Recently I had the opportunity to do a teardown of a battery-backed LED bulb, and found some interesting details on how the device operated. Essentially, the bulb contained a low voltage DC uninterruptible power supply that would automatically switch between AC power and internal battery as needed. The implications of this seemed pretty exciting. For around $12 at big box retailers, this little bulb could be a cheap and convenient solution for providing fault tolerant power to microcontrollers and other low-power devices.
The teardown was a runaway success, with quite a bit of discussion of the UPS idea specifically. Some people hated it, others loved it. But as we’ve come to expect from Hackaday readers, the comments from both sides of the aisle contained keen observations and invaluable real-world experience. From the safety of the device to the accuracy of the manufacturer’s claims, it seems like every element of the product was addressed.
I had ended the teardown with a promise that I’d continue experimenting with the tiny salvaged UPS, but even if I hadn’t, with so much feedback it seemed revisiting the subject was all but a necessity. It this little UPS really viable? Is it too dangerous to safely implement in your project? Will the thing just blow up?
So with your comments as a guide, and free of the somewhat restrictive teardown format, I set out to conduct a more thorough investigation of this little circuit that caused so much debate last month. It’s not all good news, but it’s not in the trash either. Not yet, anyway.
A Word on Safety (Or Lack Thereof)
Without a doubt, the most common complaint in the comments was that this device is not safe and shouldn’t be played with. Touching the traces on the PCB or one of the wires could give you a shock. Which of course is true, but that’s certainly never stopped a Hackaday reader before. Obviously it was never meant to be taken out of the bulb’s case, and as such it’s being used in a way that throws safety out the window.
Certainly the most obvious issue is the isolation, or more accurately, the fact there is no isolation. That’s right, the AC and DC sides of the board are connected. Now, if this was a USB charger or something along those lines, it would be a heinous design flaw. But remember, this is a light bulb. The user was never meant to see, much less touch, the internals. It’s hard to fault the designers for not worrying about isolation when the device in question was never meant to come out of its protective case.
But does the fact that the circuit isn’t isolated mean it can’t be safely used? Not necessarily. Just as the original bulb wasn’t dangerous since it was a sealed unit, if your design isn’t meant to be touched with by the user, there’s no immediate danger. Imagine something like an ESP8266 weather station that simply reads from sensors and sends its data out over Wi-Fi, such a device would likely live its life inside an enclosure anyway.
The real risk here is if you have some kind of project in mind where the user is expected to physically interact with it, or worse, plug something into it. Here the risk of electric shock is high enough that you’d want to steer clear. [Big Clive] examined a USB power supply with this problem awhile back, and illustrated the dangers:
So let’s say you’ve got a scenario where the lack of isolation won’t be a problem, perhaps the weather station example. How much current can this little device provide?
The run time for the LED light was given as 3.5 hours on the package, and since it had a 2,000 mAh battery, some back of the envelope math says that 400 mA is probably the best we can hope for. At least for an extended period of time, anyway. Testing this is easy enough, we just need to put an adjustable load on the DC side and start raising the current until the voltage dips.
The estimated figure proves to be accurate, as the device can provide around 410 mA before the voltage starts dropping off. The maximum looks to be around 480 mA, so it’s a fairly narrow sweet spot.
As a proof of concept in the teardown I soldered a Wemos D1 Mini to the DC side of the UPS board, and sure enough it powered up and worked fine. But it was also something of a risk. According to the load tester, the board is putting out nearly 3.5 V at the upper limits of its current capacity. When the current is lower, that voltage gets as high as 3.8 V. For sensitive 3.3 V devices, this could be an issue. Luckily the solution here is quite simple, just add a 3.3 V regulator.
You could also use boost converter to get up to 5 V, but with only 400 mA of current to play with on the input side, it’s not going to do much useful work.
The Real Problem: AC Mode
Up to this point, I’ve only been dealing with the board when it was running on batteries. My hope was that when I switched on the AC, things would work the same and there’d be no issues. Sort of critical if this is to be used as an uninterruptible power supply. Unfortunately, that wasn’t to be.
Once you connect AC power, the DC side starts pulsing along at (as you probably guessed) around 60 Hz. This is sort of a problem. It’s still outputting ~4 V DC, but it’s coming as a square wave; which worked fine for LEDs but certainly doesn’t play nice with microcontrollers and the like.
The solution here could potentially be a big enough capacitor to ride out the pulses. Perhaps use the output of the board to charge a super capacitor of a Farad or two.
Incidentally, hooking this non-isolated circuit up to your oscilloscope is a good a reason as any to either buy or build a stand-alone isolation transformer.
Accessories Sold Separately
After spending some more time experimenting with the board, I have to admit I’m considerably less excited then when I first pulled it out of the LED bulb. It seems there’s just no getting around the fact that you’ll still need to add additional hardware to get it truly working as a low-voltage UPS. My hope, which was inspired by the quick test with the ESP8266 during the teardown, was that this would be a turn-key solution for those looking to create AC powered devices with an integrated battery backup. But reality has a way of changing things up on you, unfortunately.
I still think it’s a very interesting board, and worth tracking down one of these battery backup LED bulbs just to experiment with it a bit. But unless you’re willing to start tacking additional components on it, its usability is severely hindered by the pulsed DC output when running on AC power.