To deliver water rewards to mice, we are using Sanworks port assemblies attached in the bottom of our freely moving behavior box. We have recently started experimenting with animals that are implanted with Neuropixels probes, and we noted a significant noise problem related to the ports.
Every time an implanted mouse licks the metal spout, we get electrical artifact in our recording that are high-amplitude and sometimes completely mask neural spiking. What's more interesting is that some lower amplitude noise remains after the mouse stops licking the spout and moves away, but this lower amplitude noise slowly decays (in 5-10 sec). Having tested a dozen of different variations, we have concluded that much of the noise is coming in through the water.
That's why the only thing that has reduced artifact amplitude was actually grounding the outer part of the metal spout to the GND connection of the port board. Replacing the metal spout with a plastic or glass one didn't actually do much, as the water couldn't be grounded.
Does anyone have similar problems or suggestions on possible solutions?