Searching for USB Power Supplies That Won't Explode | Hacker News Reader

Searching for USB Power Supplies That Won't Explode

hackaday.com

78 pointsszczys40 comments

Sometimes even a well designed and assembled power supply can be dangerous because of defective components. Big Clive on Youtube does many teardowns of cheap electronics, and here's one of a USB power supply that's dangerous because of internal damage to the transformer:

https://www.youtube.com/watch?v=3Hdn0MuCK_0

Because this is an intermittent fault it's possible that non-destructive testing would not find it.

That's not a random failure, though. The quality control for the transformer manufacturing was inadequate. This wasn't some difficult to anticipate error, either; maintaining the integrity of the winding insulation is basically the most important thing to monitor. The machines were either poorly designed or improperly operated.

Sourcing higher quality components would prevent such issues. I'll concede that nothing can be perfect, but this type of error is preventable. It starts to cost quite a bit when you move from 0.1% error to 0.01%, though.

It's not just quality control on the part of the transformer manufacturer. It was a cheap design. They could've completely isolated the two transformer windings, either with heavy tape, or wound them on two separate bobbins. That way, if the enamel on the wire wears off, the two sides are still completely isolated. I still agree with your conclusion, though.

They might not have been built by machines. Here's a video of those transformers being made: https://www.youtube.com/watch?v=HEJn74Z3wHk

I trust Anker's chargers more than anyone elses. (No association with them, just a happy customer.)

In general, chargers from big brands (Apple, Amazon, Samsung, even Ikea) are fine because they don't want to risk damage to their brand if one were to burn down a house or something.

If your charger has a brand you've never heard of, beware!

Also, don't trust the name on the case. Know the provenance.

There are plenty of fake chargers out there that look on the outside identical, down to the tiniest details, to Apple chargers, for example.

Me too - my Anker car charger delivers 1800 mAh as measured by an Android app called Ampere - I can charge my phone only during commute and use it for 24 hours. A cable I have from them is also high quality.

OTOH some Ebay cables might drop it down to 500 mAh. It was a shock to me because for years I exclusively bought Ebay chargers and cables (around 1/5 of the price of Anker) thinking they would work as specified (or not at all)...

Different gauge on the power wires give different results during charging. The original .5A of USB worked well with 28 AWG, but you need to go with 24 AWG to get 1A or more out of a charger. Sadly very few cables list this kind of info.

And even when you have the proper cable, you can run into the problem that the charger manufacturer thought the Apple way was the proper way.

Thing is that Apple has some very odd resistor setup on the data pins to indicate max A out of their chargers, while the USB charger spec basically says to short the data pins so that the device can sense it.

Try to charge a non-Apple device in an Apple type charger and you can get anything from full to .5A, depending on how clever the device engineers were...

FYI: "mAh" or milli-amp-hours is a unit of capacity, in this case it makes no sense. Perhaps you mean just "mA", a unit of current, which does make sense in context?

Sorry! You're right.. and the app is also right, I couldn't even copy from it properly :(

From what I've experienced, most power supplies are just awful and don't hold up at all to their specs, especially current in excess of 500mA.

I've been working with LEDs recently (APA102s), and I attempted just doing some testing with 4 meters (60 LEDs/m) and a cheap old 5VDC laptop power supply that in theory was rated for 5A+. Didn't work so well. Overheating, not providing enough current to keep the LEDs acting stable, power on spike, etc...

To solve my problem, I've purchased two 5V 90A Meanwell power supplies, and those seem to work pretty well. But needless to say, I cannot imagine cheap/small power adapters working well at all.

At my day job, we have some hardware that requires 800mA at peak to run. Even many 1A USB supplies we've found to offer nowhere near that current capacity. And if they do, they frequently fail after a while.

With a quick look, APA102 LEDs draw 0.04-0.2A.

Even at minimum intensity, we have 4m × 60 LED × 0.04A = 9.6A for the best case scenario.

> we have some hardware that requires 800mA at peak to run. Even many 1A USB supplies

I've been doing this for 30 years now. My rule is. Power supplies? Derate by 50% or you will be sorry. That power IC that says it's good for 1.2A? Yeah read that as 600ma.

Capacitors are the same way, rated for 16 volts? Try 10 volts. If you look at capacitors specs you'll often see MTBF at rated voltage and temperature of 2000 hours which is three months.

Usually I see a big voltage sag when you get near the rated voltage on most wall wart type supplies. I try and stick to 50% or maybe 75% if I think it's quality.

> At my day job, we have some hardware that requires 800mA at peak to run. Even many 1A USB supplies we've found to offer nowhere near that current capacity. And if they do, they frequently fail after a while.

What you could be running into there is a charger OEM that thinks Apple way is the proper way.

Thing i that Apple uses a couple of resistors on the data pins to give slightly different V between them this indicates to an Apple device that they can go 1A or higher.

But the USB spec says to basically short the data pins.

Depending on the device OEM you may have the device think it is dealing with a normal computer port rather than charging port when encountering the Apple setup, and dropping back down to .5A.

I think the best power supplies are those commissioned by big manufacturers to go with their hardware. The HP Touchpad's power plug being a good example. These brands don't want a poorly designed adapter killing their products.

I was going to say something similar. I often use USB power supplies that came from old phones/laptops for my various little robot/gizmo projects. I do this with the (naive) assumption that If I'm Apple/Dell/Samsung/etc. I don't want to taint my products because of bad power supplies.

Other than using my non-pro knowledge and tools to roughly test voltage stability, sustained amperage, and output voltages on some of the power supplies, I don't have any hard evidence that this is true- but if I'm choosing between an Apple USB power supply or a no-name power supply from Amazon for a project, I'm taking the Apple one because I can assume it's a minimum level of quality that's good enough for my use. I've got a couple of iPad USB power adapters that I guard pretty closely because the other USB power supplies I have that say they're good for at least an amp or two are from no-name suppliers.

My (Asus) Nexus 7 charger died, and none of the cheap replacement chargers I found would charge my tablet. Bought an HP Touchpad power plug off of eBay based on an XDA Developers recommendation. Worked perfectly for years.

Makes me wonder if DC wiring will ever catch on in newer homes. We don't have much need for AC anymore. The only places in a home where one needs AC is in the kitchen for large appliances, and garage/laundry room. And really that's just because of the motors. I'm not sure if those could be converted to DC motors and what powerloss that might entail. But the point stands; you could totally get away with building a new home where you drop AC to the kitchen and garage, and then 12VDC+5VDC everywhere else (12V on the wires, put a regulator to drop to 5V in the outlet for USB-C ports). The power-loss of DC in a home won't be that large. The end result is that the AC to DC transformer will be centralized to one location, where you can spend more money up front to get higher reliability, cleaner power, better safety, and better efficiency. Plus, if you have solar, you can avoid AC to DC conversation altogether, saving on a good chunk of powerloss there.

The caveat, of course, is that lots of devices don't natively support DC input. Phones, tablets, and modern USB-C laptops are fine. Lighting is easy; just do permanent LED installations. The LEDs will last a very long time since they won't have regulators in them. Desktops are an issue. I believe the only desktops that support DC input are HTPC builds, using those micro PSUs. Perhaps there's an HTPC PSU that's got enough beef to power a normal desktop? You'd also have to find monitors that accept DC input, which do exist but that will narrow your market significantly.

Rob Rhinehart is an early adopter (see: http://robrhinehart.com/?p=1331) of this type of home. Sounds crazy, but it seemed like a glimpse at a hopeful future, at least everything outside of the kitchen.

Well, uh, AC vs DC and 100-250 volts versus 5-48 volts are two completely different dimensions to the problem here.

For example, there has been some standardization of high voltage DC power distribution for datacenters, seeing as that could simplify providing battery backup, etc.

But back to the residential case: As soon as you go to lower voltage power distribution, unless you spend a lot more on copper (larger conductors), you're really hurting as far as power distribution losses: your load uses 10 times as many amps running on 12 volts DC as it did on 120 volts RMS AC. Thus your loss due to the resistance of the wire is 100 times what it was, given the same wiring. (i.e. the same series resistance -- your loss is P = V * I = I^2 * R)

And no, you don't suddenly get much greater efficiency in the power supply of the device - a lot of off-line power supply designs today are already 90% efficient, and even if you provide 12 volts (or 12 and 5) to the plug, there's going to be a lot of things you want to do in the device that does not satisfy, and if you want exactly 12, you're going to need some kind of DC-DC converter with regulation, because you cannot predict how the 12 volt supplied at the wall will go down due to other load at any given point in time -- so just like a laptop has various DC-DC converters from the 18 volts supplied by its power brick to the various internal circuits (Vcore logic, 1.8 volt logic, 3.3 volt, 5 volt USB, backlight circuits, etc) most devices will still have that. They might not require as much safety isolation, but you're not going to get them to be magically far more efficient than what we have today.

Lets not forget we could change the frequency. Running at 400Hz like aircraft do lets you shrink the capacitors, inductors, and transformers of your power supplies and your electric motors can make 10 times the power for the same weight.

Going to 3-phase instead of single phase probably also helps for power supplies. The power delivered across time is uniform so a sufficiently clever design doesn't need to store energy between cycles. Sure, its another conductor, but they can be slightly smaller so you use the same amount of copper in the end.

50/60Hz is better suited to long distance transmission, but you could put an inverter to 400Hz for the last 50 meters. Smaller, lighter, cheaper devices but with the efficiency loss in the inverter. I suspect it would be a win.

(Also, pick a new voltage, something in the ~100v+ range. High enough for efficiency, but then do the lethality tradeoff on picking the ultimate voltage. As long as its a new standard, build ground fault detection at the inverter as part of the standard. There are about 400 household electrocutions per year in the United States, so there is room to improve.)

You really don't want to use DC for distribution in your home. Go price out telcoflex copper cabling (from 0000 up to 4 gauge) for -48VDC datacenter use. Voltage drop over distance and losses to heat make it impractical in homes. You really want to pay the extra money to wire your home with 6AWG cable in the walls? That's a lot of copper.

Centralizing an AC to DC transformer is pointless for consumer stuff since things run on DIFFERENT voltages. A lot of consumer grade stuff is 5VDC internally. Or 12VDC. Or 17.5VDC. Or 19.2VDC laptops. Or 24VDC. Or 48VDC.

That's before we even get into the problem of "everything in the world is already standardized to run from 110-240VAC 50/60Hz".

If you really want to do DC distribution in your home, do it with 802.3af/802.3at switches (or "dumb" midspan injectors") and cat5e/cat6 cable. You can get LED lightings and other things that will run off PoE voltages and amperages now.

Also, with the new USB standard, its no longer only 5v, you can now use up to 20v @5A [1]. My laptop that has a USB-C port to charge uses the full 20v.

120/240VAC 50/60hz @ ~12A (steady state draw) is pretty much the only thing people agree on when it comes to consumer power.

[1]: http://www.usb.org/developers/docs/

I'm waiting for the first news stories about house fires caused by USB-C chargers. A faulty 5Vx1A charger will melt, 20Vx5A will be a bit more lively.

The tests by people like Benson Leung [1] haven't given me any faith that manufacturers will actually care about getting the specs right.

[1]: https://plus.google.com/collection/s0Inv

wait until we have USB-C charged hoverboards (and similar products), where not only can the charger and cable catch fire, but so will the low budget Li-Ion battery pack/charge controller/battery management system.

Also, the specs themselves are non-ideal. What they really need instead of the cable DRM/smarts (USB Authentication [1]) is to standardize V-Droop (Voltage drop from no-load) detection and behavior.

They need to make part of the spec a current ramp and monitoring of the voltage while doing so. If the voltage sags too much or ripple increases that means the cable or power source can't handle the load and the device needs to back off. There is very little a non-compliant power supply or cable can do to fake conformance without costing more than doing a conformant implementation. Some phones monitor V-Droop already as there are too many bad power-supplies and undersized cables out there. The downside is that it isn't as sexy to implement this.

[1]: USB Authentication section of http://www.usb.org/developers/docs/

Oddly i think the original charging spec do it that way when dealing with dumb chargers.

I've been searching for a good-quality usb hub. After a few years of intense competition, there are basically two options on the market: cheap chinese crap, and cheap chinese crap in an expensive designer enclosure.

I have a couple Anker USB 3 hubs (one is a 9 port and one a 7 port) and they have not had any issues.

It's not like any of it's made in America, so...

I have one of these and it's been fine so far.

http://www.amazon.com/EasyAcc-Bus-Powered-Self-Powered-Ultra...

To reiterate everyone else, Anker is great. I've had good results with TronSmart too. And the Amazon Basics lineup is universally reliable. They have all sorts of chargers/hubs/cables at reasonable prices.

If anyone can shed some light on the supply chain for these I'd be interested - I've seen Anker, Aukey and Tronsmart 5 or 6 port chargers that look identical (performance has been almost identical too, luckily). I can't buy Anker or Aukey in Australia (easily / for a reasonable price) but I can get Tronsmart through Geekbuying.

I've had really good luck with this one from Amazon http://www.amazon.com/AmazonBasics-Port-USB-3-0-Hub/dp/B00GG...

It's fairly inexpensive (there's smaller versions too), is powered, and backed by Amazon if there's an issue.

This is obvious to the point of not reiterating but don't ever plug your stuff into cheap/unknown chargers. I've seen the AC jump across to the low power DC side MANY times and it insta-destroys anything designed to just take the +5vdc.

I work in the event industry and we supply recharging cabinets for corporate events. We use Anker.

I generally avoid anything that doesn't have legitimate safety certifications: UL, ETL, TUV, etc. ("CE" as printed on random imported crap seems to mean nothing in particular.)

Also, you can usually count on actual, non-counterfeit chargers from well-respected phone and tablet vendors being just fine. I've never thrown out an old USB charger from an iphone, samsung, motorola, asus, just put them in a box for possible later use :)

  "CE" as printed on random imported crap seems to mean 
  nothing in particular.

http://www.icomuk.co.uk/News_Article/3794/17103/

https://en.wikipedia.org/wiki/CE_marking#China_Export

After reading Ken Shirriff's review a few years ago, I bought a bunch of Touchpad chargers off eBay (before the clones started appearing) and use them for tons of stuff.

http://www.righto.com/2012/10/a-dozen-usb-chargers-in-lab-ap...

Otherwise I stick with Anker/Aukey; am a huge fan of the 6-port "charger bricks" and have used one next to my bed (phone, two tablets, a couple of USB-powered devices like a sleep monitor) for 3-4 years now.

Here are some good analytical reviews of USB power hardware, including safety: http://www.lygte-info.dk/info/indexUSB%20UK.html

In general I buy Anker stuff - decent pricing and solid stability/quality/safety.

There's a difference between active and passive protection systems - similar to certain Lithium-ion batteries. A lot of the cheaper manufacturers skimp out on a lot of such components (hence why you hear of e-cigs blowing up, etc).

I've been very satisfied with my charger from ChoeTech. I figure that if something is Qualcomm Quick Charge Certified, then it is probably adequate to charge my devices.

For inexpensive USB power, Anker has been my go-to. They also do the Qualcomm quick charge with many of their products. Related, I liked their stuff so much, and was so impressed when ripping apart one to use for lots of 5V amps, that I bought their LED desk lamp (Lumos A2) which I've been very satisfied with. They also have the E1 out now, which is more powerful as lumens go.

Edit: tear down of the charger I like by Anker (thanks binman): http://lygte-info.dk/review/USBpower%20Anker%2040W%205-port%...

Reply on news.ycombinator.com