Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
The normal thing we normally do (in most of the companies) if there is a violation of setup and hold in same path, we've to duplicate the logic and separate these two paths. This is most effective method.
please explain how do u seperate a path into two and then do analysis on 2 different paths for setup on one path and hold on other
???????????????????????????????????
It's absolutely right that hold time is frequency independent.
Hold time equation, when there is hold time violation
Thd > Tclk-q + Trout + Tcombo
Hold time fix can be fixed by
a) Adding buffers in the data path after post synthesis
Thd < Tclk-q + Trout + (Tcombo + Tbuffer)
b) At the tape out level by reducing the voltage.
Voltage is inversely proportional to the delay of a design.
Note: We try to avoid fixing hold time at the synthesis level. Though DC has a command set_fix_hold to fix the hold time violation.
How your are missing important point ... if Setup & hold are observed in same path, when you add buffer into the data to provide porpagation delay of 0.8 ns for fixing hold violation, the setup violation will also increased by 0.8ns.
So the effective setup violation will be 1ns after fixing hold violation & you need to increase the clock period by 1ns to fix the effective setup violation (10 + 1= 11ns). (Reduce the clock frequency = 1/11ns)
How your are missing important point ... if Setup & hold are observed in same path, when you add buffer into the data to provide porpagation delay of 0.8 ns for fixing hold violation, the setup violation will also increased by 0.8ns.
So the effective setup violation will be 1ns after fixing hold violation & you need to increase the clock period by 1ns to fix the effective setup violation (10 + 1= 11ns). (Reduce the clock frequency = 1/11ns)
The most important thing to ask the interviewer is - DOES THIS HOLD AND SETUP VIOLATION HAPPEN ON SAME PATH OR DIFFERENT PATHS? IT CHANGES YOUR ANSWER. If same path, clock frequency needs to be 11ns. If different, then first lets fix hold on that path such that it doesnot affect setup here which is done by adding buffers on datapath closer to the pin launching the data.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
