|
|
Magic Mailing List |
|
From: Anand Kandasamy (anand AT bnl DOT gov) Date: Mon May 05 2003 - 11:49:18 EDT
Dear Magic Developers/Users:
We are currently using Magic 7.1 and we came across an issue
regarding the extractor. We see that the Perimeter of the
source and drain for a N-MOS device is not computed correctly.
The source/drain perimeter as per the layout is 26 Lambda's.
This is computed correctly when using Magic 6.5.1
***** .ext file from magic 6.5.1
timestamp 1052147600
version 5.1
tech scmos
style TSMC0.35um(tsmc35)from:04op
scale 1000 1 20
resistclasses 80900 152700 1017000 1017000 1 8100 45800 70 70 70 40
node "D1" 95 207.376 -13 -8 ndif 42 26 0 0 0 0 0 0 0 0 0 0 0 0 66 34 42 26 0
0 0 0
^^
node "S1" 95 207.376 -29 -8 ndc 42 26 0 0 0 0 0 0 0 0 0 0 0 0 66 34 42 26 0
0 0 0
^^
node "D2" 178 58.356 59 2 pdif 0 0 42 26 0 0 0 0 0 0 0 0 0 0 66 34 42 26 0 0
0 0
node "S2" 178 58.356 43 2 pdc 0 0 42 26 0 0 0 0 0 0 0 0 0 0 78 38 42 26 0 0
0 0
node "G1" 1011 1159.63 -45 -20 poc 0 0 61 46 0 0 0 0 0 0 144 50 0 0 248 128
0 0 0 0 0 0
node "C2" 1487 7169.58 31 -18 nw 70 34 0 0 2700 210 0 0 0 0 144 50 0 0 307
114 0 0 0 0 0 0
equiv "C2" "M2"
cap "G1" "S1" 90.7236
cap "C2" "D2" 202.04
cap "C2" "S2" 177.916
fet nfet -22 -8 -21 -7 54 30 "Gnd!" "G1" 18 "M1" "S1" 6 0 "D1" 6 0
fet pfet 50 2 51 3 54 30 "C2" "C2" 18 0 "S2" 6 0 "D2" 6 0
Whereas, using magic 7.1 the output is 38 lambda's
***** .ext file from magic 7.1
timestamp 1052147600
version 5.1
tech scmos
style TSMC0.35um(tsmc35)from:04op
scale 1000 1 20
resistclasses 80900 152700 1017000 1017000 1 8100 45800 70 70 70 40
node "D1" 521 207.376 -13 -8 ndif 42 38 0 0 0 0 0 0 0 0 0 0 0 0 66 34 42 26
0 0 0 0
^^
node "S1" 521 207.376 -29 -8 ndc 42 38 0 0 0 0 0 0 0 0 0 0 0 0 66 34 42 26 0
0 0 0
^^
node "D2" 178 58.356 59 2 pdif 0 0 42 26 0 0 0 0 0 0 0 0 0 0 66 34 42 26 0 0
0 0
node "S2" 178 58.356 43 2 pdc 0 0 42 26 0 0 0 0 0 0 0 0 0 0 78 38 42 26 0 0
0 0
node "G1" 1011 1159.63 -45 -20 poc 0 0 61 46 0 0 0 0 0 0 144 50 0 0 248 128
0 0 0 0 0 0
node "C2" 1487 7169.58 31 -18 nw 70 34 0 0 2700 210 0 0 0 0 144 50 0 0 307
114 0 0 0 0 0 0
equiv "C2" "M2"
cap "S1" "G1" 90.7236
cap "C2" "D2" 202.04
cap "C2" "S2" 177.916
fet nfet -22 -8 -21 -7 54 30 "Gnd!" "G1" 18 "M1" "S1" 6 0 "D1" 6 0
fet pfet 50 2 51 3 54 30 "C2" "C2" 18 0 "S2" 6 0 "D2" 6 0
It over estimates the PS/PD by 12 lambda's.
Here is the output in spice format
****Magic 6.5.1
M1 S1 G1 D1 Gnd! CMOSN W=1.20U L=1.80U AD=1.68P PD=5.20U AS=1.68P PS=5.20U
M2 S2 C2 D2 C2 CMOSP W=1.20U L=1.80U AD=1.68P PD=5.20U AS=1.68P PS=5.20U
****Magic 7.1
M1 S1 G1 D1 Gnd! CMOSN W=1.20U L=1.80U AD=1.68P PD=7.60U AS=1.68P PS=7.60U
M2 S2 C2 D2 C2 CMOSP W=1.20U L=1.80U AD=1.68P PD=5.20U AS=1.68P PS=5.20U
The correct PS/PD is 5.2u and not 7.6u as reported by magic 7.1 and up. I
tried
the 7.2 and it also overestimates the PS/PD of the NMOS device.
Has anyone else out there using 7.1 noticed this over-estimation of the
PS/PD numbers
for the NMOS device ?
Thanks
Anand
--
Anand Kandasamy
Research Engineer
Brookhaven National Laboratory
Upton, NY 11973 USA
Phone: (631)-344-5486
Fax : (631)-344-5773
e-mail: anand AT bnl DOT gov
|
|
|
|