most bandgap ckts contain positive feedback structure - in normal bias point, the open loop gain<1, the ckt is always stable. But during power up, the gain can be higher than 1, and thus some could latch up - that's when you find the ckt is not working (actually it has op pts while two are stable).
It has been shown by Nelson and Wilson, later by Nishi and Chua, that in order for a ckt to possess multiple operating points, a (positive) feedback structure must exist after one simplify ckt by shorting voltage source/open current source/short or open resistors.
That theorem leads to many augment methods to find multiple operating points numerically. Since no numerical method can guarantee finding all op pts, it's ckt designer's responsibility to check out the latch up condition: one must know how to detect multiple operating points through these methods; how to ensure only one op pt exists after adding leakage current. |