Forum Discussion
6 Replies
- Altera_Forum
Honored Contributor
The equivalent of what you did for Xilinx is to use LCELL.
It's not a recomended practice because the exact delay will depend on many factors: process, temperature, voltage and routing. Try to use timing constrains instead. Use set_output_delay -min and set_output_delay -max. Or use set_min_delay and set_max_delay if no clock is involved. - Altera_Forum
Honored Contributor
--- Quote Start --- Try to use timing constrains instead. --- Quote End --- Although it can remove the routing uncertainty, PVT variation will remain. In addition, timing constraints only utilize delays, that can be implemented with regular routing opions. - Altera_Forum
Honored Contributor
--- Quote Start --- Although it can remove the routing uncertainty, PVT variation will remain. In addition, timing constraints only utilize delays, that can be implemented with regular routing opions. --- Quote End --- But, we must use LCELLs, in some cases such as implementing serveral levels delay lines. My question is that what's one Lcell's delay time value when given a device? Or could you tell me how and where to find this value? And what the curve diagarm relationship between this delay value and PVT variations? - Altera_Forum
Honored Contributor
--- Quote Start --- Hi, I need a delayed outputs I don't want to use LCELL as suggested by Forum members as it is bad practice. --- Quote End --- Perhaps you should first explain why you think you need a delayed output since, in many cases, that's not a good practice either regardless of the implementation. Kevin Jennings - Altera_Forum
Honored Contributor
You can evaluate the expectable logic cell delays in timing analysis. 0.3 to 0.4 ns per logic cell is a rough estimation for Cyclone FPGAs.
- Altera_Forum
Honored Contributor
--- Quote Start --- You can evaluate the expectable logic cell delays in timing analysis. 0.3 to 0.4 ns per logic cell is a rough estimation for Cyclone FPGAs. --- Quote End --- Thanks for you reply!