LM393 Comparator input bias

at

Assuming internal workings as indicated in the schematic in the current data sheet available on the National Semiconductor web site: If the (+) input is unbiased, that side of the input diff pair won't conduct. That doesn't keep the other side from conducting, though. Don't expect fast response--there's nothing to suck base charge out of Q7 when the (-) input side turns off.

In general, it's bad practice to rely on characteristics that aren't guaranteed--or even implied by the actual specifications. (The simplified schematic in the data sheet should not be taken as a specification.) Seems to me you can get the same general performance while using the comparator in a more conventional biasing circuit, using the same number of parts (or likely even fewer parts), and the performance will be more stable over things like die and temperature changes.

Cheers, Tom

The comparator has pnp inputs requiring bias current from the negative rail. As a result, the non-inverting input of the second stage must eventually float high, preventing the output indication.

Quite lame, with undetermined response to initial power-on or short-term power removal.

It could only have been designed by accident, so you should probably consider the entire schematic to be suspect.

RL