Bandgap voltage variation in different corner

[gn7623233]

Hi,

I have designed a bandgap, it works fine.
And different corners have different curvature.
One thing i wanna know is :
Why FF (SS) corner has larger (lower) voltage than TT corner at the same temperature.
Just like the figure 5.(C) https://reurl.cc/j741dn
Is it about the doping concentration influence the saturation current (Is) or other factor?
THANKS!

 

[d123]

Hi,

Link doesn't work for me. It opens an empty, white page...

 

[dick_freebird]

Resistance (resistivity) may be moved by the corners.
You'd have to look at the models chain.

Tempco tends to be proportional to resistivity. This I
think is also true of its curvature. Although a "digital"
process's modeling of "analog" features must always
be viewed with suspicion. A NWell resistor is a bad
choice for tempco curvature reason, as well as voltco.
Bandgap designers may like the high resistor TC for
compactness, initially - until its curvature bites them.

Thin film resistors (not poly) have near-zero tempco
and curvature. In modern mixed signal technologies
these tend to reside in the lower levels of the
interconnect stack, not on field oxide. However using
these may make your bandgap block incompatible
with the "low rent" flow variants, it's a process steps
adder that may be cut for cost.

 

[gn7623233]

Hi,

Link doesn't work for me. It opens an empty, white page...

I uploaded this picture again
THANKS!

Reference: https://ieeexplore.ieee.org/document/8252532

 

[t4_v]

I would look into the schematic and see voltages and currents to compare FF vs TT to find the reason. The bias current that biases badgap (opamp that is used in the bandgap) varies with corners and it affects the final bandgap voltage.