1 EE 365 Combinational-Circuit Synthesis

2 Combinational-Circuit Analysis Combinational circuits -- outputs depend only on current inputs (not on history). Kinds of combinational analysis: –exhaustive (truth table) –algebraic (expressions) –simulation / test bench Write functional description in HDL Define test conditions / test vectors, including corner cases Compare circuit output with functional description (or known-good realization) Repeat for “random” test vectors

3 Combinational-Circuit Design Sometimes you can write an equation or equations directly using “logic” (the kind in your brain). Example (alarm circuit): Corresponding circuit:

4 Alarm-circuit transformation Sum-of-products form –Useful for programmable logic devices (next lec.) “Multiply out”:

5 Sum-of-products form AND-OR NAND-NAND

6 Product-of-sums form OR-AND NOR-NOR

7 Brute-force design Truth table --> canonical sum (sum of minterms) Example: prime-number detector –4-bit input, N 3 N 2 N 1 N 0 row N 3 N 2 N 1 N 0 F F =   (1,2,3,5,7,11,13)

8 Minterm list --> canonical sum

9 Algebraic simplification Theorem T8, Reduce number of gates and gate inputs

10 Resulting circuit

11 Visualizing T10 -- Karnaugh maps

12 3-variable Karnaugh map

13 Example: F =  (1,2,5,7)

14 Karnaugh-map usage Plot 1s corresponding to minterms of function. Circle largest possible rectangular sets of 1s. –# of 1s in set must be power of 2 –OK to cross edges Read off product terms, one per circled set. –Variable is 1 ==> include variable –Variable is 0 ==> include complement of variable –Variable is both 0 and 1 ==> variable not included Circled sets and corresponding product terms are called “prime implicants” Minimum number of gates and gate inputs

15 Prime-number detector (again)

16 When we solved algebraically, we missed one simplification -- the circuit below has three less gate inputs.

17 Another example

18 Yet another example Distinguished 1 cells Essential prime implicants

19 Quine-McCluskey algorithm This process can be made into a program, using appropriate algorithms and data structures. –Guaranteed to find “minimal” solution Required computation has exponential complexity (run time and storage)-- works well for functions with up to 8-12 variables, but quickly blows up for larger problems. Heuristic programs (e.g., Espresso) used for larger problems, usually give minimal results.

20 Lots of possibilities Can follow a “dual” procedure to find minimal products of sums (OR-AND realization) Can modify procedure to handle don’t-care input combinations. Can draw Karnaugh maps with up to six variables.

21 Real-World Logic Design Lots more than 6 inputs -- can’t use Karnaugh maps Design correctness more important than gate minimization –Use “higher-level language” to specify logic operations Use programs to manipulate logic expressions and minimize logic. PALASM, ABEL, CUPL -- developed for PLDs VHDL, Verilog -- developed for ASICs