Python is a registered trademark of Python Software Foundation.
All trademarks are the property of their respective owners.
The AFE7070 device integrates a 12-bit, 65-MSPS dual digital-to-analog converter (DAC), baseband low-pass filters, and a quadrature modulator. The AFE7070 is a convenient low power digital-to-RF conversion device suitable for a variety of transmitter applications from 100 MHz to 2.7 GHz. The integrated baseband filters have a bandwidth of 10 MHz which support RF signals bandwidths up to 20 MHz. The DAC incorporates a Numerically Controlled Oscillator (NCO) which moves the complex signal within the 20 MHz RF band window. The DAC also incorporates digital quadrature modulator correction (QMC) to suppress the spurious performance of the modulator.
The integrated quadrature modulator behaves similarly to its discrete counterparts. In an ideal modulator, the carrier (that is, LO) and the image frequency are completely suppressed. In a real modulator, DC offset imbalances within the baseband inputs manifest as carrier feedthrough (CF). Amplitude and phase imbalance in the baseband inputs degrade the sideband suppression (SBS). The QMC correction in the DAC adjusts for the modulator imperfections to suppress those components to the noise floor; however, the corrections only hold for a single frequency point and for a single set of conditions. Once frequency, LO power, or temperature change, the corrections are no longer optimized and the carrier feedthrough and sideband suppression degrades.
It is impractical to calibrate and store optimized QMC values across all frequencies and conditions. The goal is it achieve the best suppression across all conditions with three room-temperature calibrations points: the low, mid, and high frequency within the band of interest. A simple algorithm approach modifies the QMC calibration points based on current operating conditions.
The use case focuses on operation in the VHF band with the following specifications:
Initially it is important to understand the inherent performance of the device within the band of interest with respect to variation of the following parameters:
Although this use case focuses on the VHF band, the techniques and approach are suitable for any band.
This section describes CF and SBS characterization.
The CF and SBS performance are known to vary over frequency. Figure 2-1 and Figure 2-2 show the characterization data of CF and SBS performance over frequency for an unadjusted (that is, uncalibrated) case and the adjusted case where each point is optimized.
The carrier feedthrough performance at the lower frequencies is inherently good. Unadjusted performance hovers around –55 dBm. With optimization, the CF performance improves to below –80 dBm. Conversely, the sideband suppression starts off at a higher level. This is understandable knowing the polyphase circuit of the device that generates the quadrature signals of the LO is operating at the edge of its designed range. It is expected to have more phase imbalance as the frequency drops. SBS performance is around –40 dBc at the high end of the band and degrades to –33 dBc at the low end of the band.
Figure 2-3 and Figure 2-4 show the optimized QMC parameters across frequency for one device. The absolute value is not important as those values are different for each device. The key observation is the slope and variability of the parameter over frequency; this indicates how well the optimization algorithm interpolates between known calibration points.
The DCI/Q variation is not excessive. This is anticipated since the CF performance remains fairly consistent. The QMC-Gain is nominally at 1024. For this particular device there is a consistent gain adjustment of 2 steps from nominal with an occasional adjustment by one step higher or lower. The QMC-Phase parameter exhibits the biggest change. Technically, the curve fit of the response is a second-order polynomial, but if the response is broken into two pieces, corresponding to low and high band, each section is approximately linear with a slightly different slope. Using linear interpolation between three frequency points yields a good approximation of the optimized set-point across frequency. Similarly, linearly interpolated values between calibration points is used to determine the intermediate points for the DCI/Q and QMC-Gain. Figure 2-5 and Figure 2-6 show the corrected CF and SBS performance over frequency using just 3 calibrated points.