If your analysis software (Alliance, Brainlab, Winsor and others) performs a software bipolar subtraction of two amplifiers that both block, the result is a zero-trace. So the problem is most probably occurring when frequent blocking / saturation of amplifiers occurs.This happens especially during movement (also the ECG and EOG show high variability), but also during quiet, deep breathing where the mechanical breathing movements induce LF movement artifacts on the cables; this latter effect is larger if regular cables are used and if cables are not properly strain-relieved.

Possible causes and solutions:

Amplifier range set to 500 micro Volt

The 500 micro Volt range, in combination with a rather long time-constant of one (1) second, is only suggested for ?stationary? sleep studies, and then only if our special low-noise cables are utilized that reduce LF movement artifacts a factor 10. Even for these ?stationary? in-bed situation, a proper strain-relief of the cables in the neck is absolutely mandatory to prevent breathing artifacts.

Change the range of the amplifiers to 1000 micro Volt (amplification 2500 in stead of 5000).
Use low-noise cables.
Place the reference EEG electrode very securely at a place that is rather insensitive for artifact and spurious amplitudes, for instance at Cz.
Be sure the EEG cables are securely strain-relieved at the neck and alongside the body.
Use a TC of 0.3 for awake EEG and .7 for sleep EEG during review.
Check for dirt on the multiconnector of the patchpanel and the multiconnector of the recorder. Clean the contact with alcohol.

The reference electrode is not properly attached to the patient or to the recorder.

When the reference signal is lost all signals will have about the same signal recorded. Subtracting 2 of these signals will result in a zero-trace. Probably only the ECG signal will be visible because this signal is much higher in value.

Place the reference EEG electrode very securely at a place that is rather insensitive for artifact and spurious amplitudes, for instance at Cz.
Be sure that that multi-cable-connector is properly attached to the module to prevent loose contact.
Check for dirt on the multiconnector of the patchpanel and the multiconnector of the recorder. Clean the contact with alcohol.

In general you should first check the reference electrode (is the electrode OK and is it properly attached. Also, for long term recordings, a bio-check (checking the signals with the on-line viewing option of the Columbus software) or at least an impedance check of the electrodes (for instance using the OhmMate impedance meter), should be performed.