theoritically, the hystersis loop is determined by the W/L of transistors and can be calculated but the small loop u need is very small and I think it will be process dependent!
maybe u need to change the architecture of your system or the idea behind it!
but in single power OPA design
if my design range is Vh=2.55 Vl=2.45 ..
the OPA positive feedback resistor have large ration ..
I don't think OPA positiove feedback is good architecture ..
and schemitter is bad architecture ..
Schmitt trigger levels will vary dependend on power , because of feedback supplied from OPA or any other IC output - which is directly depndent on Vcc .
The simplest is to make stable power or put zener diode with resistor on OPA otuput and then use this fixed output to connect schmitt feedback resistor .
In addition you can set trigger detection point independently - 2 diodes and 2 feedback resistors will be needed .
or you can use 2 comparators (one per threshold ) and decoding logic but input noise could be a problem - capsitors will be necessary but transient time will be decreased .
no zener diode can be use in CMOS design ..
and I find main problem is my circuit act as power_on reset detect ..
and schmitt trig can work well if Vcc is "fix"
but my circuit have Vcc ramp up ..
I also use 1 Comp but hysteresis volt 100mv is too small . I think
maybe I need 2 comparator for work ..
or anyone can provide a power_on_reset circuit with hysteresis circuit for reference ..
If you just apply the schmitter trig in Power-On-Reset circuit,Why you need to design so small hysteresis range 100mv?
I think 40% of supply voltage is ok.
you can use one type comparator with hysteresis(use positive feedback).you can
adjust the W/L ratio to get the hysteresis you need.
you can find the comparator structure in any book that introduce comparator and opamp.hope this help you!