Use of PISCES POSTMINI TOOL

[dsrinivasrao]

hi colbaidh,plz give some valuable tips for seeing clock feed through in pisces.........

 

[dsrinivasrao]

Hi can u plz tell me whether we can put a inductor in pisces or not?Actually i read the pisces manual but i not able to find it anywhere in the manual??..Kindly help me

 

[Colbhaidh]

When the gate voltage changes from high to low, there is an imbalance of electrons in

the channel required to mirror the gate charge. This imbalance forces electrons out of

the channel preferentially to the source or drain terminals causing charge injection

errors. However as electrons are moving, this creates a potential drop from the centre

of than channel to the edges. If this potential drop becomes large enough, the

potential in the channel can drop below the body potential causing electrons to be

injected into the body instead of the source/drain terminals. This can actually reduce

the injection error seen at the source drain terminals.
So you need to consider two directions of current flow x-direction from channel to

source and drain and y-direction from channel to body.

Ignoring the vertical current for now, then the error voltage due to the channel charge

injection (assuming all of chennel charge is moved - half goes to source and half goes

to drain) should be:

dV = Qch/(2*Ch)

Clock feedthrough from overlap capacitance

dV = dVgs(Cov/(Cov+Ch))

Combined effect of these two is to lower the voltage at Ch. This then creates a voltage

difference between source and drain inducing a current that connects Ch to the other

terminal which recovers some of the voltage loss at Ch.

dV = Ids.dt/Ch

I would simulate at gate voltage at 1.8V and a source voltage of 1.0V with the holding

capacitor on the drain. Runa transient where the gate drops to zero and monitor the

voltage on the drain (or holding capacitor). This should drop below 1V during the

transient then recover a little.

I do not think Pisces2B can handle inductances.

 

[dsrinivasrao]

thank you very much Colbhaidh...Can i knw your full name and your profession??

 

[Colbhaidh]

I am a consultant in Industrial Science but predominatelty the semiconductor industry.
Dont think we are supposed to use real names in the forum.

 

[dsrinivasrao]

Oh in Semiconductor Industry good...I am also placed in a VLSI Company nw...Sorry but i have a doubt...Can u plz explain me whats the actual meaning of "ratio" while writing the mosfet grid..r=1 & r=1.5 whats the actual difference?Is it a ratio of voltages two consecutive grid points ???..Kindly clear this doubt....

 

[Colbhaidh]

The ratio parameter determines where grid lines are placed between the previous x.mesh or y.mesh statement and this one. So the space between mesh lines gets smaller if ratio is <1.0 and grows if ratio >1.0

 

[dsrinivasrao]

Hi colbaidh..I am making a source capacitance( Cs )now..So for making Source capacitance=Holding capacitor Ch..So i had extended the my grid on the source side similarly to waht i did when i am constructing the holding capacitor on drain side..but when i am doing DC analysis it is not converging at all...Though my Transient analysis is working alright...Can u suggest me what could be the reason??..My grid files and DC analysis file are below:

TITLE **** MOSFET (.18um channel) with both Cs=0.72um and Cd=0.72um*******


$*******Rectangular Grid ***************

mesh rectangular nx=180 ny=28 diag.fli outf=mosfet.mesh

X.mesh n=1 l=0 r=1
x.m n=180 l=1.62 r=1
Y.mesh n=1 l=-0.004 r=1
y.m n=6 l=0 r=1
y.m n=15 l=0.014 r=1
y.m n=28 l=0.24 r=1.30

$*********Eliminating coloumns*************

elim x.dir iy.l=9 iy.h=19 ix.l=1 ix.h=81
elim x.dir iy.l=9 iy.h=19 ix.l=101 ix.h=179
elim x.dir iy.l=8 iy.h=19 ix.l=1 ix.h=81
elim x.dir iy.l=8 iy.h=19 ix.l=101 ix.h=179

$ *************** regions ****************
region num=1 ix.l=1 ix.h=180 iy.l=1 iy.h=5 oxide
region num=2 ix.l=1 ix.h=180 iy.l=5 iy.h=28 silicon

$ ************** electrodes **************
$ 1 = Gate 2 = body 3=bottom of holding capacitor 4=bott

elec num=1 ix.l=81 ix.h=100 iy.l=1 iy.h=1
elec num=2 ix.l=1 ix.h=180 iy.l=28 iy.h=28
$elec num=3 ix.l=1 ix.h=80 iy.l=6 iy.h=6
elec num=3 ix.l=102 ix.h=180 iy.l=4 iy.h=4
elec num=4 ix.l=1 ix.h=80 iy.l=4 iy.h=4

$ *************** doping ******************
DOP UNIF CONC=4E17 P.TYPE
DOP GAUSS CONC=5E19 N.TYPE
+ X.RIGHT=.70 y.top=0 y.bot=0.024 char=0.018
DOP GAUSS CONC=5E19 N.TYPE
+ X.LEFT=.93 y.top=0 y.bot=0.024 char=0.018

$**************** Model Card****************
MODELS Srh FLdmob CONSrh BGn CONmob

$ ************** regrid ********************
$
$ *** first regrid on potential
$
contact num=1 n.poly
symb carriers=0
solve init
regrid poten ratio=0.2 reg=2 smooth.k=1 outf=mosfetgrid45_2
end

DC Analysis:-

title NMOS Ids-Vgs & Ids-Vds characteristics :n4b.i
$
mesh infile=mosfetgrid45_2
contac num=1 n.poly
contac num=2 Neutral
contac num=3 Alum
contac num=4 Alum

symb gummel carriers=1 electrons
method iccg damped

solve init currents outfile= out

symb newton carriers=1 electrons
method ITLIMIT=500 autonr

log outf=I_vs_V.log

SOLVE PREV v1=0.1 v2=0 v3=0 v4=0 OUTF=OUT1


solve v1=0 vstep=0.1 nsteps=30 electrode=1 currents outf=g0



MODELS Srh FLdmob CONSrh BGn CONmob


end

 

[Colbhaidh]

The MODELS statement must come before a SOLVE statement. So adjust the DC analysis file to:

title NMOS Ids-Vgs & Ids-Vds characteristics :n4b.i
$
mesh infile=mosfetgrid45_2
contac num=1 n.poly
contac num=2 Neutral
contac num=3 Alum
contac num=4 Alum
$
symb gummel carriers=1 electrons
method iccg damped
solve init currents outfile= out
$
symb newton carriers=1 electrons
method ITLIMIT=500 autonr
MODELS Srh FLdmob CONSrh BGn CONmob <<<<<<<<<<<<<<<<<<<<
log outf=I_vs_V.log
SOLVE PREV v1=0.1 v2=0 v3=0 v4=0 OUTF=OUT1
solve v1=0 vstep=0.1 nsteps=30 electrode=1 currents outf=g0
$
end

and everything converges.

Note that you are driving the gate to 3 Volts. This is too high for a 4nm gate oxide.

 

[dsrinivasrao]

hi,when i am doing the DC analysis with the gate voltage varying its working properly,but when i am doing DC analysis with source voltage varying the drain voltage is not responding accordingly..


title NMOS Ids-Vgs & Ids-Vds characteristics :n4b.i
$
mesh infile=mosfetgrid45_2
contac num=1 n.poly
contac num=2 Neutral
contac num=3 Alum
contac num=4 Alum
$
symb gummel carriers=1 electrons
method iccg damped
solve init currents outfile= out
$
symb newton carriers=1 electrons
method ITLIMIT=500 autonr
MODELS Srh FLdmob CONSrh BGn CONmob <<<<<<<<<<<<<<<<<<<<
log outf=I_vs_V.log
SOLVE PREV v1=0.1 v2=0 v3=0 v4=0 OUTF=OUT1
solve v1=0 vstep=0.1 nsteps=30 electrode=1 currents outf=g0
solve v3=0 vstep=0.1 nsteps=30 electrode=3 currents outf=h0
end

When i am solving this DC sol file, my drain voltage is not varying with the source voltage..for eg:-for my solution file h4 my source voltage= 0.5 but my drain voltage is remaining at Ovolts only..Can you plz suggest me what could be the reason?????..In my transient analysis my drain voltage is following the source voltage...kindly help me Colbaidh

 

[Colbhaidh]

You have the holding capacitors in series with the source and drain so any DC voltage applied to the source side capacitor will block this DC getting to the source. You only need one capacitor. If you want it on the source then remove the drain capacitance and apply a DC voltage to the drain or visa versa for a drain holding capacitor. Having capacitors on both source and drain does not model any real situation (that I can think of) for clock through.
Having a large holding capacitor on say the drain side and applying a DC voltage on the source and looking at transients is modelling a sample and hold circuit (e.g. an ADC or chopper amplifier input). Using a smaller holding capacitor models the transistor drain connected to the gate (i.e. MOS Capacitor) of another device e.g. a transfet gate driving an invertor or similar.

 

[dsrinivasrao]

As you rightly said that having capacitors on both the side doesn't models any real situation,but having capacitors on both sides we can predict how the charge behaves when transistor cuts off..Like say Holding cap on source is C1 and on drain side is C2..So depending on the values of C1 & C2 we expect some charge distribution..E.g. If C1=C2 the charge in the channel should get distribute 50-50 on drain &source side..If C1<C2 then the impedance on source side will be more and hence we expect more charge should flow to the drain side instead of source side..and opposite for C1>C2..But when i doing a transient analysis for C1=C2 I am seeing more error on drain side but i expect the error should be same on both side for the case C1=C2..So what i should i do to see this behaviour????

 

[Colbhaidh]

I would model one side (source or drain) with its inherent capacitance and the other side with the holding capacitance. Lets assume we have the holding capacitor on the drain side, so set up the simulation as before you added the source capacitor.
Now for the source capacitance use the contact statement in Pisces
contact num=XXX cap=XXXX

This allows you to set the voltage on the source as normal but models a source capacitance that is not within your mesh. You can now directly add DC voltages to the source and you can set the source capacitance to anything you want....but if you want to put the source capacitance as a ratio of the drain holding capacitor, you need to extract the actual value of the drain capacitor.

 

[dsrinivasrao]

Hi colbhaidh,i did the simulation set up as before adding the source capacitor and put a capacitor=1E-12 at contact=3 (my source terminal),i used here a full newton method only as you described in your earlier replies..But it is giving me a constant voltage of nearly 2.4 0r 2.5 volts always even though i am changing the voltage at source terminal..but if am using a solution file with both gummel & Newton &with no source cap value,my drain voltage is following the source voltage..Hope u understand what i want to explain you..

the mosfetgrid which i am using is:-

$*******Rectangular Grid ***************

mesh rectangular nx=108 ny=28 diag.fli outf=mosfet.mesh

X.mesh n=1 l=0 r=1
x.m n=108 l=0.99 r=1
Y.mesh n=1 l=-0.004 r=1
y.m n=6 l=0 r=1
y.m n=15 l=0.014 r=1
y.m n=28 l=0.24 r=1.30

$*********Eliminating coloumns*************

elim x.dir iy.l=9 iy.h=19 ix.l=1 ix.h=9
elim x.dir iy.l=9 iy.h=19 ix.l=28 ix.h=108
elim x.dir iy.l=8 iy.h=19 ix.l=1 ix.h=9
elim x.dir iy.l=8 iy.h=19 ix.l=28 ix.h=108

$ *************** regions ****************
region num=1 ix.l=1 ix.h=108 iy.l=1 iy.h=5 oxide
region num=2 ix.l=1 ix.h=108 iy.l=5 iy.h=28 silicon

$ ************** electrodes **************
$ 1 = Gate 2 = body 3 = source 4=bottom of holding capacitor

elec num=1 ix.l=9 ix.h=27 iy.l=1 iy.h=1
elec num=2 ix.l=1 ix.h=108 iy.l=28 iy.h=28
elec num=3 ix.l=1 ix.h=8 iy.l=6 iy.h=6
elec num=4 ix.l=30 ix.h=108 iy.l=4 iy.h=4

$ *************** doping ******************
DOP UNIF CONC=4E17 P.TYPE
DOP GAUSS CONC=5E19 N.TYPE
+ X.RIGHT=.07 y.top=0 y.bot=0.024 char=0.018
DOP GAUSS CONC=5E19 N.TYPE
+ X.LEFT=.27 y.top=0 y.bot=0.024 char=0.018

$**************** Model Card****************
MODELS Srh FLdmob CONSrh BGn CONmob

$ ************** regrid ********************
$
$ *** first regrid on potential
$
contact num=1 n.poly
symb carriers=0
solve init
regrid poten ratio=0.2 reg=2 smooth.k=1 outf=mosfetgrid45_1
end

AND the solution file which i am using (having both gummel & Newton method) is :-

title Charge Injection Test
$
mesh infile=mosfetgrid45_1
contac num=1 n.poly
contac num=2 neutral
contac num=3 neutral
contac num=4 neutral

symb gummel carriers=1 electrons
method iccg damped

solve init currents

symb newton carriers=1 electrons
method autonr

log outf=tf50nsec.log

solve v1=0 v4=0 v2=0 v3=2 outf=ab
solve v1=1.8 TSTEP=1E-9 TSTOP=10E-9 RAMPTIME=1E-9 currents
solve v1=0 TSTEP=1E-9 TSTOP=60E-9 RAMPTIME=50E-9 currents outf=t0

end

This solution file is giving me the exact results..Like when i am giving source voltage=1v my drain voltage becomes 1v,When i am giving source=2v my drain is becoming 2volts..

But when i put cap=1E-12 in the solution file(containing only newton method) i am not getting the expected values of result..The solution file which i am using is:-
title Charge Injection test
$
mesh infile=mosfetgrid45_1
contac num=1 n.poly
contac num=2 neutral
contac num=3 Alum cap=1E-12
contac num=4 Neutral

symb newton carriers=1 electrons
method ITLIMIT=200 autonr

solve init currents
log outf=tf50n_1.log

solve v1=0 v2=0 v3=1 v4=0 outf=ab
solve v1=3 TSTEP=1E-9 TSTOP=10E-9 RAMPTIME=1E-9 currents
solve v1=0 TSTEP=1E-9 TSTOP=60E-9 RAMPTIME=50E-9 currents outf=r0

end

In this file,For v1=0 v2=0 v3=0 v4=0 (i.e for solve init) it is not converging,Also for v1=0 v2=0 v3=1 v4=0 it is not converging..Remaining for all the transient values it is converging..Pleeeeeeez help me regarding this capacitor issue???? Its very important for my thesis defination...

 

[dsrinivasrao]

Hi colbhaidh,plz help me regarding how to put cap in the previous reply TRANSIENT solution..Pleeeeeeeeeeeezzzzzzzzzz its very imp....

---------- Post added at 15:49 ---------- Previous post was at 15:36 ----------

Can somebody plz tell me how to find the exact value of feedthrough voltage of a single mosfet..???

Actually i did one exp,I took one NMOS put a equal Capacitor on drain & source side..Now i applied a voltage of 0.5 amplitude & 0.2 dc offset..Whatever be the output voltage i think this is feed through voltage..Am i correct?..This i am claiming because as gate dc voltage is 0.2 which is less than Vt of nmos and hence it will be in always cut off mode..Plz correct me if i am wrong

 

[dsrinivasrao]

Hello colbhaidh sorry i had disturbed u a lot..One very imp question...Actually when i am doing transient analysis(only with newton ) and giving a capacitor value,my solution is not at all converging..So plz help me??..And whats the difference between putting a RES=1E3 and con.resis=1E3..Is it the same thing or different??

 

[dsrinivasrao]

hello colbhaidh i tried all the possible ways to put a Ch on source side,but my solution is not converging at all...Plz help me..........thank you

 

[dsrinivasrao]

Kindly help me in putting the source capacitance...

title Charge Injection test
$
mesh infile=mosfetgrid45_1
contac num=1 n.poly
contac num=2 neutral
contac num=3 Alum cap=1E-12
contac num=4 Neutral

symb newton carriers=1 electrons
method ITLIMIT=200 autonr

solve init currents
log outf=tf50n_1.log

solve v1=0 v2=0 v3=1 v4=0 outf=ab
solve v1=3 TSTEP=1E-9 TSTOP=10E-9 RAMPTIME=1E-9 currents
solve v1=0 TSTEP=1E-9 TSTOP=60E-9 RAMPTIME=50E-9 currents outf=r0

end

 

[dsrinivasrao]

hi colbaidh i got source resistance working that to when i am make my gate electrode=neutral instead of n.poly,the plots which i got are verry much expected..But when i am putting some Cs it is still converging but the charge is not splitting depending on the values of C1 and C2..Can you plz help me in this issue??

title Charge Injection test
$
mesh infile=mosfetgrid45_1
contac num=1 Neutral
contac num=2 Alum
contac num=3 Alum cap=1E-12 res=1e1
contac num=4 Neutral

symb newton carriers=1 electrons
method ITLIMIT=200 autonr

solve init currents
log outf=tf50n_1.log

solve v1=0 v2=0 v3=1 v4=0 outf=ab
solve v1=3 TSTEP=1E-9 TSTOP=10E-9 RAMPTIME=1E-9 currents
solve v1=0 TSTEP=1E-9 TSTOP=60E-9 RAMPTIME=50E-9 currents outf=r0

end

 

[dsrinivasrao]

Hi, I want to draw the graph for derivative of drain currrent (values present in log file) Vs Drain Voltage..Can u tell me the steps??