Micro USB Cable Power Tests

This writeup was brought about by someone's braindead decision to keep using a cellphone charger interface for Raspberry Pis long after it became obvious that this was a really dumb idea. Problem is that the vast majority of micro USB cables are incapable of passing the level of current required to run a Pi, which means you either need to go and buy a custom Pi-specific power supply, thereby defeating the point of using a cellphone charger in the first place, or spend forever trying to find a cable that can feed enough power to it that it won't glitch or fail.

Since I have to maintain a couple of Pi-based systems despite preferring not to (at least some packages can be moved to an ODroid even though they're nominally meant for a Pi), I ended up stuck in the position of feeding power to them through something that was never designed for it, or at least not the levels of power that a Pi requires. Buying a bunch of Pi-specific power supplies was out of the question both because of the cost and because I don't have room for a row of wall warts to power the Pis.

So it was down to finding a USB cable that you could run low-voltage, high-current DC power through, pretty much the worst way to power a device. For reference, USB power is specced to 5V±5%, so 5.25 - 4.75V. The Pi has a primitive undervolt sensor that trips at 4.63V but it's a one-shot so if at any point in the past the voltage ever went below that level it'll stay tripped, making it difficult to diagnose when the problem occurred.

The following table shows the results from various USB cables tested with two types of dummy loads, a passive resistive load and an electronic load. Input was 5.1V from a lab power supply, voltage and current was measured across a dummy load set to draw 250mA and 500mA (for the original USB power rating), 1.1A (about 5W with any potential drop factored in), and then 1.6A and 2.3A, with the 2.3A being a common maximum-current figure for many USB power bricks. The figures in the columns are the output voltage measured at the load, and in a few cases where the voltage drop with the passive load was extreme, the current. The cables without names are generic unbranded cables, the cables labelled “White” are charging cables that came with a phone or tablet, “Ribbon” are ribbon cables, “Switched” is one with an inline switch specifically meant for delivering power to a device. To help those who aren't used to metric measurements, I've also given the length in arshins. You're welcome.

There are two sets of figures because of the way the measurements were taken, the first line is for the passive load which was connected through a USB power meter that added two USB connectors and the meter draw while the second line is for the electronic load which was connected straight to the source. This means that for the good-quality cables voltages are a bit higher on the electronic than the passive load. In contrast the passive load uses the basic R=V/I to try and draw the current it needs so will switch in the appropriate resistance to draw 1.1A even if it's only getting 900mA while the electronic load will draw an actual 1.1A. This means that for the poor-quality cables the voltages for the electronic load drop a lot more than for the passive load.

If you don't understand the above paragraph, go with the values for the electronic load. If you want a lot more detail about USB cable and connector current-carrying capacity, Andreas Spiess has a good overview, note however that he's mostly working with very short cables where there's inherently less voltage drop.

Some general observations:

• It's essentially impossible to reliably power a Pi off a cable over 1m long if you're using a USB cable. You simply cannot run 5V through that length of thin wire at the current levels required by a Pi, the losses are too high (although the Anker 2m does pretty well up to 2.3A so if you're not running it at the theoretical full power of 3A that's the one cable you can use). Otherwise, if you do need to power one more than 1m from the power source, run 12V through a proper barrel-jack power cable and use a UBEC next to the Pi to take it down to 5V.
• Cable thickness says nothing about its current-carrying capacity. The Ugreen is one of the thinnest, most flexible cables of the lot but, once you exclude all of the Ankers, has the second-best current-carrying capacity of any of the ones tested, including quite substantial cables.
• Going with name-brand cables is a huge win. All of the name-brand cables easily outperformed all of the generic cables.
• Generic D can't in all honesty be called a cable, it's actually a very long thin resistor. The voltage drop is such that it falls outside the USB spec at around 80mA of current.
• The switched cable was the second-worst-performing of the lot. This is one specifically intended for supplying USB power and yet it's incapable of reliably doing so. It also produced erratic measurements that changed on each run, a sample is given above. Given that even the long thin resistor produced consistent measurements, I'm guessing the inline switch is the cause of the problem, making the improved measurements over time a side-effect of the switch heating up slightly with the current it's passing, however this doesn't explain the resistance measurements which never actually stabilised but drifted all over the range shown.
• The Logitech cable came with a Logitech remote and is meant only for data transfer so don't treat it as a black eye for Logitech, it was never meant to run power. Even then though, it beats more than half the generic cables.
• Anker make damn fine cables. I'm not being paid or sponsored by Anker to say this, just reading the data and commenting on it.
• Unrelated to the cable, Anker power supplies have remarkably good regulation, a flat 5.1V across the whole current range up to 2.5A. Coincidentally, they also make the best cable to go with it, their off-the-shelf micro USB cable (the full 1m one, not the 10cm stub cable) easily outperformed even ones specifically advertised as being built for high current capacity.

If anyone has any USB cables they'd like tested, feel free to fax them to me. In particular if you think you've got something that can beat the Anker, I'd be interested in seeing it.

Mini USB Cable Power Tests

Because there are still devices powered off mini USB, here's a much shorter table for those cables. These date back to 500mA max USB devices, so there's less expectation that they handle higher currents well. As before the cables without names are generic unbranded cables, “Braided” is a plastic-covered braided tinned-copper cable as was fashionable years ago, so with the conductor braided rather than a protective nylon braid.

USB-C Cable Power Tests

Finally, a few USB-C cables thrown in just for comparison. These are all USB-A to USB-C, there are no tests for USB-C to USB-C cables because those do USB-PD and that seems to interact in odd ways with the electronic load I'm using. None of the USB-A to USB-C cables are E-Marked (not that you'd expect them to be, but you never know).

You can immediately see the difference between these and the mini USB's above, but then again the mini USB's were designed to pass 500mA max rather than the power loads expected of USB-C.