gf180mcu 5V/6V devices

[d-m-bailey]

@RTimothyEdwards :
Question about the recent update to differentiate 5V and 6V devices.

In gds, the V5_XTOR/FET5VDEF 112/1 layer differentiates between 5V and 6V devices, but when drawing in magic, the default is 5V, correct? mvndiff + poly -> mvnfet. To change the device to a 6V device, each device must be redrawn with hvnfet correct? I suppose most people will draw the devices with the device generator, so it's probably not a big issue.

Also, the decompose statements for the 6V devices seem a little off.

  decompose hvnfet poly mvndiff
  decompose hvpfet poly mvndiff

Should the hvpfet diffusion be mvpdiff?

Also looking at 486fb8633c3d95ae475a5ccd68d0066c25919bb5 I can't find the layer derivation for hvnfet. All poly/mvndiff devices are extracted as 5V whether or not there is FET5VDEF, I think.

[RTimothyEdwards]

(1) Magic will read in 5V and 6V devices correctly now, although it is possible that some devices in the PDK libraries are wrong because magic read them using the earlier PDK.
(2) Painting poly over mvdiff will result in a 5V FET, so it is "default" in that sense. However, if you use the generated devices, both 5V and 6V options are available.
(3) Yes, the hvpfet diffusion is mvpdiff. There is no hvpdiff because there are no special rules (that I am aware of) that distinguish between diffusion around 5V and 6V devices. The only relevant DRC rule appears to be the restriction on gate length, so I only need a special layer to distinguish the 6V FET gate for extraction and for the gate length rule check.
(4) I think you are referring to the cifinput section, where I see that there is no rule to read 6V nFETs, so I will correct that. I think that's the only missing definition, and hvnfet is represented in all other sections (cifoutput, extract, drc, etc.).

[RTimothyEdwards]

The fix for hvnfet in cifinput will go in the next commit.

[d-m-bailey]

@RTimothyEdwards Just to be clear, the current decompose statement for hvpfet is

decompose hvpfet poly mvndiff

It should be mvpdiff instead, right?

[RTimothyEdwards]

Oh, right. Didn't see the typo.