regions may be misregulated. Developing better therapies for these disorders will require distributed neural interfaces capable of interacting with the brain at the circuit level; that is, reading and writing to multiple connected parts of the brain at once. Such interfaces do not yet exist clinically, and existing interfaces are not straightforward to parallelize because of their bulky physical footprints, power and communication bandwidth needs, and surgical burden.
We believe there is an opportunity to develop a new class of minimally-invasive, distributed neural interfaces that enable simultaneous access to multiple brain regions. We’re working to build new technologies that allow for recording and stimulation of neural activity with a level of precision vastly exceeding what is possible with currently available systems. A key aspect of this is reducing the surgical burden of device implantation – in part by reducing implant size. We’re doing this in two primary ways:
- By minimizing the physical size of our electronics, leads, and packaging, and
- By removing the battery and reducing power requirements such that the implant can easily run via wireless power transmission.
Commercially available electrophysiology electronics aren’t a good match for these approaches, as they tend to be power-hungry (10s of mW) and physically large (typically >5 mm/side, often >10mm.). We decided to address this gap by developing our own custom electrophysiology chip, in partnership with imec, an R&D organization with an exceptional track record of pushing the envelope on state-of-the-art neural interfaces, particularly in miniaturization and low power consumption.
Our overall goals for this chip center on minimal size and low power, while maintaining functionality similar to existing general-purpose headstage designs capable of both recording (spikes and LFP) and stimulating. Some of the features include:
- Low power: 1.1 mW total power consumption during normal recording
- Physically small: 2 x 4mm (0.3mm pitch BGA)
- Capable of both recording (spikes and LFP) & stimulation (biphasic pulses)
- 32 electrode sites, 16 simultaneous recording channels at 18.75kHz
- 1 current source for stimulating on arbitrary pairs of electrodes
- Onboard impedance monitoring and stim voltage transient measurement
- Digital onboard data processing and spike detection allows the device to operate via low-bandwidth wireless interfaces.
- Fabricated in TSMC 55nm process
We designed this chip with the intent of future integration into a fully wireless, battery-free implant. Therefore it has flexibility to work under various power or data constraints – for example, amplifiers can be disabled to save power; data can be filtered and downsampled, and channels can be omitted to save bandwidth, and communications interfaces are flexible enough to support easy integration with most microcontrollers. At the same time, we are also actively developing tiny, low-power electronics for reliably and robustly transmitting power and data through tissue, and expect to have more to share on that progress in the future.
We anticipate our first chips arriving in late 2025 and we are interested in finding collaborators for whom such a chip would open new and exciting avenues. At this early stage, we’re especially interested in collaborators for whom this technology would pair well with their existing work in fields such as wireless power delivery and communication, or those designing custom implanted neural interfaces. But we want to hear from you regardless if this chip might be useful in those areas, or somewhere completely different. If you’re interested in using these chips or just want to learn more, contact us at ochre-inquiries@starfishneuro.com and let us know how they might help you—or if you have any other questions or thoughts.
Additionally, there is tons of work yet to be done here, and we are always looking for exceptional people to join our team and help us accelerate our vision. If this vision sounds exciting to you, please click on Join Us, and consider applying to work with us.
Posted May 20, 2025 by Nate Cermak