Troubleshoot Electrical Noise¶
- create protocol with clamp device
- configure daq to run at maximum rate possible (500kHz for MIO-6259)
- set protocol to loop and start
- switch plot to FFT mode
- turn off and unplug everything, turn back on one at a time.
- check for noise first with a model cell on the headstage, wrapped in foil
- next, remove the foil and check again
- remove the model cell, replace with an ACSF-loaded pipette, check again
- place the pipette in the recording chamber with ACSF flowing
- experiment with different grounding/shielding conditions
- experiment with different sampling rates, downsampling, and possibly lowpass filters (we commonly sample at 400kHz and downsample 10x to 40kHz)
Common noise sources¶
- some MultiClamps produce a wide band of HF noise that is not attenuated by the bessel filter. The best way around this is to oversample (>300kHz) and then downsample.
- most cameras produce noise. The noise may vary depending on the state of the camera (initialized/not, acquiring/not, frame rate...). Camera noise can be addressed by making sure the scope is grounded, and the firewire connection uses a ‘quality’ interface.
- fluid management (pumps, aspirators, drips) is often the cause of random spiky noise
- temperature probes in the recording chamber
- brush motor pumps
- any cables running parallel to the headstage cables
- cell phones. just turn ‘em off.
- 60Hz. It’s everywhere! Most important is to ground your microscope. Make sure the headstage has good contact with electrodes and that both electrodes are in good condition. 60Hx noise is often radiated directly to the electrode by the experimenter (your body makes an efficient antenna). To avoid this, either move all noise sources (power supplies, laptops, noisy screens/keyboards) away from the rig and experimenter or put a grounded metal screen between the preparation and experimenter.
In general, noise can be reduced significantly by oversampling and then downsampling. See the DAQ interface in the prorocol runner [link].