The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Modeling >> Semiconductor Devices >> Source/drain symmetric in MOSFET model
https://designers-guide.org/forum/YaBB.pl?num=1137464811

Message started by 337see733 on Jan 16^th, 2006, 6:36pm

Title: Source/drain symmetric in MOSFET model
Post by 337see733 on Jan 16^th, 2006, 6:36pm

Dear all,
Anyone who has experienced in solving the compact model's source/drain gummel symmetric issue?
I've used symmetric Vdeff but still there are some intrinsic problem of the fundamental equation. Any gurus, please advice. ;)

Title: Re: Source/drain symmetric in MOSFET model
Post by rf-design on Jan 17^th, 2006, 1:21am

Do you want to simulate a potential based model or want to setup equation based on surface potential?

Title: Re: Source/drain symmetric in MOSFET model
Post by Geoffrey_Coram on Jan 17^th, 2006, 6:35am

Why don't you take the CMC's next standard MOSFET model, PSP?
http://pspmodel.ee.psu.edu

One starting point was the Penn State SP model, which used a "symmetric linearization" method to handle charge partitioning along the channel. The BSIM models are fairly symmetric in current (but not its 2nd derivative) but have trouble with the channel capacitance.

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Jan 17^th, 2006, 5:10pm

Wow, so fast got respond in the community d! Great and thanks! ;D

Basically, I'll need to setup/develop the current equations based on surface potential...I noticed how PSP handled using the symmetric linearization method, which i also tried to follow. But, it seems to me I'm still missing some pieces of the trick/puzzle here and there. I'm suspecting that there could be some intrinsic problem due to the way how i handled the intrinsic regional surface potential at inversion(which is simply following Philip's "old" model in strong inversion...). So, i'm hoping to find some clue from anyone who worked on this before....Thanks...

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Jan 17^th, 2006, 7:41pm

Upzz....for lunch..

I realized with the PSP's approach...it can work on the old Philip's strong inversion...Now, my difference potential between source and drain is symmetric...but my Id still not symmetric!...sigh...... :-/

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Jan 18^th, 2006, 1:40am

Upzz...before going back..

Hmm...i might need to take back the previous claim...still have more problems...... :'(
Anyone?Anyone?Anyone? Please enlighten...ha ha ha ;D

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Jan 18^th, 2006, 5:51pm

Upz....anyone??

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Jan 18^th, 2006, 10:21pm

Dear ALL,

When i look at the normalized equation, i ran the gummel test on the simplest equation, normalized(mobilty,width and length) it seems that it was already not symmetry? How come??? 1st figure is Id vs Vx, 2nd, DId vs Vx, 3rd, D2Id vs Vx

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Jan 20^th, 2006, 12:31am

8-)

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Jan 20^th, 2006, 3:14am

Anyone do this test on the model? Do they tune to get the symmetry behavior? ;)

Title: Re: Source/drain symmetric in MOSFET model
Post by Geoffrey_Coram on Jan 20^th, 2006, 5:03am

I'm not clear on what you think your problem is. In your first set of figures, the 1st derivative looks like two straight lines, so maybe you didn't take enough points to compute a good numerical derivative.

Are you concerned that Id(Vds) != Is(-Vds)? Or are you trying to pass the Gummel symmetry test with a couple derivatives?

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Jan 20^th, 2006, 8:28pm

Thanks Geoffrey fur your reply. ;D

The 1st derivative looks like two straight lines are not cause by lack of numerical points. It was actually the way my data is computed. I just sweep one site of Vx, then do a image reflect onto the other site. So, my transition point will always go through zero.

Quote:

Are you concerned that Id(Vds) != Is(-Vds)? Or are you trying to pass the Gummel symmetry test with a couple derivatives?

I thought both are actually related? If my Id(Vds)!=Is(-Vds) then I'll failed the Gummel symmetry test, right? :-/ Now, I'm more concerned with the passing the Gummel symmetry test at the 2nd derivative....

Title: Re: Source/drain symmetric in MOSFET model
Post by Geoffrey_Coram on Jan 23^rd, 2006, 4:24am

337see733 wrote on Jan 20^th, 2006, 8:28pm:

I just sweep one site of Vx, then do a image reflect onto the other site. So, my transition point will always go through zero.

I think you really need to sweep Vx on both sides to properly apply the Gummel symmetry test.

Quote:

I thought both are actually related? If my Id(Vds)!=Is(-Vds) then I'll failed the Gummel symmetry test, right?

Yes, but in a previous post, it seemed that you were having a problem with Id(Vds) = Is(-Vds), never mind the derivatives.

Also, I don't think it's relevant yet, but may become so: the Gummel symmetry test as originally specified fails if diode junction currents or gate leakage currents are included. Colin McAndrew put together a "modified Gummel symmetry test" starting on page 6 of this document:
http://www.eigroup.org/cmc/next_gen_cmos/phase3/freescale_ngsmm_evaluation_hisim2.2.0b_noibmdata.pdf

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Jan 23^rd, 2006, 5:14pm

Quote:

I think you really need to sweep Vx on both sides to properly apply the Gummel symmetry test.

Oh, sweep both sides, u mean simultenaously Vd-Vx and Vs+Vx ? If that is the case, i do sweep together.
Cause the way i formulate the current is basically reference to source. Therefore, the input will looks like:
vx=0..1.2
Vd=2vx;
Vg=vg-vx;
Vb=vx-vb;
where current are function of input voltages and width and length,
Id=f(Vd,Vg,Vb,W,L);
After which, i'll flip both the vx and Id(without the 1st point,which is vx=0) to plot for gummel symmetry test result.

Title: Re: Source/drain symmetric in MOSFET model
Post by Geoffrey_Coram on Jan 25^th, 2006, 10:06am

No, I meant that you need vx to go from -1 to +1.

You should have vb and vg fixed during the sweep (and run the test at several vg and several vb values).

VD should be +vx and VS should be -vx.

I think if you try to reference to source, you'll confuse the issues.

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Jan 25^th, 2006, 5:40pm

Quote:

No, I meant that you need vx to go from -1 to +1. You should have vb and vg fixed during the sweep (and run the test at several vg and several vb values). VD should be +vx and VS should be -vx.

Yes, I think i should try to do this. ;D

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Feb 6^th, 2006, 1:43am

I've include the Vs in everywhere of the equation(codes)...sgh...There is some problem with the transition from vx to -vx...

I assume this is coding problem rather than the equation itself...Because, I checked at a single bias of Vd and Vs or (Vs and Vd interchange), the output current are the same.

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Feb 6^th, 2006, 6:58pm

Quote:

No, I meant that you need vx to go from -1 to +1. You should have vb and vg fixed during the sweep (and run the test at several vg and several vb values). VD should be +vx and VS should be -vx.

Yes, I think i should try to do this.

Finally, i did that. But, the answer looks similar to the one i do with the flip-reflect approach!

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Feb 6^th, 2006, 10:43pm

Hmm...This is symmetry, right? ;D

Title: Re: Source/drain symmetric in MOSFET model
Post by Geoffrey_Coram on Feb 7^th, 2006, 9:45am

337see733 wrote on Feb 6^th, 2006, 6:58pm:

Finally, i did that. But, the answer looks similar to the one i do with the flip-reflect approach!

But with the flip-reflect approach, you haven't actually tested anything. If you want to test that f(x) is an odd function, you can't test it for positive x and then plot it for negative x using the flip-reflect approach.

Your last post looks like you've solved your problem now, though.

Title: Re: Source/drain symmetric in MOSFET model
Post by 337see733 on Feb 7^th, 2006, 5:23pm

Quote:

Your last post looks like you've solved your problem now, though.

Yes...finally......still testing it to verify. ;D Thanks a thousands...