Whilst not disputing or detracting from any advise given above I have read much recently about the various battery packs becoming "smart" and controlling the power/current output to smooth out spikes etc.
I wonder if this may have some affect on peak demands from PAW being smoothed out.
My packs are older and pre this terminology and I have never experienced this problem. I use an Anker Powedrive 5 in the aircraft but I believe even the latest incarnation of these have the "smart" technology.
I've no idea if this has any bearing on the problem but thought it worth a mention.
Alan
It is the device that decides how much power to draw, regardless of supply. If the device decides to try and draw more than the supply can handle you will get problems. All schemes such a PowerIQ are there to identify to the end device that is it connected to a fully leaded supply and therefore is free to pull as much as it can. This is done by fiddling with the D+/D- lines. Mine do similar things looking to tell the end device that it is free to pull as much power as it wants. If something works totally outside the specs (looking at you Samsung...) you can put a customised profile on one or more ports on my units to counter this.
The Pi doesn't have the D+/D- pins connected, as it's not looking to charge but be a "Power Thief" under the standards, just a device hungry for power. The specs say any device attached to a USB port should be limited until identified, although in general chargers don't do this but the likes of laptops etc. do. Most of the time such chargers will simply connect the output power direct without interruption (not a good idea for a number of reasons) so things like the Pi will work fine up to a point.
What is unknown is if the negotiation fails under PowerIQ (as it must, as the pins aren't connected) does it actually impose a limit or just let the end device draw what it wants. As they say there is per port protection of some kind, and they can increase the voltage on a per port basis too (up to 12v for the latest ones), these newer supplies must regulate the supply on a per port basis. If I get time I'll get hold of one and stick it on the bench to see what happens, be an interesting experiment.
The Pi (and PAW) usage is 'spiky' needing a supply be to be able to have fast transient response. Chargers generally don't handle that too well, they are designed to supply a steady state to charge a battery after all. The end device is powered by the battery, and the supply charges the battery. The electronics in the likes of a phone or tablet handle this in the background, the input power doesn't go into the device direct. This is why so many of the cheap wung-hung-lo chargers can damage the charging ICs on modern devices, they identify they can do everything but physically can't and their output is so poor it knackers the charging IC over time.
I did spot their VoltageBoost description...
VoltageBoost™ is an Anker-exclusive technology that compensates for cable resistance by smoothing voltage output.
Other chargers have their speeds reduced by cable resistance. With the effects of cable resistance mitigated by VoltageBoost™,
your device receives its fastest possible charging speed, every single time.
...which is meaningless, smoothing output has noting to do with resistance. I suspect this got a bit mangled in translation!