The 9103 Picoammeter uses a continuously sampling A/D when measuring current. These samples are then averaged using an low-pass infinite impulse response (IIR) filter.
When using the 9103 to sample current, you have control over the filter response and the degree of smoothing (both in Actuel in when programming the unit). The filter setting will make little difference for most constant signals, but for dynamic and periodic signals, the filter can be set to attenuate noise, or to provide detail and catch peaks.
A filter coefficient that is user-programmable determines the amount of smoothing the filter will apply. The higher the value, the more smoothing of the signal.
The filter can be set to 0, 2, 4, 8, 16, 32, and 64. A value of 0 is essentially the same as bypassing the filter. A value of 64 is the greatest amount of filtering. For most cases, values of 4, 8, and 16 will work best. Higher values may produce more accurate results for stable signals, but it will take longer for measurements to stabilize.
In the examples below, a 1 Hz sine wave is sampled at a 25 mS rate, yielding 40 discrete data points per cycle. Each data point is comprised of multiple filtered A/D readings.
The filter settings used in the examples are 2, 4, 8, 16, and 32.
There’s quite a bit of noise present when using low filter values of 2 and 4, while values of 16 and 32 reduce the noise but also attenuate the signal somewhat. For this application, a value of 8 produces the most accurate result.
In general, any low-pass filter will of course mask high-frequency data. While the 9103 is not typically used to measure periodic signals, the filter’s effect on your application may be significant. When in doubt, start with the filter set to 8 for some noise reduction without significant smoothing or signal attenuation.
In Part 2 we’ll discuss use of the additional first-level filter implemented in the high-speed model of the 9103.