How would you write then a NMOS with m=2 in the ne...
# xycem
Murat Eskiyerli
08/05/2021, 4:29 PMHow would you write then a NMOS with m=2 in the netlist? @User
t
Tim Edwards
08/05/2021, 8:59 PMDisregarding the complication of fingered devices, you would want to specify "M=2 mult=2".
m
Murat Eskiyerli
08/05/2021, 10:07 PMStill need to check Xyce manual, but I am not sure if Xyce subcircuits automatically understand m parameter as denoting multiplicity.
Murat Eskiyerli
08/06/2021, 8:38 AMI checked and I can't find any reference to a default M parameter for subcircuits in Xyce. The way I changed the interface, a single MULT=2 parameter would be sufficient to define parallel MOSFET devices. Of course, adding a default M parameter to subcircuit definitions might have been quite useful.
t
Tim Edwards
08/06/2021, 1:07 PMParameter M for subcircuits is implicit.
m
Murat Eskiyerli
08/06/2021, 1:08 PMAre you sure? I can't find any reference in Xyce manual.
Murat Eskiyerli
08/06/2021, 1:10 PMIn any case, trying to keep track of two parameters is error-prone.
Murat Eskiyerli
08/06/2021, 7:03 PMThis the reply I got from Eric Keiter regarding multiplicity factor issue:
Unfortunately, Xyce doesn't yet support the m= multiplier for subcircuits. Xyce currently only supports it on the device level, in some devices.
e
Eric Keiter
08/12/2021, 7:59 PMHi all, sorry that I’m late to this discussion. I can confirm that Xyce does not currently support the m= multiplier for subcircuits. This is an issue that is in our issue tracker but hasn’t been resolved yet.
m
Murat Eskiyerli
08/13/2021, 8:39 PMHi Eric, I wonder if a netlist preprocessor could be the way forward including the array notation as in the Cadence virtuoso.
t
Tim Edwards
08/20/2021, 1:43 AM@Eric Keiter: Lack of M = for subcircuits is a serious problem especially for technologies like SkyWater where all the low-level devices like transistors are represented by subcircuits. Fortunately magic doesn't output M by default, so at least magic-extracted layouts should be okay. Everything else, though. . .
e
Eric Keiter
08/20/2021, 3:30 PM@Tim Edwards thanks for the info. This has come recently with our internal users as well for another PDK. So, we’ll have to fix this soon. It has been in our issue tracker since earlier this year, but wasn’t prioritized until now.
m
Mustafa Ahmad
10/11/2021, 11:33 AM@Amro Tork
a
Amro Tork
10/12/2021, 8:00 AM@Eric Keiter I was wondering when you would be able to have a version with m support for transistors? We are building a regression here to compare ng-spice and Xyce simulation results on basic circuits:
Amro Tork
10/12/2021, 8:00 AMe
Eric Keiter
10/13/2021, 5:12 PM@Amro Tork I believe Xyce supports the
M=a
Amro Tork
10/13/2021, 5:48 PMThanks @Eric Keiter. Appreciate all the help
Amro Tork
10/13/2021, 5:48 PMI’ll wrap a testcase for you to show that m seems to not be working.
e
Eric Keiter
10/13/2021, 5:48 PMthanks!
a
Amro Tork
10/13/2021, 6:50 PM@Eric Keiter Just to confirm here
Amro Tork
10/13/2021, 6:50 PMIs M multiplayer or divisor?
Amro Tork
10/13/2021, 6:50 PMImage from iOS.jpg
Amro Tork
10/13/2021, 6:51 PMLike it takes W and divide it by M or W total is W * M
Amro Tork
10/13/2021, 6:51 PM?
Amro Tork
10/13/2021, 6:51 PMJust to confirm
e
Eric Keiter
10/13/2021, 7:53 PMM is supposed to be a scalar multiplier. It should increase the size of the current coming out at each terminal of the device. So it effectively scales the area.
a
Amro Tork
10/13/2021, 10:55 PM@Mustafa Ahmad please read Eric note above
Amro Tork
10/13/2021, 10:55 PM@Eric Keiter thanks for your help. We will update the test case and send it to you.