LM139A: Unused Pin Connections

Hello Juan,

It was originally recommended that the unused inputs be tied to V-, but that causes a possible issue:

If the comparator offset voltage is near zero, there could be conditions (over  temperature or supply voltage)  where the offset will cross through zero and cause the output to "chatter" on it's own noise. This can cause the output to randomly transition.

Since this is an open-collector output device, as long as the output does not have a  pull-up resistor, the output will not be 'flopping" around. However, it can cause very slight random changes to the supply current (<5%).

So now we recommend that the inputs be biased as to keep the output in a known state. So one input can be grounded and the other input tied to some other convenient voltage within the valid common mode range to force the output state.

In 99% of the cases, the above will not be an issue and you can just ground the inputs. The amount of devices "chattering" will depend on the offset distribution of the lot...maybe a percent or two. These are NOT precision devices...

The output is up to you. Leaving the output unconnected allows possible future "upgrades" to a totem-pole output devices and reduces possible output leakage current paths. One argument for grounding the output is reducing possible ESD susceptibility due to a "floating" pin.

Obviously, a totem pole (sink-source) output device should never be grounded and should be left floating. Have to say that...

So it is up to you...Do whatever is easiest for you knowing the above "risks".

Hello Paul,
This is extremely helpful and it all makes sense. I am sold on your idea and will force a known output by connecting the inputs to some specific voltages

One final question. If I tie the non-inverting input to a voltage lower than the inverting input then the output state will be forced to low. I could then leave the output unconnected. This would have the advantage of "reducing possible output leakage paths" while also reducing "possible ESD susceptibility due to a floating pin" because the output would be both unconnected and also "driven" low.

Is this basically a way of having my cake and eating it too? I would rather have no floating pins, and it would be great if I am not burning any extra power with a pullup. It looks like I can't have both if I force the output to be high impedance (non-inverting higher than inverting) but looks like I can get both the other way around?

Hopefully that was clear! Thanks,
Juan

Hello Juan,

Yes. You can force the output low, but I really doubt it will help for a direct ESD strike. ESD zaps usually happen during board handling when there is no power applied -so the output will not be "low". You have other bigger problems if ESD strikes are getting that deep into your board to reach an unused comparator pin...

Let the output float.... No current can flow through an unconnected pin. And as I said - it also allows you to "upgrade" later if you need to replace/swap with a R-R out device.

In fact - if you are in the prototyping phase - you can make the traces to the inputs available to cut so it is available as a "spare" comparator should it be needed during prototyping for future "unforeseen needs".