Designer’s Guide Community

Alm Junior Member Offline Posts: 18	Why is CGBO defined per unit length and not area? Jan 22^nd, 2007, 7:08am Hi, The gate overlap capacitances CGSO and CGDO are defined per unit width. However the cap CGBO is deined per unit length? As length changes the the gate-bulk cap will obviously be affected. But will it not be affected by width as well? When calculating CGBOeff we use CGBO * Leff, but not W. Why? Regards, Alm P.
Back to top	IP Logged

Croaker

Senior Member

Offline

Posts: 235

Re: Why is CGBO defined per unit length and not ar
Reply #1 - Jan 23^rd, 2007, 6:33pm

CGBO is the cap formed by the poly outside the active that extends onto the field region. So no, the width of the MOSFET isn't related to this capacitance. If you draw a MOSFET layout this should make sense. Typically this cap is small relative to the gate-to-channel cap because the field oxide is thick compared to gate oxide.

Back to top

IP Logged

Alm

Junior Member

Offline

Posts: 18

Re: Why is CGBO defined per unit length and not ar
Reply #2 - Jan 26^th, 2007, 6:18am

Yes, that explanation helped a lot.

But there is some component of Cgb that depends on the width. For example for a simple CS amplifier I find the folowing values:

Op. Regn Width Cgb
Sat 70u 2.262 f
Sat 140u 4.3978 f
Sat 210u 6.5337 f
cutoff 210u 135.8588 f

There is another interesting thing I noted:

Vgs Op Regn Width Cgb
0.8 V Sat 210 u 38.99 f
0.5 V Sat 210 u 6.5337 f

i.e. as Vgs increases Cgb decreases. Why does this happen?

Regards,
Alm P.

Back to top

IP Logged

Croaker

Senior Member

Offline

Posts: 235

Re: Why is CGBO defined per unit length and not ar
Reply #3 - Jan 28^th, 2007, 12:54pm

You are right. It depends on the mode of operation. Basically, there are two capacitors in series between the gate and bulk. One is the oxide capacitance (Cox), and the other is the capacitance of the depletion region under the oxide (Cd). In depletion and weak inversion, Cd and Cox are similar in value. As the channel becomes more inverted Cd grows (a tiny change in the surface voltage makes a big change in the depletion charge) until Cox is pretty much the gate capacitance you see.

* The gate-to-bulk cap is no longer important because of a shielding effect; the depletion charge doesn't change much...all changes in charge are due to inversion and come from the source and drain. The only gate-to-bulk cap left that you see is the poly over field one.

In case my rambling made no sense:
In weak inversion Cgb = Cox*Cd/(Cox+Cd) + Cgb_field
In strong inversion Cgb = Cgb_field

For the change with width, the shielding effect is not perfect. The depletion charge may change a little, and so Cd, which is width dependent plays a small part. Cd=W*L*esi/tdep

Back to top

« Last Edit: Jan 28^th, 2007, 5:40pm by Croaker »

IP Logged

ywguo Community Fellow Offline Posts: 943 Shanghai, PRC	Re: Why is CGBO defined per unit length and not ar Reply #4 - Jan 28^th, 2007, 7:05pm Hi, Marc, Your comments are very good. But Alm said that Quote: There is another interesting thing I noted: Vgs Op Regn Width Cgb 0.8 V Sat 210 u 38.99 f 0.5 V Sat 210 u 6.5337 f Why the Cgb is larger at Vgs=0.8V than that at Vgs=0.5V? Thanks Yawei
Back to top	IP Logged

Croaker Senior Member Offline Posts: 235	Re: Why is CGBO defined per unit length and not ar Reply #5 - Jan 29^th, 2007, 7:08am I noted that too. It contradicts what he said and contradicts what I understand, so I assumed it was just a typo on his part. :o Quote: i.e. as Vgs increases Cgb decreases. Why does this happen? The above is correct, but his numbers got mixed up.
Back to top	IP Logged

Alm

Junior Member

Offline

Posts: 18

Re: Why is CGBO defined per unit length and not ar
Reply #6 - Jan 30^th, 2007, 6:04pm

Hi,

I don't think it is a typo.

Here is my spice file.

*circuit CS amp
m1 2 3 0 0 nmos w=70u l=1.6u
rd 1 2 500
rs 3 5 1
vb 5 4 0.5
vbac 4 0 ac 0.03 0
.include model3
.op
.end

I varied w to 140 and 210 and vb to 0.8 and 1.5 to report the earlier results.

here is the model file

.MODEL nmos NMOS ( LEVEL = 3

+ TOX = 3.21E-8 NSUB = 1.481713E15 GAMMA = 0.7819712

+ PHI = 0.7 VTO = 0.6375544 DELTA = 0.8987787

+ UO = 800 ETA = 9.999885E-4 THETA = 0.0689157

+ KP = 7.038973E-5 VMAX = 3.383919E5 KAPPA = 1.8146864

+ RSH = 0.047758 NFS = 6.04175E11 TPG = 1

+ XJ = 3E-7 LD = 1.192306E-11 WD = 6.837626E-7

+ CGDO = 1.64E-10 CGSO = 1.64E-10 CGBO = 1E-10

+ CJ = 2.845126E-4 PB = 0.8911576 MJ = 0.5

+ CJSW = 1.264309E-10 MJSW = 0.05 )

Back to top

IP Logged

Croaker

Senior Member

Offline

Posts: 235

Re: Why is CGBO defined per unit length and not ar
Reply #7 - Jan 31^st, 2007, 4:32am

Well, these statements seem to contradict each other. Is it a typo or not?

Quote:

Vgs Op Regn Width Cgb
0.8 V Sat 210 u 38.99 f
0.5 V Sat 210 u 6.5337 f

i.e. as Vgs increases Cgb decreases. Why does this happen?

For now I'll assume it's not a typo and mention that what I said is true under higher frequency gate voltages. If it's a low frequency, the series cap combination is at Cox below threshold and starts dropping until around threshold it grows again until it reaches Cox. This should be covered in any devices book in the chapter about MOSFETs.

Back to top

IP Logged

Alm

Junior Member

Offline

Posts: 18

Re: Why is CGBO defined per unit length and not ar
Reply #8 - Jan 31^st, 2007, 6:13am

Croaker,

Thanks for your patience. It was indeed a typo. Probably too much of hurry on my part. Apologize about that.

I think I understand your earlier expression that a small change in surface voltage makes a large change in depletion charge, which is why when vgs = 0.8V the cap reduces.

Thanks again.

-Alm

Back to top

IP Logged

Croaker Senior Member Offline Posts: 235	Re: Why is CGBO defined per unit length and not ar Reply #9 - Jan 31^st, 2007, 8:00am Yes, the depletion cap becomes huge and thus when it is in series it doesn't matter much.
Back to top	IP Logged