I am looking for the general approach (technique/strategy, componentry used, etc.) to implement what I am calling a power supply router, although perhaps there is a more precise/accurate term for it.

By "power supply router" (PSR), what I mean is this:

  • A device, say some PCB (doesn't really matter) will connect to this PSR and expect it to be its power supply, but...
  • The PSR itself is not an actual power supply; it is simply a circuit/component/thing that uses a simple routing algorithm to draw power from one of several actual power supplies at any given time
  • In other words, the router will strive to always provide power (if the device is toggled to be "on" via power button, etc.) from one of many power supplies, and can switch the actual power supply it is using at any point in time (only during operation of course) according to this algorithm

And a diagram for illustration:

enter image description here

In that diagram Power Supply #1 (PS1), #2, #3, #4 and #n are the actual power supplies (batteries, solar panels, whatever). The device connects to the router, expecting it to supply it with power, but the router will only be connected to one PS at any given time.

The algorithm I am looking to implement is very simple:

  • For all the power supplies in the system, find the first supply that isn't "nearly dead" and set it as the current supply
  • Once the current supply is nearly dead, obtain the next non-nearly dead supply and set that as the current supply
  • Only when all power supplies are dead/nerly dead is the system officially out of power

I'm sure the "nearly dead" part requires better explanation. Here, the router simply needs to be able to determine whether the given supply is almost dead, perhaps based on the amount of volts or amps its currently outputting, or based on some gas gauging component like what cell phones use to show you that your smart phone is at "2%".

I'm wondering if this type of "routing multiple power supplies" need is a well known pattern in EE, with well known solutions? Either way, what are my options here for implementing such a router & routing algorithm as a circuit? What components could be used?

  • \$\begingroup\$ How about just diodes in series with each power supply? Simple, and inexpensive. \$\endgroup\$
    – R Drast
    May 6, 2015 at 13:19
  • \$\begingroup\$ vishay.com/docs/70786/70786.pdf there are multiple discrete solution app notes available, ranging from just diodes as @RDrast said. There are more complicated active switching circuits. TI also has some nice power management IC solutions. \$\endgroup\$
    – scld
    May 6, 2015 at 13:52
  • \$\begingroup\$ Thanks @RDrast - any chance I could twist your arm into explaining why diodes wired in series would suit my needs? Thanks again! \$\endgroup\$
    – Zac
    May 6, 2015 at 14:01
  • \$\begingroup\$ If you put a diode in series with each power supply positive lead, then only the supply delivering the highest voltage will power the load. If that voltage drops, the supply with the next highest voltage will power the load, etc. I use that technique all the time around the plant here to use backup batteries on our 24VDC supplies. If the line power drops out, the backup batteries seamlessly supply power. When the line comes back, the powersupply silently takes over again. \$\endgroup\$
    – R Drast
    May 6, 2015 at 14:16
  • \$\begingroup\$ Thanks @RDrast - last follow up question regarding this solution, I promise! So are you saying to "chain" these supplies together, in other words: PS1 -> Diode1 -> PS2 -> Diode2 -> ... -> PSn -> Dioden? Or is it still like my diagram above, where each PS->Diode combo routes into the circuit at some common connection, but does so from its own distinct path? Thanks again! \$\endgroup\$
    – Zac
    May 6, 2015 at 14:27

1 Answer 1


Something like this. Of course, I am assuming that all of your supplies are the same voltage; all 12V, or 5V, or something. And this obviously won't work for AC. The way to select precedence, is to make your primary supply the highest voltage (not by much), and then step down a tenth or quarter volt for every other supply. Take into account the diode forward drop.


simulate this circuit – Schematic created using CircuitLab


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.