Bench modeSteps, parts, and safety only. Big type for a phone at the bench.
Microvolt
Phase 6: Beyond the scalpProjectA semesterLab-fundedTier 3Needs a lab

Project B: Build a lab a rig

An Open Ephys compatible headstage, a behaviour box, an optogenetics LED driver, a Bpod-style state machine. Labs are full of unbuilt instruments, and building one is how an undergraduate becomes indispensable.

AssumesProject A: The ADS1299 boardFirmware and SPI for a biopotential chipThe research ladderSpineAnalog / mixed-signal hardwareElectrophysiology / wet lab

You are skimming: the title, the first figure, and the short version. Switch to Read in the header for the full page, or Deep to open every deep dive.

Every electrophysiology lab has a list of things it wishes it had: a headstage with a different connector, a box that delivers a stimulus and logs a lever press with millisecond timing, an LED driver with a proper current source, a synchronization signal between two recording systems. Graduate students do not have time to build them and commercial versions cost thousands. A student who has built the ADS1299 board and written its firmware can build any of them, and the lab that gets one will not let that student go. This project is the research ladder’s most powerful rung.

Predict before you look

A lab needs a behaviour box that presents a tone, waits for a nose poke, and delivers a water reward, with all events logged to a millisecond. What is the hardest part?

Timing and synchronization. The components are a speaker, an infrared sensor, and a solenoid valve, an afternoon’s wiring. The value is in a state machine that runs with deterministic timing on a microcontroller, logs every transition with a timestamp, and emits a synchronization pulse the neural recording system records on a spare channel, so that behaviour and neurons can be aligned to the millisecond afterward. That is what the commercial systems (Bpod, Bcontrol) sell, and it is what you build.

Choose from the list

Ask the lab. Then pick the one that is most wanted and that you can finish in a semester. Typical candidates:

A headstage. The Intan RHD chips do for spikes what the ADS1299 does for EEG: 32 or 64 channels of amplification and digitization on a chip, with an SPI interface. Open Ephys publishes headstage designs and an acquisition system that reads them. A headstage variant with a different connector, a different form factor, or an added accelerometer is a well-bounded board project with a real user.

A behaviour box. The state machine above, on a Teensy or RP2040, with a Python interface for defining tasks. Open designs exist (pyControl, Bpod’s firmware); adapting one to the lab’s specific task and hardware is the job.

An optogenetics driver. A precision current source for a high-power LED, with a fast enable input, adjustable current, and an output monitor, so that the light delivered is known and repeatable. Commercial ones cost a thousand dollars; the circuit is an op-amp, a MOSFET, and a sense resistor done carefully, with attention to thermal drift.

Synchronization. A box that takes trigger inputs from several systems (camera, behaviour, recording) and distributes a common timing pulse, with a log. Boring and universally needed.

A jig or a fixture. A 3D-printed holder that positions an electrode repeatably, a perfusion chamber, a head-fixation adapter. Mechanical, quick, and often the thing the lab most wanted.

Do it like an instrument

Requirements from the user (the graduate student who will use it). A schematic and a board, or a drawing. Firmware with a version number. A characterization: timing jitter measured with a scope, current accuracy measured against a meter, noise floor if it records. A one-page manual. A repository. Then hand it over and watch them use it; the first hour of real use produces the change list.

Why this matters more than it looks

The student who builds a lab’s rig is the one whose name goes on the methods section, who is asked to present at the lab meeting, who is trusted with the next project, and whose recommendation letter says “indispensable.” It is also exactly what the research hardware companies (Open Ephys, Intan, Ripple) hire for.

  1. Ask the lab for its list. Pick one. Write half a page of requirements with the user.
  2. Design, build, and bring up, in the Phase 2 style.
  3. Characterize: timing, accuracy, noise, as applicable.
  4. Write the one-page manual and put everything in a repository.
  5. Hand over. Watch the first hour of use. Fix the change list.
  6. Ask for the next item on the list.
Recall
What separates a lab rig that gets used from one that gets shelved?
Requirements written with the actual user, measured characterization (timing, accuracy, noise), a one-page manual, and a repository, then watching the first hour of real use and fixing what it reveals.
Recall
Why is synchronization the hard part of a behaviour box?
Behaviour events must align to the neural recording at millisecond precision, which requires a deterministic state machine, timestamped logging, and a shared timing pulse recorded by both systems.