Logic Design for ArrayBased Circuitsby Donnamaie E. WhiteCopyright © 1996, 2001, 2002, 2008, 2016 Donnamaie E. White, WhitePubs Enterprises, Inc. 


External SetUp and Hold TimesLast Edit July 22, 2001 Computing Hold Time in the Register ExampleTo compute the hold time of the design, the propagation delay of both paths needs to be determined while factoring in the effects of tracking.
The general equation for the example is as follows (using 10% tracking for the first macro in PATH1, and 20% tracking for the second macro):
where Tpd is nominal. For the example, the driver macros (D1) are identical and placed within the same chip quadrant. For the Q5000 Series, this has been specified to mean a 10% tracking (0.9). For identical macros on the same row in the same quadrant, this reduces to 5% tracking or 0.95. TRK1 will be taken at 10%. The second tracking factor (TRK2) applies to FF15. This accounts for unlike edges and unlike structures between D2 and FF15. It does assume that both are placed within the same quadrant. For macros that are not identical, placed in the same quadrant on the array, tracking is defined as 20% or 0.8. TRK2 will be taken as 20%. These numbers only apply to the AMCC Q5000 Series. The timing diagram for Figure 65 is shown in Figure 67. Minimum Library (Q5000 Series)For the example, for a military circuit, using bipolar worstcase multipliers (AMCC Q5000 Series), the equation for the Minimum Library minimum values becomes: 0.86 * [0.9 * Tpd_{ D1} + 0.8 * Tpd_{ FF15}  Tpd_{ D2}] > Th_{ FF16} The Minimum Library is specified with worstcase multipliers of 0.86 (MAX), 0.78 (TYP) and 0.70 (MIN). Figure 67 Timing Diagrams In Figure 67, FF15 Tpd C>Q (clock to output delay) may be 20% faster than nominal. The total effect of these two variations may be to cause the D input to FF16 to change within the hold time window. Military Library (Q5000 Series)The equation for the Military Library maximum value becomes: 1.45 * [0.9 * Tpd_{ D1} + 0.8 * Tpd_{ FF15}  Tpd_{ D2}] > Th_{ FF16} The Military Library is specified with worstcase multipliers of 1.45 (MAX), 1.32 (TYP) and 1.19 (MIN). The computation should be performed using the MAXIMUM (Military or Commercial, as appropriate) worstcase library and then using the MINIMUM library to determine the worstcase external hold time. The ratio of the data and clock path delays will determine which library will produce the larger (most positive number) hold time. While success can be estimated using FrontAnnotation, final calculations must be made after BackAnnotation using the actual metal delays. The layout and/or design must be changed if the appropriate test fails. Note that tracking may be part of the adjustment factor specifications, in which case the equations would need to be altered to reflect this different specification. Check with the array vendor.
Exercises


Copyright © 1996, 2001, 2002, 2008, 2016 Donnamaie E. White , WhitePubs
Enterprises, Inc. 