LVS-related question: For the IHP PDK, what is the...
# ihp-sg13g2
b
LVS-related question: For the IHP PDK, what is the suggested way to deal with multiple VSS nets? For example, say I have VSS and VSSA, both of which connect somewhere in my layout to substrate via taps (thus creating a short between the two nets). Some PDKs offer an "isosub" layer to mask out e.g. the substrate taps from VSSA. Others allow you to specify which nets can be shorted via the substrate (I like this better). What is the proper approach for the IHP PDK? @Krzysztof Herman @Amro Tork
k
@Boris Murmann since we do not have any isosub layer I guess that the second approach would be the choice however let me consult.
a
@Boris Murmann I had a question just to clarify, you won’t put one of them in Deep NWELL, both will be connected to p-sub?
*or
b
I haven't considered using deep n-well, and our layouts are not done that way right now. This is an option that is also needed for other VSS nets, e.g. VSSIO, regardless of having deep n-well.
a
Ok
Will see if klayout would support such connection
b
From the KLayout documentation, I see that there is a soft connect command. This may do the job.
According to this post, Calibre uses the same soft connect function to ensure that a bulk tie does not associate the substrate exclusively with one node (thereby allowing multiple nets to have substrate taps): https://www.edaboard.com/threads/stamping-soft-connection.334208/
a
Yes, I’m familiar with how Calibre does that
b
It looks like the soft-connect feature of KLayout was just released recently: https://github.com/KLayout/klayout/issues/1598, requiring version 29.0.
a
Thanks Dr. @Boris Murmann, I wasn't really sure. I was going to look at today. Thanks for all the help.
h
@Boris Murmann @Amro Tork @Krzysztof Herman In IHP PDK there is a method to split the substrate (but well hidden, because it is a bit ill-named):
DigiSub
is the layer you want to use.
In related news, an area of digital standard cells needs to be surrounded by
DigiBnd
to activate the tighter DRC rules.
a
Thanks Dr. @Harald Pretl for letting us know. I don't think we have implemented in DigiSub in Klayout LVS. Let us check with IHP team.
h
@Amro Tork Please implement it, because that is exactly what is used in the closed PDK with the other tools.
a
I have to get confirmation from IHP team. Thanks for the feedback Dr. @Harald Pretl . Appreciate if you could open an issue on the PDK as well.
h
@Amro Tork My pleasure, here you go: https://github.com/IHP-GmbH/IHP-Open-PDK/issues/86
a
Thanks Dr. @Harald Pretl
b
Thanks @Harald Pretl. Will this allow you to split the substrate just into two areas or more? For example, I have VSS, AVSS and IOVSS connected to local taps and need to mask out the "soft short" that this creates for the LVS.
a
Dr. @Boris Murmann Yes, it will mask out the "soft shorts" based on Dr. @Harald Pretl requirements once we implement that behavior in LVS.
Will implement it soon.
h
@Boris Murmann It splits the substrate into as many areas as you want (each area inside a
DigiSub
is a substrate island). Once it is implemented by @Amro Tork 🙂
🙌 1
Hence the confusing name
DigiSub
— it is not only for digital 😉
a
We could request a change in name from
DigiSub
to
subiso
To make it more intuitive.
@Krzysztof Herman Could we do this?
h
@Amro Tork I would like to keep the name although confusing, because in this way at least it is consistent for all IHP SG13* PDKs (open and closed). @Krzysztof Herman
a
Ok
b
While we're on this subject, what is the intended use/purpose of the substrate layer? I see it in some of the existing IHP cells, but not in others. It is clearly just an LVS layer, but I wonder how it is supposed to be used.
a
Dr. @Boris Murmann and Dr. @Harald Pretl Digisub and Burried Layer isolation are now implemented in the LVS. Wanted to keep you posted. cc @Krzysztof Herman @Sergei Andreev
🙌 1
b
Thank you!
a
Anytime Dr. @Boris Murmann