How can I know whether a start-up circuit is needed or not

[CHL]

Hello
I wonder how I can know if a start-up circuit is needed or not intuitively (for example, in a BGR circuit). Although textbooks explain there are two conditions that current in a circuit can be zero or some amount, I don't understand how to check it intuitively. Can you explain the way to think about a start-up circuit?

Thanks

 

[erikl]

Not quite intuitively, I think: Circuits which can have two different operation point (OP) conditions, have two feedback loops: a negative feedback loop, and a positive one. In equilibrium OP condition, the positive loop gain must be less than the (absolute) negative loop gain (usually < 1).

You'd have to find out, if two such feedback loops exist:


If so, you need a start-up circuit.

 

[dick_freebird]

If you want your career in circuit design to last more
than one development cycle, "assume the worst" are
words to live by.

If you see any signs of bimodality, DC convergence failure
or any not-entirely-consistent behavior, "worst" is there
in front of you and does not even have to be assumed.

 

[dick_freebird]

You'd have to find out, if two such feedback loops exist:

If so, you need a start-up circuit.

I have seen 4-transistor PTAT loops (only one feedback)
exhibit 3 stable states - "off", "as planned" and "latched".

An "off" bangap w/ D/N*D+R pair can exhibit a lot of scatter
in functionality without a startup because leakage balance
determines the difference voltage and defectivity / mismatch
can be the deciding factor rather than the expected current
/ sizing ratio.

This might not be exposed until you get to Monte Carlo
analyses where such effects are laid onto the ideal circuit.