Skip to content

Gain_settings

Nick Steinmetz edited this page Sep 4, 2018 · 2 revisions

The amplifier gains may be altered programmatically, from a range of settings. This allows you to trade off dynamic range versus resolution. A low gain setting will have less voltage resolution but can handle a wider range of voltages without saturating (the same encoding bits are stretched to cover the wider range), and the reverse for high gain. The table below shows the possible gain settings and the resolution/range of each.

A typical setting is 125 or 250 for LFP band, and 500 for AP band.

Gain setting Resolution (µV/bit) Range (mV)
50 23.4 -12 to 12
125 9.38 -4.8 to 4.8
250 4.69 -2.4 to 2.4
500 2.34 -1.2 to 1.2
1000 1.17 -0.6 to 0.6
1500 0.78 -0.4 to 0.4
2000 0.59 -0.3 to 0.3
2500 0.47 -0.24 to 0.24

How these numbers are calculated: the ADCs are 10 bits, and their voltage range is -0.6 to 0.6V, giving them a resolution of 1.2mV/bit (=1.2V/2^10bits). So for gain=500, for example, the true values are multiplied by 500 before being digitized according to that resolution and range, yielding a resolution of 2.34µV/bit (=1.2mV/bit / 500) and a range of -1.2 to 1.2mV (=-0.6V/500 to 0.6V/500).

Important note! The Phase3A probes (which version do I have?) produce data that is offset from zero, due to this behavior not being optimized (will be improved in a future version of the probe). The amount of offset is variable from channel to channel, and depends on the gain setting. This effectively reduces the dynamic range in one direction and increases it in the other. Unfortunately the direction of offset is negative, so the dynamic range is smaller in that direction. In other words, with a gain of 500 the AP band dynamic range may actually be something more like -0.9 to 1.5mV.

Amplifier saturations recover quickly in the AP band (~2ms) but obviously should still be avoided wherever possible.

Another important note! There is a “gain correction” that should be applied to record the most accurate data. It applies a multiplicative factor to each individual channel to optimally match its true gain with the nominal value. This correction is optional in both acquisition softwares, because it takes a long time (~5 minutes) to perform (for Phase3A; in Phase3B (which version do I have?) it is fast but the files must be supplied), each time you connect the probe. However, for real data you should always apply it. See the IMEC-supplied user manual for details about the precision of the gain values with and without correction.

You can’t perform that action at this time.