Continue to Site

Welcome to EDAboard.com

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.

Why don't ATPGs like clocks generated through flip-flops?

Status
Not open for further replies.

ConradATPG

Member level 1
Joined
Dec 1, 2010
Messages
36
Helped
2
Reputation
4
Reaction score
2
Trophy points
1,288
Activity points
1,509
Why don't ATPGs support clocks that are generated through flip-flops? Is it that the models and simulation engines cannot support it or is there some other reason?
 

You need to specify all derived clocks for ATPG tools to recognize. Some tools such as Cadence - Encounter Test has special engines to recognize derived clocks but these don't always work properly.

Derived clock names, through net names and period need to be defined. Also if there is any interaction between the main clocks and derived clocks, synchronizers need to be placed and defined so that ATPG tools can insert the proper latches.

-- adam
 

I gusss the potential reason is there is no benefit to do so. Using external clocks that external scan enable is in sync with is much simpler solution than making the externally generated scan enable and scan-in data synchnorize to internally generated clock.
 
Last edited:

Couldn't the ATPG find these cases, tell the user and assume that the clocks work?

Walk forwards from all clock inputs specified by the user to find flip-flops clocked from clock pins. Then walk backwards from unclocked flip-flops clock input to find their clock source. You would need to take into account the values on the circuit that select the clock source.

---------- Post added at 22:21 ---------- Previous post was at 22:20 ----------

It may be a simpler solution, but in the real world some designs do this for some reason. It would be good if an ATPG could support it.
 

To ensure a proper shift operation, you would anyway need a single clock. Why not use the same for capture?
 

You don't need to use a single clock to shift the scan chain.

Your scan chain connector could add lockup latches on clock domain boundaries. Alternatively
you could sort the flip-flops so those in each domain are together in the chain and schedule
the clock pulses so those closest to scan out are shifted first. Lastly, you could have your your
tool that constructs the clock trees fix it so shifting will work.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top