The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> Gm calculation of a MOS with constant current source
https://designers-guide.org/forum/YaBB.pl?num=1355401527

Message started by VINAY RAO on Dec 13^th, 2012, 4:25am

Title: Gm calculation of a MOS with constant current source
Post by VINAY RAO on Dec 13^th, 2012, 4:25am

I have forced 500uA through the MOS as shown in figure 4. I kept Vds constant and varied Vgs from 0 to 1.1 v. And extracted Gm of MOS from the spice which is shown in fig 5.
As per theory, when Vgs is far less from Vt i.e. means in WEAK INVERSION(WI) Gm is constant and depends only on the bias current (here it is 500uA). Also Gm has to reduce from its maximum value when we move from WEAK INVERSION (WI) to STRONG INVERSION (SI).

Question : Here in simulation that too in WI, why Gm starts increasing with Vgs first and later only in SI it starts falling?

Also in WI simulated values are not matching with theoretical values:
WI: Gm=(Id/nUt)=19.3mS. But in simulation Gm=400uS.
SI: at Vgs=0.6: Gm=2Id/(Vgs-vt)=8mS but in simulation its around 1.4mS.
Theoretical values are kind of overestimating (Moreover velocity saturation effects are less at Vgs=0.6). Is this the case with every one?

Regards,
Vinay Rao.

Title: Re: Gm calculation of a MOS with constant current source
Post by raja.cedt on Dec 13^th, 2012, 4:38am

hello,

I didn't understand what is your Goal here?

How did you make Vds constant while varying the current? (did you adjusted W/L along the sweep). All your explanation valid only if you have Vds across the transistor, but looks like your Vds is very low (because when you are increasing Vgs current though transistor will increase, but current source makes vds low to maintain same current through both of them).

May i ask you to show Vds when your vgs is increasing?

Thanks,
Raj.

Title: Re: Gm calculation of a MOS with constant current source
Post by VINAY RAO on Dec 13^th, 2012, 7:25am

Thanks for your quick reply Raj. Here I attached the response: 1st row- ID through MOS, 2nd row- VDS, 3rd row: Gm.

Purpose: I wanted to record the response of Gm w.r.t. Vgs when Id is constant. In simulation, I found 2 ways. First- keep constant current by varying W/L, second option to have constant current is to force ID by some ideal current source. Now I find second option is illogical to opt. But anyhow, how can I change W/L simultaneously with varying Vgs in simulation to record Gm vs Vgs in all 3 regions (WI, MI & SI)?

VDS: Here its coming around 6 volts. Though it's bit illogical-one question is striking me. Is that ideal current source which is charging Cdb and Cgd cap and increasing VDS to 6v?

Gm: Its true that in this case those equations becomes invalid. But if we observe in WI, we got large VDS value and still Gm is increasing w r t Vgs. How can I interpret this phenomenon?

Regards,
Vinay Rao

Title: Re: Gm calculation of a MOS with constant current source
Post by raja.cedt on Dec 13^th, 2012, 8:35am

hello,

How can you keep a mosfet current constant by changing vgs while keeping Vds constant.
. Please find the attached file.

Answers to your Questions.
1. Why VDS went to 6V? it's because when you have very low Vgs say 0, transistor current will be zero, but some how it should get 500uA due to current source so By making Vds extremely high it get's the final state. If you have some real current source (say Pmos) it would have been stopped at vdd rather than going to hig.

2. Why gm is increasing with Vgs: In week inversion ideally Gm depends on only ID and there is week dependence on Vds, since you have horrible Vds i don't have any comments.

Please remember Vth changes a lot with Vds, if you are working with below 130nm.

Have fun.

Title: Re: Gm calculation of a MOS with constant current source
Post by VINAY RAO on Dec 13^th, 2012, 11:26pm

Hello Raj,
What you said is conceptually true but I am facing problem when I try to simulate your setup. Here is my spice code for your setup.

x1 d g 0 0 NMOS lf=0.06u wf=p nf=2 m=1 con=1 * MOS with W=p as a parameter
.param p=0.5u
Eopamp g 0 d ld MAX=1.1 MIN=0 1e6 *OPAMP with a gain of 1e6 and (g,0) as O/P and (d,ld) as I/P.
Vds ld 0 DC 1.1 * Connecting 1.1 v to –ve terminal ld of OPAMP
Id d s DC 0.5m
Vdummy s 0 DC 1
.DC p 0.5u 8u 0.5u
.plot DC I(x1.mos) V(d) v(g)
.probe gm=par('gmo(x1.mos)')

When I am using opamp, both vd (drain voltage) and Id are becoming negative which is shown in figure (1st row-ID, 2nd row- Gm, 3rd row-Vg and 4th row- Vd.). Hence drain is becoming source and vice versa. Why Vd is becoming -ve in this setup instead of following opamp's other terminal voltage? Vg (gate voltage) also would have gone into negative if I could have given negative minimum instead of zero to the opamp.

Gm: Gm is increasing w.r.t. W and this is what expected.

Id: At drain, we have one current from ideal current source and one more negative current flowing into the MOS. Is it not equally saying both the current are getting added at node DRAIN? If so then is it not violating KCL? Or am I misinterpreting the hspice expression I(x1.MOS) ? Where I am missing the string ?

Title: Re: Gm calculation of a MOS with constant current source
Post by raja.cedt on Dec 14^th, 2012, 4:09am

hello,
Very difficult to debug the spice net-list, you should take care of that.

For example idc should be from node s to d rather s to d. Take care of the spice deck, why don't you use schematic rather than writing net-list.

Thanks,
Raj.