What is the acheivable gain flatness for RF DACs across the nyquist zone?
The AD9739 and AD9739A, for example, do not have built in inverse-sinc correction.
Across the 1st nyquist zone, gain variation is about 4-5dB. Figure 1 is the best indicator for this (see below). If the signals are in the 2nd nyquist zone (mix-mode), the gain variation is much tighter. In either case, the sinc-roll off can be compensated for in the FPGA to get better response, if needed. Also, the gain response can degrade based on the characteristics of the chosen transformer (shown in figure 2 below) interfaced to the analog outputs of the DAC. We recommend the Mini-circuits TC1-75-33G for the 9739. Our measurements were taken using this transformer.
Figure 1. Gain flatness of the AD9739A across nyquist zones.
Figure 2. Ideal vs. measured (after trasnformer) gain flatness
As per AD9739A evaluation board, RF transformer is 75ohm and connector is 50 ohm, how the impedance is matched in this case?
The 75Ω balun, TC1-33-75G2+, is used on this board intentionally. Since the balun sees a 50Ω load from the DAC side, such mismatch will introduce 14dB of return loss. Compared to 50Ω baluns, the test engineers empirically proved that using 75Ω balun can improve SFDR better. They tried different baluns with the same footprint.
This board is designed to give us the best SFDR performance. However, while it gives the best SFDR performance, it doesn’t help with MER due to the reflected waves. That is, the full power is not delivered to the load and that causes poor MER measurements.
Retrieving data ...