7.3.6.3.2 Automatic SYSREF Calibration

The ADC12DJ3200 has an automatic SYSREF calibration feature to alleviate the often challenging setup and hold times associated with capturing SYSREF for giga-sample data converters. Automatic SYSREF calibration uses the t_AD adjust feature to shift the device clock to maximize the SYSREF setup and hold times or to align the sampling instance based on the SYSREF rising edge.

The ADC12DJ3200 must have a proper device clock applied and be programmed for normal operation before starting the automatic SYSREF calibration. When ready to initiate automatic SYSREF calibration, a continuous SYSREF signal must be applied. SYSREF must be a continuous (periodic) signal when using the automatic SYSREF calibration. Start the calibration process by setting SRC_EN high in the SYSREF calibration enable register after configuring the automatic SYSREF calibration using the SRC_CFG register. Upon setting SRC_EN high, the ADC12DJ3200 searches for the optimal t_AD adjust setting until the device clock falling edge is internally aligned to the SYSREF rising edge. TAD_DONE in the SYSREF calibration status register can be monitored to ensure that the SYSREF calibration has finished. By aligning the device clock falling edge with the SYSREF rising edge, automatic SYSREF calibration maximizes the internal SYSREF setup and hold times relative to the device clock and also sets the sampling instant based on the SYSREF rising edge. After the automatic SYSREF calibration finishes, the rest of the startup procedure can be performed to finish bringing up the system.

For multi-device synchronization, the SYSREF rising edge timing must be matched at all devices and therefore trace lengths must be matched from a common SYSREF source to each ADC12DJ3200. Any skew between the SYSREF rising edge at each device results in additional error in the sampling instance between devices, however repeatable deterministic latency from system startup to startup through each device must still be achieved. No other design requirements are needed in order to achieve multi-device synchronization as long as a proper elastic buffer release point is chosen in the JESD2048 receiver.

Figure 66 provides a timing diagram of the SYSREF calibration procedure. The optimized setup and hold times are shown as t_SU(OPT) and t_H(OPT), respectively. Device clock and SYSREF are referred to as internal in this diagram because the phase of the internal signals are aligned within the device and not to the external (applied) phase of the device clock or SYSREF.

Figure 66. SYSREF Calibration Timing Diagram

When finished, the t_AD adjust setting found by the automatic SYSREF calibration can be read from SRC_TAD in the SYSREF calibration status register. After calibration, the system continues to use the calibrated t_AD adjust setting for operation until the system is powered down. However, if desired, the user can then disable the SYSREF calibration and fine-tune the t_AD adjust setting according to the systems needs. Alternatively, the use of the automatic SYSREF calibration can be done at product test (or periodic recalibration) of the optimal t_AD adjust setting for each system. This value can be stored and written to the TAD register (TAD_INV, TAD_COARSE, and TAD_FINE) upon system startup.

Do not run the SYSREF calibration when the ADC calibration (foreground or background) is running. If background calibration is the desired use case, disable the background calibration when the SYSREF calibration is used, then reenable the background calibration after TAD_DONE goes high. SYSREF_SEL in the clock control register 0 must be set to 0 when using SYSREF calibration.

SYSREF calibration searches the TAD_COARSE delays using both noninverted (TAD_INV = 0) and inverted clock polarity (TAD_INV = 1) to minimize the required TAD_COARSE setting in order to minimize loss on the clock path to reduce aperture jitter (t_AJ).