I'm designing a simple circuit to use a lithium-metal coin battery as a backup power supply, and I'm considering something simple like this:


simulate this circuit – Schematic created using CircuitLab

However, during normal operation, \$V+\$ is expected to be at a higher voltage than the coin cell, so I expect some reverse current through D2. According to this source, this absolutely can damage the coin cell if it is too high - coin lithium batteries can't be guaranteed to work after subject to a total reverse charge of 3% total capacity, and a 1μA current will do that in a few months.

My question is, how can I find out how much current will go through D2? Most diode datasheets are completely uninterested at the currents going through the diodes when a small reverse bias is applied, mentioning only maximum reverse current when operating very close to the maximum voltage of the diode. Others have graphs like this, that show reverse current growing roughly exponentially, but don't show any details when close to the origin:

enter image description here

So my question is, is there a good model for how diodes behave when subject to a small reverse voltage bias? Can I have a better estimate of reverse current than simply using the maximum value?

(before you mention, assuming I take Shockley's diode law seriously, the reverse current would be close to the saturation current \$I_S\$ for a large range of negative voltages, but if that were the case I'd expect datasheet graphs as the one provided to look a lot different, and for \$I_S\$ to be mentioned in every single diode datasheet. Since this is not the case, I am assuming the diode law doesn't hold very well for reverse bias)

  • 1
    \$\begingroup\$ Your asking for a good model but what I think you really need is a diode with a specified maximum reverse current. A model is only as good as the parameter values used with it, and it will be very difficult to get the parameters that correspond to a worst case diode rather than a typical diode. Even the graph you show probably represents typical data rather than worst case data. \$\endgroup\$ Feb 22, 2019 at 21:02
  • 1
    \$\begingroup\$ What OP really needs is diode with a specified maximum reverse current given at low voltages, because lot of datasheet do not give that info. A model can help extrapolating the curves that are given to voltages close to zero volt. \$\endgroup\$
    – Huisman
    Feb 22, 2019 at 22:02
  • \$\begingroup\$ I was about recommending the BAS116 and one manufacturer even provides maximum values ... but at 75V ... No idea how much is it scaled down when the reverse voltage is 0.4V in the case of OP \$\endgroup\$
    – Huisman
    Feb 22, 2019 at 22:04
  • 2
    \$\begingroup\$ I know they were chosen for their low forward voltage drop but the OP should know that schottkey diodes have particularly high leakage currents. \$\endgroup\$
    – DKNguyen
    Feb 22, 2019 at 22:19
  • \$\begingroup\$ You might look at "ideal" diode parts like this one from Maxim. It does have at least typical values vs. voltage for the reverse leakage: datasheets.maximintegrated.com/en/ds/MAX40200.pdf \$\endgroup\$
    – John D
    Feb 22, 2019 at 23:15

2 Answers 2


One of the best simple models I've seen is here:

This shows that the reverse current quickly enters a predictable saturation region with mainly the temperature effects to be aware of. However there is a much higher resistive leakage current that impacts the total reverse current.

This diagram shows both Is (the saturation reverse current) and Il (the resistive surface and bulk leakage).

enter image description here

It's worth noting that the resistive leakage value can be permanently changed (increased) by overheating the device during soldering. This makes it very difficult to predict total reverse current in any simplified model. Resistive reverse leakage may be in the 10 - 1000 MOhms range.

Note: If your goal is to provide BBU for something low current like an RTC chip with a CR2302 battery, then you don't need to use diodes at all. You can use a transmission gate to switch from one supply to the other. For example the Philips 74LVC1G3157 is capable of up to 12mA with off state leakage below 1 uA maximum.

  • \$\begingroup\$ that's exactly the kind of model that I am looking for, but it doesn't seem to fit the roughly exponential increase in reverse current shown in the graph I provided. Any clues as to why and at which values the linear/exponential transition occurs? \$\endgroup\$
    – FrancoVS
    Feb 27, 2019 at 17:24

Make sure to factor in the self-discharge of the cell (maybe 1% per year) as well as leakage currents on the PCBA...you may very well be leaking that 1uA away elsewhere on the board. Another option is to put a tantalum capacitor in line with your cell. The leakage of that capacitor will be enough to ensure you don't damage the cell.


Your Answer

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

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