MOANA — Wideband power amplifier for magnetogenetic stimulation
A 5 kW amplifier flat from DC to 5 MHz, built on GaN devices, so magnetogenetic stimulation can address many frequency channels at once instead of one at a time.
- Output power
- 5 kW
- Output bandwidth
- DC – 5 MHz
- Devices
- GaN, 5 MHz hard switching
- Channel range
- 5 kHz – 5 MHz
The power amplifier — the GaN output stage delivering kilowatts across a DC-to-5 MHz bandwidth. A collaboration with Rice; the wider MOANA effort is theirs.
Magnetogenetics makes neurons magnetically sensitive, so a magnetic field can drive them without a coil pressed against the head and without the intensity TMS needs. What it asks of the electronics is different, and harder in one specific way: frequency selectivity. Different channels respond at different frequencies, so addressing them independently means producing many frequencies at once, cleanly, at kilowatt power.
That is a bandwidth problem, and bandwidth is exactly what a high-power stage normally has none of.
Why it is a different machine from MPS-TMS
Both MPS-TMS and this are modular synthesizers producing a commanded waveform at high power. The difference is the switch.
MPS-TMS uses SiC devices: enormous current capability, switching rates in the tens to hundreds of kilohertz. That is the right trade when the target is a TMS pulse — a few hundred microseconds, kiloamps into a coil.
Magnetogenetics inverts the requirement. Currents are far smaller, but the output has to stay faithful up to 5 MHz. GaN transistors switch roughly an order of magnitude faster than SiC at the cost of current capability, which is the correct trade here. The prototype hard-switches at 5 MHz and holds output bandwidth from DC to 5 MHz at 5 kW.
Two devices, the same architecture, opposite ends of the current-versus-speed trade. Which is the general point about power electronics for neuroscience: the biology sets the bandwidth, and the bandwidth picks the semiconductor.
Showing the bandwidth
The demonstration I like is encoding the Arecibo reply message into the magnetic field spectrum. It is not a neuroscience result — it is a fidelity claim, made in a form that is hard to fudge. Reproducing an arbitrary structured pattern across the full band means the amplifier is not just reaching 5 MHz but is faithful throughout, which a single-tone measurement would not show.
A more detailed description of the hardware is on IEEE Xplore.
The ideas behind it