Critique my circuit: Poor man's UPS

I have a project that I want to keep running for at least a few hours if the utility power fails. My initial thought was to use a UPS unit designed for a computer, but that's relatively expensive, and it turns out that it doesn't run all that long, I get an hour or so with APC's smallest unit. I imagine they're optimized for larger loads and shorter periods of time.

So I came up with this idea. I've only done initial testing; it seems to work. The next test will be to see how long it will actually run the circuit.

Would appreciate any input from the forum.

Circuit theory: With power supplied via J1, Q1 is kept saturated, pulling Q2's base to +9V and preventing it from conducting. Without input power at J1, R2 pulls Q2's base to ground, causing it to conduct and power the load from the battery. Q2's VCE was < 50mV when powering the load from the battery, so I was pretty happy with that. Didn't have any Schottky diodes handy, but I may try one for D2, just to squeeze a little more out of the battery. Maybe D2 can be eliminated altogether.

Thanks in advance!

If you're OK with the diode voltage drop already, then those two diodes alone are enough to do what you need.

macegr:
If you're OK with the diode voltage drop already, then those two diodes alone are enough to do what you need.

I should have mentioned, the wall wart I'm using actually supplies 400-500mV less than six fresh AAs. So with the two-diode approach, even with the wall supply on, the load would be powered from the batteries until they drop below the wall wart voltage. Not sure how much of the battery's capacity that would represent, but I wasn't sure I wanted to lose that.

I'm considering redesigning the load circuit to operate at 3.3V instead of 5V. Then I can use 4 x AA cells, and if I stick with the same wall wart, then the two-diode approach is a winner.

If voltage drop across the diodes is a concern, you could also go the mosFET approach. At least that's what I remember the folk over at picoUPS do.

That pre-engineered solution may be better (even if it's oversized) since it'll also charge a 12V battery (i.e. set and forget). You get 12V out, could use a switchmode VR from dimension engineering to deal with the Vdrop after that. That's the route I'm going to take on a phone base station in my house that will also serve as a USB phone charger.

Why not use some recharge-ables and leave them in the circuit all the time. At least then you won't have the system failing because someone forgot to change the batteries. Then you power supply just has to be regulated to the proper voltage for the batteries. NiCads peak out at approx 1.25 V/cell and run through a good part of their discharge cycle approx 1.2 V/cell so you coud regulate to 5V and supply that 5V after the regulator, straight to the arduino and the batteries.

Constantin:
If voltage drop across the diodes is a concern, you could also go the mosFET approach. At least that's what I remember the folk over at picoUPS do.

Interesting link, thank you!

kf2qd:
Why not use some recharge-ables and leave them in the circuit all the time. At least then you won't have the system failing because someone forgot to change the batteries. Then you power supply just has to be regulated to the proper voltage for the batteries. NiCads peak out at approx 1.25 V/cell and run through a good part of their discharge cycle approx 1.2 V/cell so you coud regulate to 5V and supply that 5V after the regulator, straight to the arduino and the batteries.

I was barking up that tree before I hit on the idea of plain old alkaline cells. I'm not up on the finer points of recharging NiCds or NiMHs etc. So I looked into various ICs designed for the job, wasn't sure I found anything entirely suitable for properly maintaining the battery while it was in-circuit. At any rate that seemed to be a more costly alternative. Alkaline cells have advertised shelf lives of seven years these days. Changing them out every 2-3 years is a pretty nominal cost, and if there is an extended outage, I'm hoping that will serve as a sufficient reminder to replace them!

Hmmmm, the MCU could track battery replacement date and cumulative usage...

Hmmmm, the MCU could track battery replacement date and cumulative usage...

and voltage? If they drop below a set value the Arduino can warn you before the backup is too low to be useful.

Pete

el_supremo:

Hmmmm, the MCU could track battery replacement date and cumulative usage...

and voltage? If they drop below a set value the Arduino can warn you before the backup is too low to be useful.

Pete

Of course! Well, it'd be an open-circuit voltage from the battery. Not sure how useful that'd be. Of course could also give the MCU the ability to either force a switch to the battery, or to test them with a load. But that's more complexity sigh ... does sound fun, though!

Well, it'd be an open-circuit voltage from the battery. Not sure how useful that'd be.

With alkaline batteries, if you stick with one brand, model, and size I suspect you will get reasonable results. I have#. But, to get a "percent life remaining" you will have to develop a "battery profile".

# I recommend Rayovac. If you're interested, I'll get the model and size I've been using.

When it's convenient, I'd appreciate that info. Not sure if I'm going there yet for this project, but would be interesting to know for future reference!

Let's see ... there it is ... Rayovac 815 (AA). The datasheet is here...

http://ca.rayovac.com/technical/pdfs/pg_battery.pdf

They very consistently start life at just above 1.6 volts, drain quickly to about 1.5 volts, drain slowly to 0.9 volts, drain quickly to 0.8 volts, at which point they are essentially dead. The "long drain" (1.5 to 0.9) is a smooth curve. I suspect the curve can be accurately estimated with three to five points. The open-circuit voltage has been consistently a bit above the closed-circuit voltage (I recall it was 0.05 volts; the test device uses about 10 ma). I use this technique for measuring the battery voltage.

Around here, you can get them in large consumer packs; my last purchase was a pack of 36 cells. I've had better luck (total life and consistent profile) with the Rayovac 815 batteries than with the others I've tried (I recall testing four other brand / models).

Thanks very much! Sounds like you've really done the homework! I've tended toward Duracell or Energizer, but not for any particular reason. Rayovac tends to be a little cheaper around here, but not as common. I'll have to keep my eyes open. Decent datasheet, I did check out Duracell's at one point, don't remember it being quite as good. I do have some Rayovac NiMH AA cells which are quite old, they just keep going and going, oh wait that's the other guy's line XD

Only the best for my wife! :smiley:

I've tended toward Duracell or Energizer, but not for any particular reason.

I've tended towards Energizer because they're sold everywhere: grocery stores, home improvement stores, even the pet store. Their mascot is a bunny because, like rabbits, once they get a toehold you can't get rid of them. Their slogan is "keep going and going..." not because the batteries keep going but because their marketing department does. Their mascot is pink to distract us from what they're really doing. ]:smiley:

Rayovac tends to be a little cheaper around here, but not as common.

Ditto.

I do have some Rayovac NiMH AA cells which are quite old, they just keep going and going, oh wait that's the other guy's line XD

But for a different reason.