The Wireless Android POS/NFC Reader Suck Index

Now that contactless is everywhere, wireless contactless readers have become very fashionable and popular. Nobody wants wires or checkout lines. All of these systems are built around an Android based reading device connected to the internet payment service via Bluetooth, WiFi or 4G with a main terminal, an iPad or a laptop running POS software. Convenient though they may be, compared with hard wired NFC reader performance they all suck with different levels of suckiness:

1. stera: this lovely little ‘NFC antenna under the screen’ piece of shit from SMBC, GMO and Visa Japan is so slow that checkout staff put their hand over the stera screen/reader to keep customers waiting until the device is ready to go. This is followed by the instruction ‘don’t move your device until the reader beeps.’ It’s a 2~4 second wait until it beeps. This is 2014 era ‘you’re holding it wrong’ garbage nonsense. I teased one store manager about the hard wired JREM FeliCa readers that were swapped out with stera, “Those were too fast,” he said. Too fast?!
2. PAYGATE: Another payment provider associated with GMO, slightly faster than stera but still slow, PAYGATE does’t like Apple Pay Suica•PASMO Express Transit very much. Have of the time it ignores it altogether forcing customers into the 2016 era ‘manually bring up Apple Pay Suica’ authenticate and pay maneuver. Another ‘you’re holding/doing it wrong,’ when the fault is on the checkout system side. Passé and totally unnecessary.
3. AirPay: It’s weird that the cheap AirPay hardware performs better than PAYGATE or stera, it’s even weirder that AirPay performs better than Rakuten Pay which uses the very same reader but is stera shitshow slow.
4. Square Terminal has gotten lots of media attention in Japan. Too early to experience it in the field yet but I’m not hopeful. Square Terminal is Android based after all and the NCF antenna under the screen design is the worst performing reader design out there. As one Brazilian reader wrote: “I just don’t like the ones running Android because at least here the software is less reliable and I managed to crash a few one by just taping my phone.”

Yep, that observation matches my experience. Payment network providers need better Android readers, the current crop is too slow getting the payment transaction ready to tap. In this era of endless subcontractor layers in the development process, creating a fast reliable Android based NFC wireless reader might be a tall order, if not impossible. The all over the place wireless NFC reader experience certainly doesn’t boast well for open loop advocates.

UPDATE
I ran across another crappy reader experience (above) and retweeted it. A reader had some questions about it, answered here by an anonymous expert. It basically comes down to poorly executed reader polling or not following Sony polling recommendations for FeliCa cards. This is what is happening in the above retweet. It is also what is going on with PAYGATE Station readers, half of the time the proper code hasn’t loaded correctly although this issue seems to be fixed in new PAYGATE Station checkout installations. Which brings us to the point I was trying to make: these performance issues can be fixed with reader firmware updates or transaction system software updates, but never are.

Wildcard polling involves the reader making a request for system code 0xFFFF and expecting the card/device to list all the system codes that it supports. Wildcard polling won’t work on an Apple Pay device in Express Transit mode – instead, the system code must be explicitly polled for (0x0003 for CJRC, 0x8008 for Octopus). You can cause Suica/Octopus to be automatically selected by sending SENSF_REQ (Polling command, 06) for those services explicitly.

I have verified that doing so with Apple Pay will cause the emulated card to be switched out as appropriate – the IDm value will also change, since Apple Pay emulates each card separately, instead of with a common IDm as with Osaifu Keitai. If you read the Sony documentation, you will see that developers are cautioned to also poll for the specific service codes they want to access if there’s no response to a wildcard poll.

Perhaps your reader doesn’t do this, but it’s fairly big omission…it should be doing explicit polling. Simply polling for service code 0x0003 should wake up Suica if selected as an Express Transit candidate, even if you don’t send any other commands. I’ve verified this with an RC-S380 reader and NFCPay.