TY - JOUR
T1 - Analysis and optimization of BiCMOS gate circuits
AU - Kuroda, Tadahiro
AU - Sakata, Yoshinori
AU - Matsuo, Kenji
PY - 1994
Y1 - 1994
N2 - A comprehensive view of an optimization strategy for BiCMOS gates is described. A simple gate delay model is proposed. BiCMOS gate delay, when optimized, is found to be expressed as A + B√F, where F is fanout and A and B are coefficients. Since the coefficients can be extracted by SPICE simulation, the delay prediction can be precise, while keeping the delay formula simple enough for circuit designers to derive useful expressions. A procedure for optimizing BiCMOS gates is studied. BiCMOS gate delay can be calculated quickly and optimized efficiently just by looking up a design table which is obtained from SPICE simulations. The procedure for making the design table is technology-independent. Once obtained, the design table can be applied to any design with the same device technology. A sizing strategy of cascaded BiCMOS buffers is derived from the simple delay model. In a 0.8 μm, 9 GHz, BiCMOS process, a BiCMOS-BiCMOS cascaded buffer is optimized when the scale-up factor between two consecutive stages is e2.3 (≈10.0). A BiCMOS-CMOS cascaded buffer becomes the fastest when the scale-up factor, e1.6(≈5.0), is employed. The optimization procedure and the sizing strategy can be used for several variants of the basic BiCMOS gate, because the delay model is based on basic circuit models for the variants.
AB - A comprehensive view of an optimization strategy for BiCMOS gates is described. A simple gate delay model is proposed. BiCMOS gate delay, when optimized, is found to be expressed as A + B√F, where F is fanout and A and B are coefficients. Since the coefficients can be extracted by SPICE simulation, the delay prediction can be precise, while keeping the delay formula simple enough for circuit designers to derive useful expressions. A procedure for optimizing BiCMOS gates is studied. BiCMOS gate delay can be calculated quickly and optimized efficiently just by looking up a design table which is obtained from SPICE simulations. The procedure for making the design table is technology-independent. Once obtained, the design table can be applied to any design with the same device technology. A sizing strategy of cascaded BiCMOS buffers is derived from the simple delay model. In a 0.8 μm, 9 GHz, BiCMOS process, a BiCMOS-BiCMOS cascaded buffer is optimized when the scale-up factor between two consecutive stages is e2.3 (≈10.0). A BiCMOS-CMOS cascaded buffer becomes the fastest when the scale-up factor, e1.6(≈5.0), is employed. The optimization procedure and the sizing strategy can be used for several variants of the basic BiCMOS gate, because the delay model is based on basic circuit models for the variants.
UR - http://www.scopus.com/inward/record.url?scp=0028436218&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0028436218&partnerID=8YFLogxK
U2 - 10.1109/4.284708
DO - 10.1109/4.284708
M3 - Article
AN - SCOPUS:0028436218
SN - 0018-9200
VL - 29
SP - 564
EP - 571
JO - IEEE Journal of Solid-State Circuits
JF - IEEE Journal of Solid-State Circuits
IS - 5
ER -