"Diode OR" with higher battery voltage - suggestions?

[gxti]

Show me your cleverest circuit for drawing power from a 12V source when available, or a 24-28V battery backup otherwise. The output will go into a 34063 based switching regulator, which will consume less than 0.5W.

Here's my first attempt. It works (in simulation, at least), but it's not very much fun to look at and could use a few less parts.


ltspice file: http://partiallystapled.com/~gxti/circuits/2012/12/16-diode_or_1.asc

[PuterGeek]

My immediate smart ass reply was a toggle switch. 

But a SPDT relay would work pretty well. The 12V supply powers the coil to switch between battery and 12V. You would only need three parts; relay, spike suppression diode (across the coil) and a Schottky so only 12V can power the coil.

Quick question, why are you using such different voltages?

A lower battery voltage or increasing the supply voltage would allow charging the battery with a little more circuitry.

[c4757p]

I like the relay idea, personally.

[gxti]

I actually am charging the battery, using a boost converter. The load in this case is the microcontroller that supervises the whole system. The 12V is from an ATX power supply which is used because it's cheap and more efficient than anything I could hope to build. The battery is 24V so I can use a simple boost converter to charge it and a simple buck converter to power the 12V load from battery. If I used a 12V battery then I'd need a buck-boost or SEPIC converter to get charge in and out because the battery voltage can be higher or lower than 12V.

I will probably use a relay since I need one anyway. But if there's a simpler solid-state solution than what I posted I might still be interested. But what do you mean about "a Schottky so only 12V can power the coil" if the coil is only connected to the 12V input anyway?

[onlooker]

Try this on.

[PuterGeek]

I actually am charging the battery, using a boost converter. The load in this case is the microcontroller that supervises the whole system. The 12V is from an ATX power supply which is used because it's cheap and more efficient than anything I could hope to build. The battery is 24V so I can use a simple boost converter to charge it and a simple buck converter to power the 12V load from battery. If I used a 12V battery then I'd need a buck-boost or SEPIC converter to get charge in and out because the battery voltage can be higher or lower than 12V.

I will probably use a relay since I need one anyway. But if there's a simpler solid-state solution than what I posted I might still be interested. But what do you mean about "a Schottky so only 12V can power the coil" if the coil is only connected to the 12V input anyway?

You can get 24V from the ATX supply by using -12V (pin 14) for ground and +12 (pins 10 & 11) for power.

Are you using a charge control or just connecting the converter to the battery? Most batteries last longer if properly fed.

[gxti]

Thanks onlooker, looks great. I'd leave the top diode in though so that the 12V rail won't source current to the 28V input if it is disconnected or fails short or something.

As for ATX, the -12V rail is legacy and usually supplies less than an amp max. The +12V rail is where all the muscle is. In fact, many "enthusiast" supplies have a single 12V rail as the output of the main converter, and then hang smaller buck converters off that to make all the other ATX voltages. Motherboards and video cards draw the bulk of their power from the 12V rail and convert it to the needed voltage closer to the load. In any case, 24V still isn't enough to charge a 24V lead-acid battery.

Yes, the boost converter is controlled. I have voltage and current feedback, and the voltage is tapped from separate wires right to the battery terminals. I'm planning on doing "hysteretic" charging where it turns on until the voltage reaches a high-water mark, then shuts off until it hits a low mark. But it will be current limited both during a fast charge and float charge to keep the battery happy.