ContentsPrevious Chapter Next Chapter

5 Propagation Delay Effects

5.1 Hazards


Using a simple example circuit, the reason for the occurrence of time-limited errors in digital circuits can be demonstrated:

Figure 5.1: Sample Circuit


Figure 5.2: K-Map ref. to the circuit of Fig. 5.1.


In a steady-state situation this circuits works correctly. But looking at the timing behaviour for the input signal change that has also been indicated in the K-Map

x1: '1' '0'

with constant values x0 = x2 = '1', then a short signal change at output y0 can be observed.

The timing diagram clearly shows that due to the inversion of signal 2 the signal 4 will be delayed by one gate delay compared to signal 5. Therefore at the input of the OR gate for a short time the unexpected case will occur, in which both input signals 4 and 5 are '0'.

Consequently output y0 can be switched to logic '0' for the time corresponding to one gate propagation delay (in the following timing diagram a uniform delay time of 10 ns is assumed).

Figure 5.3: Timing Diagram.

The appearance of this spurious pulse (glitch) at the OR-output (signal 6) cannot be precisely predicted. Gates may "absorb" very short signals because of their internal capacities (see also: dynamic noise immunity).

Because of the circuit design (in this example) the danger of a delay-dependent erratic behaviour exists. This risk that may exist in a circuit realization is called a "Hazard".

Definition:

When a circuit structure allows during an arbitrary input signal change the appearance of a spurious pulse, then the circuit realization contains a Hazard. This pulse that occurs due to the hazard is called a glitch or spike.

Example:

The following example will show how decisive a hazard can be for the function of a circuit:

A combinational circuit will be used to produce the signals that are needed to control an RS Flip-Flop. The RS FF shall be maintained in the "Store" state (R=S='0' in NOR realization), but a glitch (hazard) occurs on the R line:

Figure 5.4: Hazard-caused erratic behaviour of a flip-flop.

The RS Flip-Flop indeed remains in the "Store"state, but the glitch generates an unwanted reset of the Q output.

The sources of these problems are erroneous circuit structures, their localization can be quite difficult. Very often a complete redesign of the circuit is necessary for the removal of these hazards.


ContentsPrevious Chapter Next Chapter