Connecting pin B of resnsppoly - N+ poly resistor with salicide

[Shlapenka]

resnsppoly problem

Greetings all,

Here is my problem: in my design i use resnsppoly – N+ poly resistor with salicide, it has 3 pins - PLUS, MINUS and B (picture No.1). I connected it as is shown in picture No.2, the B pin to ground. In layout mode resistor is as shown in picture No.3. Its pins plus and minus are connected as according to my schematic, but what about the pin B? I cant see it anywhere.

Am i mistaking that pin B is ground? If it is ground pin where is it in the layout?

I use Cadence gpdk090v4.4.

Thank you so much for your help and time!

Pictures:

No.1:

**broken link removed**

No.2:

**broken link removed**

No.3:

**broken link removed**

 

[erikl]

Re: resnsppoly problem

in my design i use resnsppoly – N+ poly resistor with salicide, it has 3 pins - PLUS, MINUS and B (picture No.1).

If this is N+ in n-well, B ("Bulk") is the N+ contact. It should be connected to VDD - at least if you have pMOSFETs in the same n-well.

If it's N+ on substrate (like nMOSFET S/D contact), B is also the N+ contact. In this case, it could be connected to any potential. For lowest parasitic capacitance of the resistor to GND, it should also be connected to VDD.

 

[Shlapenka]

Re: resnsppoly problem

im sorry i mistyped it is actualy an resnsppoly – N+ poly resistor without salicide


it has these layers:

Poly
Pimp
SiProt & Resdum (Marker Layer)
Cont
Metal1

and is placed on substrate.

 

[erikl]

Re: resnsppoly problem

im sorry i mistyped it is actualy an resnsppoly – N+ poly resistor without salicide

This doesn't affect the B contact.

it has these layers:

Poly
Pimp
SiProt & Resdum (Marker Layer)
Cont
Metal1

and is placed on substrate.

So my second info above applies.

 

[Shlapenka]

Re: resnsppoly problem

so if ive got it right, i must make like an ntap wich will be the N+ contact?

 

[diemilio]

Re: resnsppoly problem

Hi Shlapenka, hi erikl.

erikl, are you sure of what you posted? cause it seems to me that the B terminal is should always be connected to the substrate potential and not to any potential or even the N+ contact. The resistor symbol that Shlapenka is using seems to be a distributed model for the resistor so the B terminal is to model the overall capacitance to subs. Now the only way of really knowing is by taking a look at the hspice/spectre model itself... Shlapenka, could you provide this, or at least part of it?

diemilio

 

[Shlapenka]

Re: resnsppoly problem

here is the info form specification:

Spectre Netlist

Spectre Model Name = “gpdk090_resnspoly”
R1 (B MINUS PLUS) resnsppoly_pcell1 segL=8u segW=1.5u
Subckt resnsppoly_pcell1 B MINUS PLUS
Parameters segL=8u segW=1.5u
R0 (PLUS MINUS B) gpdk090_resnsppoly l=segL w=segW
Ends resnsppoly_pcell1

DIVA LVS Netlist

DIVA Device Name = “resnsppoly”
; resnsppoly Instance /R1 = auLvs device R1
d resnsppoly PLUS MINUS B (p PLUS MINUS)
i 1 resnsppoly PLUS MINUS B " r 2137.33 w 1.5e-6 l 8e-6"

CDL Netlist

CDL Device Name = “resnsppoly”
RR1 PLUS MINUS B $[resnsppoly] r=2137.33 w=1.5u l=8u

Assura Netlist

Assura auLvs Device Name = “ resnsppoly ”
c resnsppoly RES PLUS B MINUS B B I ;;
* 3 pins
* 3 nets
S (p PLUS MINUS) ;
i R1 resnsppoly PLUS MINUS B ; r 2137.33 w 1.5e-06 l=8e-06;

 

[diemilio]

Re: resnsppoly problem

Hi,

Look for the spice/spectre model of gpdk090_resnspoly... you can find it pretty easily when you go to the Analog Design Environment and then setup --> Model libraries --> select the model file that contains the resistors (usually is the same file for everything, transistors, resistors, caps, etc...) and then look for gpdk090_resnspoly

diemilio

 

[Shlapenka]

Re: resnsppoly problem

here is what i found in spectre model:

**************************************************************************
* Non-Salicided P Poly Resistor
**************************************************************************
inline subckt gpdk090_resnsppoly (PLUS MINUS B)
parameters \
l = 0.6e-6 \
w = 0.6e-6 \
simM=1 \
Pwl=1.0 Pww=1.0 \
geo_fac=.01*sqrt(Pwl*Pwl/(l*w*1e12) + Pww*Pww/(w*w*1e12)) \
delmis= geo_fac * pres_mc / sqrt(simM) \
w_mc=w/(1+delmis)

gpdk090_resnsppoly (PLUS MINUS) resnsppoly l=l w=w_mc
ends gpdk090_resnsppoly

model resnsppoly phy_res subtype=poly rsh=rnspp cj=1.33e-3 cjsw=3.15e-10 tc1c=9.26e-4 mr=0.71*10m

 

[erikl]

Re: resnsppoly problem

Hi Shlapenka, hi erikl.
erikl, are you sure of what you posted? cause it seems to me that the B terminal is should always be connected to the substrate potential and not to any potential or even the N+ contact. The resistor symbol that Shlapenka is using seems to be a distributed model for the resistor so the B terminal is to model the overall capacitance to subs. ...
diemilio

Hi diemilio, hi Shlapenka,
actually I'm not so sure any more, because the info "resnsppoly – N+ poly resistor without salicide" made me think this is an unsalicided poly resistor over an N+ region (I've used such resistors, they have lower parasitics to substrate due to the possible isolation by a reverse-biased diode).

However, probably an N+ doped poly resistor is meant, and with Shlapenka's info "Pimp" I'd think diemilio is right saying it's probably a poly resistor over substrate, probably over P+ implant. So if there's really a "B" layout contact necessary (I couldn't find out from the models shown), it should of course be connected to GND.

Rgds, erikl.

 

[diemilio]

Re: resnsppoly problem

Well, it definitively isn't a distributed model as I though because the subckt is not composed of a series of resistors and capacitors (as distributed models usually are). The funny thing is that it seems that it seems the Model is not doing anything at all wit the B terminal!!! If you pay close attention the model is composed only of a set of equations that use the resistor's parameters and then a "call" to another model called resnsppoly but it only takes as inputs the terminals PLUS and MINUS!!

From the models I've seen that involve a bulk terminal the usually look something like this...

subckt blablabla (PLUS MINUS BODY)

equations equations equations

R1 (PLUS NEWTERMIAL)
C1 (NEWTERMINAL BODY)
R2 (NEWTERMINAL MINUS)

maybe more resistors and caps

equations equations equations

ends blablabla

So its funny that the terminal B is not used at all in the model you're showing...

diemilio

Added after 11 minutes:

I think you should run a set of parametric simulations connecting a voltage source at the B terminal and stepping its voltage to see how your results change... maybe that way you can get a better idea of what's going on...

 

[oermens]

resnsppoly problem

add substrate contact close to the device in layout.

 

[deepak242003]

resnsppoly problem

We can add substrate terminal to Hspice connecting it to ground... for resistor over substrate..
R1 A B RNNPO L=4U W=5U $SUB=VSSQ

I dont have any idea how it works on Spectre format..

 

[diemilio]

Re: resnsppoly problem

Well, now that I think about it... the B terminal might not be doing anything in the spectre model, but it might be there so that when you run parasitic extraction you get the parasitic caps from poly to subs. I'm not sure of this though, could someone confirm??

diemilio

 

[erikl]

Re: resnsppoly problem

Well, now that I think about it... the B terminal might not be doing anything in the spectre model, but it might be there so that when you run parasitic extraction you get the parasitic caps from poly to subs. I'm not sure of this though, could someone confirm??

Hi diemilio,
IMHO parasitic extraction has nothing in common with SPICE or SPECTRE models; it works purely from the layers' geometries and the extraction rules (which provide - inter alia - the length- and area-dependent values for cap calculation).

The SPICE or SPECTRE models, however, establish the basis for the parasitics' consideration in circuit analysis, i.e. for the pre-layout simulation, isn't it?

In the model texts presented above, I didn't find a reasonable calculation model for a parasitic cap between the poly resistor and the B (bulk) terminal.

Regards, erikl