1
\$\begingroup\$

I see many circuits online of dual power supplies connected in parallel through output diodes on their outputs, to provide redundancy in case the main one fails. Usually the primary supply is an AC-DC adapter and the secondary is a battery backup.

I'm designing a 12 V UPS for my router but am not sure why a diode is needed in series with the primary voltage source.

I understand that the diode in the secondary supply rail is needed to prevent the main 12 V supply feeding into it. In my case, the primary rail is a simple mains-to-12 V DC adapter. I don't see any issue in the battery backup "feeding" back to that if it loses power.

This is my design so far. It's missing some capacitors on the load output. Note that the diode I'll use won't be a 1N4148, that's just the default Circuitlab had when I drew it.

enter image description here

As I understand it, the primary 12 V supply will take priority because the voltage on both sides of the diode will be 12v. With 0 V across the diode, it won't conduct and the main supply will power the load. If the primary 12 V system loses power, then the voltage at that point drops and the diode begins to conduct, allowing the boosted-up backup battery to power the load. With a diode voltage drop of, say, 0.5 V this would give 11.5 V.

In this case it's also "backfeeding" into the main primary supply, but I don't see the danger this causes. I'm wondering if I can skip the diode on the primary voltage source in order to avoid the voltage drop that comes with it.

\$\endgroup\$
10
  • 5
    \$\begingroup\$ Yes you can skip this extra diode in series with V1, but then ask yourself this question: what happens if V1 fails as a short-circuit, ie: the V1 output shorts out to GND? \$\endgroup\$ Commented May 1 at 21:45
  • \$\begingroup\$ Like Fabio already said, a shorted output is a reasonable fail mode so now your screwed when you need it most. Also, it's good practice even if the main power supply never fails but just becomes unpowered - it will act as a light load to your battery supply. It's mains powered; why care about the trace power loss from the diode anyway? The diode is a dime a dozen and the cheapest insurance that's going to protect your expensive batteries and boost controller. \$\endgroup\$
    – MOSFET
    Commented May 2 at 0:27
  • 1
    \$\begingroup\$ If you don't know how the primary voltage source is design, don't assume it can handle reverse voltage. The most common failure mode ought to be Vin to the primary source missing (due to a power outage etc) and then you end up in a situation where it has Vin=0V, Vout=12V. Voltage regulators tend to break when that happens. \$\endgroup\$
    – Lundin
    Commented May 2 at 9:19
  • \$\begingroup\$ (And obviously don't use 1N4148 but I take it this is just a "pseudo schematic" and in the real application you'll rather use a Schottky which can also handle the current.) \$\endgroup\$
    – Lundin
    Commented May 2 at 9:21
  • \$\begingroup\$ @Lundin, "And obviously don't use 1N4148 but I take it this is just a "pseudo schematic"". Had you read the question through before posting? OP already clearly stated "Note that the diode I'll use won't be a 1N4148, that's just the default Circuitlab had when I drew it". \$\endgroup\$
    – TonyM
    Commented May 2 at 9:28

2 Answers 2

1
\$\begingroup\$

If you assume the power supply tolerates being back-fed, you might as well assume the boost regulator tolerates being back-fed too, and no diodes are necessary. But this is not the case.

You need two diodes because you can't back-feed voltage/current into a powered off-supply. Or you can, but the power supply may get destroyed.

You seem to assume a powered-off supply somehow disconnects itself from the output and does not care if something else keeps the output at 12V (or 11.5V) when mains input to power supply is off.

While some power supplies may tolerate that, some might even be especially designed for it, but some power supplies will not tolerate it and get damaged.

Some power supplies may even have a mechanism to try discharging the output voltage to have a faster and definite turn-off ramp instead of slowly decaying voltage.

You don't know beforehand if the power supply tolerates being back-fed or not.

Therefore you need to assume it does not handle being back-fed, just like you are assuming your battery powered boost converter does not tolerate being back-fed with 12V.

\$\endgroup\$
0
\$\begingroup\$

Short: They are not needed

Longer: If you look at the schematic of most common switching mode supplies there is a already a diode in the output, so backfeeding isn't even that big of an issue.

It's a question of what failure mode you want to protect from and what your primary power source is tolerating.

schematic

simulate this circuit – Schematic created using CircuitLab

This is a simplified schematic of a typical buck converter commonly used in many power supplies. What you don't see is the implicit reverse Diode in the MOSFET M1.

So if you feed in Voltage at M1, U1 gets a signal on FB (which could prevent it from stating up again if it just has had a paower failure on the primary side). We're also backfeeding into Vin via the implicit reverse Diode in M1.

It can work, it can be no issue at all. It can be no iddue for a few times or under certein circumstances.

Backfeeding in a power supply is normally not a case that's defined in the datasheet. So the device can do whatever it wants if it happens, it's an undefined state.

So putting in that second diode in (it can be a shottky diode or an "ideal diode" if you don't want the voltage drop) is generally just the safe option.

\$\endgroup\$
3
  • \$\begingroup\$ "What you don't see is the implicit reverse Diode in the MOSFET M1" Wouldn't the body diode of an N channel have its anode against Vout? So how exactly is that one doing you any good? You'd need a P channel with drain against Vin and source against Vout like a reverse polarity protection circuit. \$\endgroup\$
    – Lundin
    Commented May 2 at 9:16
  • \$\begingroup\$ This diode wouldn't do you any good. Maybe I should phrase that differently. This Diode can mess you up badly because it might power the "primary side" of the converter making it want to power up again... \$\endgroup\$
    – kruemi
    Commented May 2 at 9:59
  • \$\begingroup\$ By "Ideal Diode" i assume you mean some MOSFET - but looking online at other similar circuits I haven't found a way that people use this without also using a diode on the output. \$\endgroup\$
    – QuickishFM
    Commented May 2 at 15:12

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.