On Feb 27, 12:24 pm, "Edward Schram" <[email protected]> wrote:
> I am starting to learn Electric and have managed to get an inverter, nand2
> and and2 completed and simulated in LTSpice.
>
> My questions are:
>
> When making the basic components what is the best way to make them "standard
> cells"?  Do I just make sure that the Vdd and gnd  are y units apart on each
> component?  Metal 2 for inputs and outputs?
>
> Is there am automated way to get Electric to turn a schematic view into a
> layout if it is designed using the standard cells in the library?
>
> Thanks for any help.
>
> Ed

The idea behind standard cells is that they are usually fixed in one
dimension (commonly height). For example all the cells in a library
can have a center-to-center (measured from center of power rail to gnd-
rail) of say 50 or 60 lambda. If you design all your cells in this
manner, then you can easily see that they can be abutted together into
a row. Then wiring tracks can be laid out above/below power/gnd lines
and connections can be easily made through vertically.

There is no automated way to turn a simple schematic into a layout.
For basic cells, you have to do the effort. Electric does have
automatic placement/routing but that depends on already having the
basic cells in place. Look at Lab 1 in 8-bit microprocessor  labs in
http://cmosedu.com/cmos1/electric/electric.htm to how to design a
sample standard cell library. Make sure to align your I/O pins to gnd/
vdd substrate contacts.

Also most standard cell libraries provide gates of different strengths
(having the ability of driving different loads). For example, a
inverter of strength 1 could be defined to drive a maximum load of X
units or to drive another inverter. Then an inverter of strength 2
would be designed to drive twice as much load as that of an inverter
with strength 1.

Standard cells can be optimized for various parameters: area, delay,
power, or a combination of those and other criteria (equal rise/fall
times). Depends on what you want. Generally, a good way to design
libraries is to size the transistors to give lowest "average" delay.
Average delay is defined as rise (propagation delay + fall propagation
delay)/2. How do you determine this ratio? You simulate the standard
FO4 inverter circuit (a cascade of 5 inverters with each inverter
being 4 times as big as its predecessor). The 3rd inverter is the
device under measurement. You run Spice simulations to figure out the
P/N ratio that gives the lowest average delay.
For example, in the AMI 0.5 um technology process from MOSIS this is
1.4:1. If you had designed for equal rise/fall delays, you would have
ended using a ratio of 2:1 which consumes more area.

I'll upload the f04.zip file and you can go through it to see the
sample circuit/optimization setup etc. Try to see if it runs in
LTSpice. I used it on HSpice and it ran fine.

-- 
You received this message because you are subscribed to the Google Groups 
"Electric VLSI Editor" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/electricvlsi?hl=en.

Reply via email to