Update doc

This commit is contained in:
Matthias Koefferlein 2024-06-25 23:45:09 +02:00
parent 100760e219
commit 9865b14b0c
1 changed files with 27 additions and 26 deletions

View File

@ -3082,6 +3082,8 @@ it is called on. The input layer is returned and available for further processin
This method requires a polygon layer. It will apply a bias per edge of the polygons
and return the biased layer. The layer that this method is called on is not modified.
</p><p>
The alternative method <a href="#size">size</a> works like <a href="#sized">sized</a> but modifies the layer it is called on.
</p><p>
In the single-value form, that bias is applied both in horizontal or vertical direction.
In the two-value form, the horizontal and vertical bias can be specified separately.
</p><p>
@ -3109,32 +3111,6 @@ layer.sized(300.nm).raw.merged(2)
Bias values can be given as floating-point values (in micron) or integer values (in
database units). To explicitly specify the unit, use the unit denominators.
</p><p>
The "inside" option and the "steps" option implement incremental size. Incremental
size means that the sizing value is applied in n steps. Between the steps, the sized
shape is confined to the "inside" layer by means of a boolean "AND" operation.
</p><p>
This scheme is used to implement latch-up rules where a device active region has to
be close to a well tap. By using the well layer as the "inside" layer, the size function
follows the well contours. The steps have to selected such that the per-step size value
is smaller than the minimum space of the well shapes. With that, the sized shapes will
not cross over to neighbor well regions. Specifically, the per-step size has to be less
than about 70% of the minimum space to account for the minimum corner-to-corner case
with Euclidian space measurements.
</p><p>
"inside" and "steps" can be used with positive sizing values only.
</p><p>
"outside" acts like "inside", but instead of confining the sized region to the
inside of the given layer, it is confined to be outside of that layer. Technically,
a boolean "NOT" is performed instead of a boolean "AND".
</p><p>
An example for the "inside" option is this:
</p><p>
<pre>
ntap.sized(30.um, inside(nwell), steps(100))
</pre>
</p><p>
<a href="#size">size</a> is working like <a href="#sized">sized</a> but modifies the layer it is called on.
</p><p>
The following images show the effect of various forms of the "sized" method:
</p><p>
<table>
@ -3152,6 +3128,31 @@ The following images show the effect of various forms of the "sized" method:
</tr>
</table>
</p><p>
The "inside" option and the "steps" option implement incremental size. Incremental
size means that the sizing value is applied in n steps. Between the steps, the sized
shape is confined to the "inside" layer by means of a boolean "AND" operation.
</p><p>
This scheme is used to implement latch-up rules where a device active region has to
be close to a well tap. By using the well layer as the "inside" layer, the size function
follows the well contours. The steps have to selected such that the per-step size value
is smaller than the minimum space of the well shapes. With that, the sized shapes will
not cross over to neighbor well regions. Specifically, the per-step size has to be less
than about 70% of the minimum space to account for the minimum corner-to-corner case
with Euclidian space measurements.
</p><p>
"inside" and "steps" can be used with positive sizing values only.
A steps value of 0 will not execute any sizing at all.
</p><p>
"outside" acts like "inside", but instead of confining the sized region to the
inside of the given layer, it is confined to be outside of that layer. Technically,
a boolean "NOT" is performed instead of a boolean "AND".
</p><p>
An example for the "inside" option is this:
</p><p>
<pre>
ntap.sized(30.um, inside(nwell), steps(100))
</pre>
</p><p>
The effect of the "inside" option is shown here:
</p><p>
<table>