DocumentCode
2525080
Title
Gate delay modeling for static timing analysis of body-biased circuits
Author
Baek, Donkyu ; Shin, Insup ; Shin, Youngsoo
Author_Institution
Dept. of Electr. Eng., KAIST, Daejeon, South Korea
fYear
2012
fDate
May 30 2012-June 1 2012
Firstpage
1
Lastpage
4
Abstract
Body biasing is a well-known circuit technique to compensate for increasing process variations. A static body biasing, in which fixed amount of bias voltage is applied after chips are sorted out according to process corners, is a convenient and viable approach in ASIC designs. The key question in this approach is how we generate new gate delays without re-characterizing all the gates, which is expensive. We notice, for the first time, that the ratio of delay (of original gate and the gate after body biasing) from the time input starts to change to the time output crosses Vdd=2 can be regarded invariant, i.e. it is a function of input transition time and load capacitance, but not a function of gate type. Therefore, once the ratio is characterized using a sample gate such as an inverter, it can be used to derive delay of all the other gates. We also propose some refinement techniques to improve accuracy. Experiments with industrial 32-nm library indicate that the average error over re-characterization is about 4% with maximum error being 11%, when maximum body bias voltage is assumed. The errors decrease as smaller bias voltage is applied.
Keywords
application specific integrated circuits; integrated circuit design; logic gates; ASIC design; body-biased circuit; gate delay modeling; inverter; process variation compensation; size 32 nm; static body biasing; static timing analysis; Capacitance; Delay; Inverters; Libraries; Logic gates; Optical wavelength conversion; Propagation delay;
fLanguage
English
Publisher
ieee
Conference_Titel
IC Design & Technology (ICICDT), 2012 IEEE International Conference on
Conference_Location
Austin, TX
ISSN
pending
Print_ISBN
978-1-4673-0146-6
Electronic_ISBN
pending
Type
conf
DOI
10.1109/ICICDT.2012.6232836
Filename
6232836
Link To Document