Attached is part of the circuit that includes the supervisor and the DC-DC switcher. As you can see I'm going to monitor the battery voltage. When it's 0.9 V - the battery is practically dead - I want the output of the TPS3808 U21 shuts down the switching power supply U23. Vcc is 3 V and generated by U23. When the RST pin of the 3808 is pulled down, it disables U23 and enables another power supply to guarantee VCC always be at 3 V.
I wouldn't mind to have a 0.84 threshold so that I could squeeze a bit more juice out of the battery. The thing is that, even though the circuits powered by Vcc always draw very little current (less than 2 mA), once in a while they may need a bit more juice (about 20 mA). Therefore, at voltages below 0.9 volts there probably is little juice or no juice at all... and the RST would toggle up and down often when that short pulse of current is requested from the battery. It's not exactly a problem because the other power supply backs it up but I would prefer disabling U23 to have a clean switch from one power supply to the other.
The TPS3808G09 has a Vit of 0.84 and the next is the G12. Is using two resistors the only way to get 0.9 V as the trip voltage? You said the G09 is typically used to supervise the 0.9 V rail. Do you advise me to use it as 0.9 V is its nominal supply voltage? Frankly, I would prefer a fixed value than having to use two resistors.
I bypassed the sense pin with a 10 nF cap as the manual suggests and used a 390 nF for a reset delay time of approximately 2 seconds. I still have the option to pull Ct up with the 100 K resistor if I need a shorter (300 ms) delay. Therefore, I can use one or the other.
As that 20 mA pulse of current will be less than 1 second, would the G09 be the correct choice? Looking at the timing diagram, it would switch back to the power supply U23 "long" after (1 second later) the voltage went above the 0.9 V, right?
Any other ideas or suggestions would be welcome.