Adjustment Interface - 2.3 English

After a value has been established in the system_timer, it begins to increment by a fixed amount on a cycle-to-cycle basis. This fixed amount is labeled as timer_increment on the block diagram.

The amount of the increment can be adjusted in one of three methods. The desired method is selected using the signal ctl_D_ptp_st_adjust_type_N input. The update value is provided by D_ptp_st_adjust_N and the operation is triggered by toggling the value of D_ptp_st_adjust_vld_N.

Updates can be performed at any time, although changes should only be performed once every 10 clocks (system clocks) to avoid issues with retiming.

Method 1: Phase Adjustment (type = 2’b00)

The system_timer can be shifted in time by a specific amount (that is, a phase shift) using the system timer phase adjustment method. The provided adjustment is signed and is in units of 2^-8 ns which are directly added to system_timer in addition to the usual timer increment. It is a one-time adjustment and you must repeat the operation again to add another amount. The maximum shift is (+/- 2¹⁷ - 1) × 2^-8 ns, or approximately +/- 0.5 µs.

Method 2: Coarse Frequency Set (type = 2’b01)

In this mode, you can define a new increment value that is to be added to system_timer on a clock-by-clock basis. The provided increment value is interpreted as an unsigned integer in units of 2^-8 ns (maximum 10 bits). When a coarse adjustment is made, the bottom 32 bits of the increment value are zeroed out (the internal increment value has a granularity of 2^-40 ns). To make more fine-grained adjustments (< 2^-8 ns) to increment_value, the next method must be used.

Method 3: Fine Frequency Adjust (type = 2’b10)

This method allows you to make sub-nanosecond adjustments to the interval value. The provided increment value is interpreted as a signed integer in units of 2^-40 ns. It is added to the existing timer increment. This is a signed value, so subtraction is possible if a negative adjustment is provided.