Improvements for the Tiny Tapeout analog switch: <...
# analog-design
u
@Tim Edwards moving the discussion here, so Sylvain will also be able to participate
Tim's last suggestion was:
You need the switch gated by 3.3V, but you can put the switch circuit next to the pad and wire it up to 3.3V manually without requiring openlane to deal with it.
The problem we're trying to solve: any accidental connection between the analog pins and 3v3 could kill the chip
t
Well feel free to do it, but I don't have time for that, it's not a trivial change and we don't even have a 3v3 rail anywhere close to those IOs yet. And there is no space for that so this all need to fit under the PDN rings basically without interfering with the routing path.
If the issue is just "user error" and it will work fine if the user keep things under 1.8V then I just see that as a "nice to have".
u
I was suggesting, as a workaround, to put solder bridges that connect the analog pads to GND
(in the breakout board)
so user error isn't easy to make - you first have to remove those bridges before you can make a user error
t
I would definitely put weak pull downs on the break out board at least.
u
I think what Tim suggested was to get the 3v3 directly from the pad next to the switches, so it won't need another core ring
t
what pad ?
u
perhaps something like this:
message has been deleted
red boxes running on Metal 5,
t
I'm not sure what this is supposed to show.
And where are those box connected to ?
u
and they get their power from vddio
t
how ?
u
(which is supposedly available somewhere on the pad)
t
It's not exposed to us.
So yeah, it's "somewhere" but we have no clue of where and how to get there.
Pretty sure this would also make all tests fails as we would ahve metal out of the allowed area.
Feel free to play with it if you have spare time, but as I said, I don't have time for that, especially when the proper solution is just to add the 3v3 support, adding 3v3 PDN and using 3v3 switch and that will solve the issue by itself.