When the system averages a number of samples, the division results in a fractional part of the answer that is lost in producing the integer valued result. Oversampling is a technique where extra samples are summed, but not divided down to produce the average. Suppose the system were oversampling by 16 times – that would mean that the values returned were actually 16 times the average. Using the oversampled value gives additional precision in the returned value.

The number of averaged and oversampled values are always powers of two (number of bits of oversampling/averaging). Therefore the number of oversampled or averaged values is two ^ bits, where ‘bits’ is passed to the methods: SetOversampleBits(bits) and SetAverageBits(bits). The actual rate that values are produced from the analog input channel is reduced by the number of averaged and oversampled values. For example, setting the number of oversampled bits to 4 and the average bits to 2 would reduce the number of delivered samples by 16x and 4x, or 64x total.

The sample rate is fixed per analog I/O module, so all the channels on a given module must sample at the same rate. However, the averaging and oversampling rates can be changed for each channel. The use of some sensors (currently just the Gyro) will set the sample rate to a specific value for the module it is connected to. The example above shows setting the sample rate for a module to the default value of 50,000 samples per channel per second (400kS/s total).

There are a number of options for reading Analog input values from an analog channel:

1. Raw value - The instantaneous raw 12-bit (0-4096) value representing the full -10V to 10V range of the ADC. The typical 0V-5V swing will provide values in the approximate range 0-1024. Note that this method does not take into account the calibration information stored in the module.
2. Voltage - The instantaneous voltage value of the channel. This method takes into account the calibration information stored in the 9201 module to convert the raw value to a voltage.
3. Average Raw value - The raw, unscaled value output from the oversampling and averaging engine. See above for information on the effect of oversampling and averaging and how to set the number of bits for each.
4. Average Voltage - The scaled voltage value output from the oversampling and averaging engine. This method uses the stored calibration information to convert the raw average value into a voltage.
5. Accumulator - The purpose and use of the accumulator is discussed below.

The analog accumulator is a part of the FPGA that acts as an integrator for analog signals, summing the value over time. A common example of where this behavior is desired is for a gyro. A gyro outputs an analog signal corresponding to the rate of rotation, however the measurement commonly desired is heading or total rotational displacement. To get heading from rate, you perform an integration. By performing this operation at the hardware level it can occur much quicker than if you were to attempt to implement it in the robot code. The accumulator can also apply an offset to the analog value before adding it to the accumulator. Returning to the gyro example, most gyros output a voltage of 1/2 of the full scale when not rotating and vary the voltage above and below tat reference to indicate direction of rotation.