Calibrations Explained

You can find the calibrators we used here.

The goal of this calibration is to understand how the Roland Octa Capture amplifies the signals into its channels before reading them into the computer, with the goal of being able to measure the physical input in its real units (SPL, force, acceleration, or velocity).

The model we have come up with assumes a linear slope for each level of gain with a zero y-intercept (0 volts in reads 0 arbitrary units in). This slope should converge to one number for all gains if the arbitrary units input is scaled exponentially by gain.

To test this model, we sent sine waves of known amplitude from a function generator into the octa capture at different gain levels. We first plotted the amplitude of the arbitrary units ready by the computer vs the voltage input for each gain level, then scaled the arbitrary unit exponentially by gain so that the lines converged and solved for "m" in the equation above.

The plot above shows the different best fit slopes for each constant gain level, and how the calibration slope increases for higher gain levels. More testing needs to be done to determine the stability of the amplifier over time, but if you calibrate the octa capture at a certain gain level, but if a channel is calibrated to same gain level that data is recorded, you should get an accurate reading out. If you calibrate to a different gain level from what you measured, the maximum error from calibration will be on the order of 1dB.

Each line on the plots above represents the amplitude in SPL from the microphone calibrator at a set gain calculated using the calibration coefficients (m in the equation above) determined from the function generator for different gain levels. The zero point on the x-axis represents the calibration coefficient at the same gain used to record the microphone calibrator, which should theoretically return the expected amplitude of the calibrator. However, because the calibration was performed with the function generator, there is a steady offset from the expected amplitude. If you trust your microphone calibrator, the coefficient should be calculated using that, then the gain level should be kept constant.