How to find the capacitance of a wire

[hardyboy_86]

Hi,
can any body suggest me How to calculate the the total capacitance of a Route???

 

[erikl]

Somewhere in your PDK docu (e.g. in the "analog characterization" description of your process) you should find metal-to-metal and metal-to-substrate capacitance values for each metal layer. Together with the calculated area of the route - if necessary including its fringing capacitance - you can estimate its total capacitance.

A simpler method would be analysis by simulation: here, the SPICE (or SPICE-like) algorithm will perform the calculations for you - and much more accurate. The usual method is to inject a unit (1V) ac voltage at the concerned node and run an ac analysis. The i(f) (of this node) plot will reveal you its total capacitance: C = (1/ω)*(i/v) = (1/2πf)*i (for v=1).

May be you can find its C_total directly in the sim data base, if your sim save options allow for it.

 

[timof]

Somewhere in your PDK docu (e.g. in the "analog characterization" description of your process) you should find metal-to-metal and metal-to-substrate capacitance values for each metal layer. Together with the calculated area of the route - if necessary including its fringing capacitance - you can estimate its total capacitance.

A simpler method would be analysis by simulation: here, the SPICE (or SPICE-like) algorithm will perform the calculations for you - and much more accurate. The usual method is to inject a unit (1V) ac voltage at the concerned node and run an ac analysis. The i(f) (of this node) plot will reveal you its total capacitance: C = (1/ω)*(i/v) = (1/2πf)*i (for v=1).

May be you can find its C_total directly in the sim data base, if your sim save options allow for it.

Erikl - can you explain your idea in more detail? I am not getting it... In order for SPICE simulations to show the parasitic capacitance of the wires (i.e. interconnects), it needs to be run in post-layout mode - i.e., after performing parasitic capacitance extraction. But, once you ran the parasitic extraction, you know all the parasitic capacitance values - why would you need to run SPICE simulations then?

 

[helpmejerry]

In my experience it is very difficult to calculate the total capacitance of a metal plate used four routing especialy in a 6-8 metal technology. This should get you started: https://eee.guc.edu.eg/Courses/Elec...Scale Integration/lect/VLSI_lecture3_ver2.pdf

Usually we check the functionality of a designed system via. spice or spectre and after the layout is done we do a post layout simulation to verify the results for designs up to tens oh MHz. That being said we do on occasion insert capacitance and resistance for pads and other critical parts. There are also papers on modeling firinge capacitance of interconects in VLSI etc.

 

[erikl]

Erikl - can you explain your idea in more detail? I am not getting it... In order for SPICE simulations to show the parasitic capacitance of the wires (i.e. interconnects), it needs to be run in post-layout mode - i.e., after performing parasitic capacitance extraction. But, once you ran the parasitic extraction, you know all the parasitic capacitance values - why would you need to run SPICE simulations then?

Hello Max,
you're right: all the parasitics are available after extraction. However, a SPICE simulation
1. relieves you from the task of adding several of them, especially if they are distributed over several metal layers - and SPICE won't forget any of them ;-)
2. if such a node is connected to a device terminal, there will probably be voltage-dependent caps adding to Ctotal. SPICE can add those voltage-dependent caps, which might be desirable if you wish its Ctotal in a real application.

 

[timof]

Hello Max,
you're right: all the parasitics are available after extraction. However, a SPICE simulation
1. relieves you from the task of adding several of them, especially if they are distributed over several metal layers - and SPICE won't forget any of them ;-)
2. if such a node is connected to a device terminal, there will probably be voltage-dependent caps adding to Ctotal. SPICE can add those voltage-dependent caps, which might be desirable if you wish its Ctotal in a real application.

Hi Erikl -

I see your point.

On (1) - does not your design environment allow you to see the total capacitance (without having go through a tedious manual addition of all the coupling capacitance components)?
In DSPF (or SPICE netlist) generated by parasitic extraction tools, total net capacitance are reported and can be easily seen directly...

On (2) - running SPICE simulation is an advantage if you'd like to see the total (interconnects and devices) capacitance, but disadvantage if you'd like to see only parasitic capacitance...

Max
----------

 

[erikl]

On (1) - does not your design environment allow you to see the total capacitance (without having go through a tedious manual addition of all the coupling capacitance components)?
In DSPF (or SPICE netlist) generated by parasitic extraction tools, total net capacitance are reported and can be easily seen directly...

Hi Maxim,
probably you're right with the new extraction tools: my knowledge still relates to old tools like DIVA and assura: then, the parasitics were distributed over the metal layers, and one had to add them manually (I remember to have written an awk script for this purpose ;-) ). I think DSPF (or SPEF) standard formats weren't available then.

On (2) - running SPICE simulation is an advantage if you'd like to see the total (interconnects and devices) capacitance, but disadvantage if you'd like to see only parasitic capacitance...

Sure: if you want to see the contribution of the (possibly not optimal) routing only ...