I am trying to design a circuit to replicate the behavior of the LTC4071 battery charger/protector IC (see datasheet here). In particular I want to replicate the low battery disconnect function which consumes 500 nA when running and only 0.1 nA when in battery disconnect mode. I have tested the LTC4071 on a dev board and found that it behaves as described. My issue is that the package is to large for my needs and so I require a custom circuit which can be made with lower profile components.

Below is the LTC4071's block diagram and circled in red is the low battery disconnect functionality. Seems simple right? When VCC is connected to an external power supply the comparator outputs low which turns on the p-mosfet MP1 connecting the battery to VCC. When the power supply is disconnected the battery continues to drive VCC. If the battery level drops low enough for the comparator to output high then the p-mosfet turns off and the battery is disconnected from VCC.

LTC4071 block diagram

However when I mock up an equivalent circuit on a bread board using a low power op amp (MCP6241) and p-type mosfet (IRF4905) I do not get the required results, this simulation describes the problem.

Simulation diagram

Here, Vbat is below the battery disconnect threshold and so the op amp's V+ is greater than V- so its output is high which should turn off the p-mosfet and hence disconnect the battery from the system. However, this doesn't happen since as the op amp's supply voltage decreases so does the output voltage which in turn partially turns on p-mosfet until the system reaches equilibrium at an intermediary voltage.

I could connect the op amp's supply directly to the battery, rather than to Vcc, but then the op amp's quiescent current would constantly draw from the battery. I can't find any op amps or comparators that draw less than 100 nA so how did the IC designers get this to work using just 0.1 nA!?

  • 4
    \$\begingroup\$ If a 3x2mm DFN is too big for your application, there is no way rolling your own version with discretes is going to fit either. \$\endgroup\$
    – vir
    Commented Nov 9, 2022 at 22:15
  • \$\begingroup\$ @vir You're probably right and I will likely end up using the LT4071 anyway, but I'd still like to know how they do it! Unless of course it's some crazy secret IP they have, but the fact that the datasheet includes the system diagram just makes me think it must be possible! \$\endgroup\$
    – John
    Commented Nov 9, 2022 at 22:21
  • \$\begingroup\$ Did you see on the block diagram that the reference voltage is connected to the inverting input of the op amp? \$\endgroup\$
    – vir
    Commented Nov 9, 2022 at 22:27
  • \$\begingroup\$ I did notice that but I actually assumed that was a diagram error, I'm not sure how it can work the other way round. Unless I'm completely misunderstanding something :/ It's safe to say my circuit doesn't work the other way round either... \$\endgroup\$
    – John
    Commented Nov 9, 2022 at 22:30
  • 1
    \$\begingroup\$ The comparator in the LT4071 is not a "simple" comparator it's got some schmitt trigger functionality included (as evidenced by the symbol in the middle). \$\endgroup\$
    – brhans
    Commented Nov 9, 2022 at 22:46

1 Answer 1


After some more thought I think this is what might be going on.



The above now works completely as expected. I will test this out on a breadboard soon.

It looks like the datasheet might have in fact had the inverting and non inverting inputs the right way round but had left out a crucial n-mosfet which is used to actually turn on the p-mosfet.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.