https://open-source-silicon.dev
Join Slack
<@U016EM8L91B> Hi Tim, I have a question regarding the `sky130_fd_pr__res_high_po` (i.e. `poly resis...
# magic
m
@User Hi Tim, I have a question regarding the 
sky130_fd_pr__res_high_po
(i.e. 
poly resistor - 319 Ohm/sq
option in the 
Devices 2
in Magic menu). How do you calculate the resistance of the device. For example when I choose the 
sky130_fd_pr__res_high_po_0p69
device and choose 
0.690 um
both for the W and L (i.e. 1sq), then according to the table the resistance should be 
738 Ohms -> NOM
. But it is 
375.417 Ohms
in Magic. Would you mind help me figure it out? Overall, I think I’m confused about how to calculate the resistance of an element (layer).
a
R = L / W* Rtho
But I think proper resistances is listed on the table on the website. Looks like the _po does not have Rtho, and uses specific values.
m
@User So the question is how to find those “specific values”?
t
There is a fairly large resistance (~20 ohms) associated with the end contancts on the device, so while the body of the device is 320 ohms per square, the total resistance will be higher. I do not know how you came up with 738 ohms; for W = L you have one square, so it should be 320 ohms +20 + 20, or about 360 ohms (these figures are approximate).
m
@User Please find the attachment. I got the value from this table row #7. You can find it here. But how did you find the contacts’ and P+ poly’s resistance? I searched the whole “Device Details” part of the documentation for the 20 and 320 Ohms and found nothing.
t
The sheet resistance of the P+ poly is given in another table. The value for the terminal resistance in magic is wrong, as it should have been taken from the Rend value in the SPICE model multiplied up by the width in microns (and divided by 2, since Rend includes both terminals). Which means that is should be around 194, so it looks like a decimal place got moved. I will fix this in the .tcl file.
Note that there are still discrepancies with the table you cited, since the resistor equation used by the device generator is relatively simple and does not take into account 2nd order effects that are in the model. If you extract the layout, however, you will get a correct value for the drawn W and L, which if used in simulation will be passed to the complete model and give you a realistic simulation. Ultimately it's the simulation that counts.
m
Thanks a lot for the clarification @User. Currently the SPICE’s result is about 0.5KOhms more than the Magic’s (i.e. what it calculates in the 
Value
field). Is it what you expected? I’m using 
res_xhigh_po_0p35
device with 
W=3.50um
. And one other question, why 
paint res0p35
does not draw anything? Do I need to set anything else (other than the box)?
Open in Slack
PreviousNext
Powered by

What is Linen?

Linen is a search-engine friendly community platform. We offer integrations with existing Slack/Discord communities and make those conversations Google-searchable.

Powered by