open-source-silicon.dev

Current mirror script from the meeting today:

Hi, thanks for the explanation in the video from yesterday. I have a question, when you define the routes, you don't need to specify the metal that you want the route on? The library chose the most suitable one? If that is true, I assume that in a more complex design you don't need to worry about routing because it is ""automatic""

By automatic, I mean by default GLayout when routing selects the metal layers of the ports that you want to connect

If the metal layers are the same as in the case of the current mirror example yesterday, then the route will be the same metal layer as both the ports

If the metal layers are different, then straight route with automatically place via stacks to route

But thats as far as the automatic aspect goes, the routes don’t go as far to prevent interfering routes or automatically choose routing layers to prevent coupling parasitics, its up to the programmer to deal with that

Okay, I assume that once I design a more complex circuit I will see how to manage that situation with Glayout. Thanks

Hi again :sweat_smile: I ask you because maybe you know the reason of the issue. I have been ""redoing"" the code step by step to be sure that I understand all. However I've seen something strange:
While this works perfectly (when none of them have a dummy cell), if one has a dummy and the other one they overlap (The only thing that I have changed in the code is the False for a True in the dummy). Why is that? Thanks

Hi Eva,
movex only moves them by a certain amount of units in the x direction. The previous examples worked because the components had the same dimensions on the x axis, However, on the example you are having issues on they have different dimensions. When you place two transistors, they are both centered at the origin

You have to add the distance from the east edges of the components to what you are currently passing to movex