Output delays are similar to input delays, except that they refer to the output path minimum and maximum time outside the device to be functional under all conditions.
The output delay values for the both types of analysis are:
Output Delay(max) = Tsetup + Ddata(max) + Dclock_to_FPGA(max) - Dclock_to_ExtDev(min)
Output Delay(min) = Ddata(min) - Thold + Dclock_to_FPGA(min) - Dclock_to_ExtDev(max)
The following figure shows a simple example of output delay constraints for both
setup (max) and hold (min) analysis, assuming the
clock has already been defined on the
set_output_delay -max -clock sysClk 2.4 [get_ports DOUT]
set_output_delay -min -clock sysClk -1.1 [get_ports DOUT]
The output delay corresponds to the delay on the board before the capture edge. For a regular system synchronous interface where the clock and data board traces are balanced, the setup time of the destination device defines the output delay value for max analysis. And the destination device hold time defines the output delay for min analysis. The specified min output delay indicates the minimum delay that the signal will incur after coming out of the design, before it will be used for hold analysis at the destination device interface. Thus, the delay inside the block can be that much smaller. A positive value for min output delay means that the signal can have negative delay inside the design. This is why min output delay is often negative. For example, the following code example indicates that the delay inside the design until DOUT has to be at least +0.5 ns to meet the hold time requirement.
set_output_delay -min -0.5 -clock CLK [get_ports DOUT]