AWS IoT Button | Hacker News Reader

AWS IoT Button

aws.amazon.com

285 pointsvasinov137 comments

I hacked one of these to power the nightlight in my kid's bedroom (https://github.com/ecaron/hue-website-nightlights). The downside is that the button barely lasts for 1000 presses = 90 days (all kids like pushing buttons). And it is heart breaking because its just a AAA battery that they decided to solder in place...

I'm excited that this button launched. But $20 is too much for something without a swappable battery. I'll either keep intercepting the ARP packets from the $5 unit or buy https://www.particle.io/button.

Is desoldering an option?

It's tricky, metal tabs are soldered right to the battery terminals. It's a lithium battery so you don't want to heat it too much so desoldering will be annoying and potentially dangerous. What you can do instead is cut both tabs to remove the battery then solder wires to the tabs on the board and connect a proper battery holder. Might need to move the battery outside it since the holder + battery won't fit.

For folks who use words like "desolder" it's not tricky at all.

This (and the Dash) come in a glued shell with a 1.5v Energizer Li AAA solder-welded to the back. Getting inside without completely destroying the curb-appeal takes a hex bit, some scraping, and a little patience.

Once you get to the board, the battery is easily replaced with a LiPo. The MCU is 3.3v, but Amazon Prime has you covered on cheap LDO/buck converters (like, "a buck" converters).

Wait, what? So, we are basically moving from this which could last decades to stuff which stops working after 1000 presses?

And people say we are not moving to planned obculence ( sorry, don't know the English word for it)

The word is "obsolescence." (I wouldn't be able to say it any any other language!)

Planned obsolescence = obsolescencia planeada in Spanish, designing purposely a product that won't last to get money through client's pockets. Examples with ink for printers was on courts.

It sort of makes sense for the dash button's original purpose. You probably wouldn't order stuff like paper towels / toilet paper / dish soap more than once a week or so (less for for most people), so 1000 presses would last you over 19 years.

Using the exact same hardware for a general purpose remote control seems like a mistake - an extremely limited product at best, an unnecessary source of ewaste at worst. A modified dash button with a replaceable battery would me much nicer, though I'm sure it's far cheaper to just sell a rebranded dash button.

> so 1000 presses would last you over 19 years

I've been wondering about that. They say you can use it to control lights/netflix. If you don't want 10s delay it has to do wakeups every 30s or so depending on the wifi setup to stay connected to the network. And wifi transmission is quite power expensive. If it does that, I wouldn't give it more than a year (maybe more like a few months). If it doesn't, the latency must be terrible.

An ESP8266 can go from deep sleep to publishing a request in less than three seconds. The expensive operations are the DHCP request and the DNS query, but that can be cached for hours, often even days.

Well maybe not 10s, but 3s is still a terrible latency if you want to control something like lights for example. And it's not all that uncommon for dhcp lease to last only a few hours (and we are talking about a case here when you use the button once every few days).

I used to be irritated with cheap led bulbs that took almost a second to light up. And with a wall switch you are pretty sure whether you flipped it or not. With such button during those 3s you'd be wondering whether you should press it again or not (unless it beeps or something like that).

From the Amazon official page (https://aws.amazon.com/iot/button/):

    How long will the battery last?
    
    The battery should last for approximately 1,000 presses.  
    When the device battery runs out of charge, there is no way to recharge or replace the battery.

How can you sell a button that has only 1000 presses in it for $20 with a straight face? Looks like overcomplicated, overpriced garbage to me.

From an environmental perspective, it is completely retarded I would say.

I don't get why people like Amazon, none of their services are any good. Although to be fair, Amazon (the retail) doesn't exist in my country.

> none of their services are any good

Care to back that statement up? :)

(no argument on the environmentally subpar approach to the button)

Well, it's a subjective statement as it is my opinion. But for me as a Swede:

* The closest retail in Amazon is in GB or Germany, orders takes weeks to reach me which is a mystery since I've ordered stuff from Germany before and they usual deliver within a week.

* Closest AWS datacenter is in Ireland, latency is high for nordic countries.

* Their documentation is very confusing, you can read all about their web services without understanding one. Leading to sites like this: https://www.expeditedssl.com/aws-in-plain-english

But sure, their services works and is probably good for many, but I've never found them any useful at all.

Isn't there an AWS Amsterdam datacenter which would be better latency?

Frankfurt, but latency doesn't always correlate directly with physical distance.

Try Frankfurt. It may have better latency for you, even if it's further away.

AWS is a pretty popular service ;)

Amazon cloud drive is fairly performant and a good price. I use it as a backup repo. Amazon prime shipping is also pretty great and the prices are pretty decent usually. They're also the biggest cross platform ebook vendor with the most selection.

I can't believe they do it either. It's a button. Its main purpose is to be pressed. They literally say it will cost you 2 cents per button push.

And that is assuming your time spent on ordering and setting it up is worthless.

I guess the original design was to order things infrequently, and if you ordered a new detergent every 10 weeks that's fine.

Just cancelled my order based on this.

me too

me three

Fourth.

I kinda wanted one in the hopes that if it worked well, I could theoretically get a lot more to trigger actions for things with my Z-Wave setup.

20$/each isn't amazing but its not exactly terrible for this. But 20$/each/year (or less!), horrible.

It's being presented as testing hardware, so you can get to know their stack. At $19, it's probably being subsidized substantially.

In the same way free AWS instances are not adequate for hosting a production website, this is a toy for devs to get to know their stack...

The real question for this device is, is it hackable, i.e. can you change the software running on it, or open it up and solder different hardware to spare pins?

It looks like people have done this, but it's surprising amazon would sell a devboard without catering to it: https://github.com/dekuNukem/Amazon_Dash_Button

> this is a toy for devs to get to know their stack...

The difference being that AWS has real tiers. The dash button is exactly the same even if you "go pro" (what does that even mean? You persuade your BigCo employer to join the program?).

I suspect it's a last-gasp attempt at making some money out of a program that has not become as popular as they thought.

"At $19, it's probably being subsidized substantially."

Doubtful

If it's testing hardware, then why doesn't it have a replaceable battery? This is just madness.

Wow, I'll wait for v2 then.

> no way to recharge or replace the battery.

No way at all in the entire universe not even with hand tools and a soldering iron?

I think what they mean is no reasonable way. Who wants to spend $20 on a button they will have to tear apart and desolder?

Follow-up: This thread is getting a fair amount of heat, and let me foremost say that I love the button and think it'd work for the vast majority of people. Just as a child's nightlight, he pushes it probably 10 days a day (potty breaks, bad dreams, is it time to get up yet, etc). I'd also wager that my ARP-interception causes the board to consume more just the norm.

10x a day means it'll last about 4 months... do you really want to have to pay $20 every 4 months for this thing?

Certainly not. I'll keep using the $5 counterpart and intercepting the packets in an inglorious-but-it-works way.

Agreed. It should at least reorder itself when battery runs low (and copy it's configuration over so I don't have to redo it)

I love my particle. I'm even using it in production now for measuring remote process levels (ultrasonic level sensors, flow meters). I'll probably have 100-200 in the field by the end of the year.

You can replace the battery with a rechargable LiPo using a $2 security driver, spudger and a little elbow grease.

> The battery should last for approximately 1,000 presses. When the device battery runs out of charge, there is no way to recharge or replace the battery.

This limits use cases to infrequent-press applications.

But also limits the damage :)

https://twitter.com/AvidanRoss/status/727222223417667584?lan...

"Amazon dash button lost in my house is being tapped and ordering bulk paper towels. Time to hunt down a wifi signal."

https://twitter.com/InternetOfShit

As pointed out in the comments there, the button will not ship a second order until the first one has been delivered. You will get a notification email each time it is activated so you can cancel an order. It's also trivial to deauth a particular button.

This is interesting, since from teardowns it looks like there's literally just a AAA battery in there:

https://mpetroff.net/2015/05/amazon-dash-button-teardown/

Yup, but they decided to weld the contacts of the board to the battery. I'm guessing that they calculated it was cheaper (or more business savvy) to get users to replace the button every X months/years than to make the button in such a way that the battery is user replaceable and also indicates charge level.

It also means they can reasonably assume there aren't 10 year old buttons using an ancient protocol out there to support.

Sometimes planned obsolescence is a good thing.

You'd have to order an awful lot of toilet paper to have it only last 10 years. Even at a rate of one-press-per-week, it'll last getting on for 20 years.

The sample use cases for these buttons is more frequent use than the Dash "order me something" ones.

But do they use a different protocol?

It's a thing with an ID that sends a single event. How much can the protocol evolve?

Maybe the next version will have a thumbprint scanner so it knows who initiated the event. Maybe multiple buttons will be in a mesh network and send data about the other devices they're in contact with. I highly doubt we've seen the final feature set for IoT buttons.

But why would you deprecate the ability of a button to send a single event since that's what buttons do. Coordinating them in a mesh should be done outside of the device so it still works as a simple signaler.

Because fuck the environment!

Is it even safe to weld an alkaline battery? I don't know.

Why not just solder the contact? If it's soldered, you should be able to swap the battery with a soldering iron. Sure, heating the iron and cleaning it will take much more time than swapping a battery in a spring-loaded housing, but I've spent just as long opening battery compartments held shut with tiny stripped screws.

Yes. The welding heat is VERY localized to the spots where the electrodes made contact. Very small spots and very localized heating for a secure connection. In this regard, less heat gets "into" the battery than would be the case from soldering, the operation is very fast (cycle times in the low number of seconds and welding times in the few milliseconds range), and more than strong enough for the application.

The better ones use a double pulse for a cleaner weld without the need for any flux.

Our Puck (https://flair.co/products/puck) is sorta like this but with a small display. Instead of a generic button its a bit like a generic knob. So far positioned more as a heating and cooling device but I know I am excited to open it up to developers.

At 3x the price (or 12x the price of a hacked button), it's not really in the same market as the dash button though

agreed :)

No, it sounds way better:

Puck Specifications

Communication

=============

Bluetooth Beacon for occupant awareness

Wifi (Puck can act as its own gateway, no hub needed)

915MHz Radio for device to device communication

Infrared (IR) record/playback

Sensors

=======

Temperature

Humidity

Pressure

Ambient Light

Interaction

===========

Rotation and Click <---- Rotation is not on Amazons

E-ink Display <---- added

You'll hit breakeven on the price at 2,001 presses and from that point on it'll be cheaper.

That was one really nifty device.

Your site mentions device to device comms. Is this user programmable?

Also display only shows temp. Are there other options?

You will be able to reflash all three radios on the device (although you will need to keep the wifi's bootloader for the foreseeable future). Otherwise use should be able to swap out the bluetooth image and the subgigahertz stack.

The UI is reprogrammable without touching code also with 'microapps'. You could think of them as super lightweight shortcuts for adjusting things. You can see some examples on our twitter feed: https://twitter.com/flair_co

This looks fantastic!

> So far positioned more as a heating and cooling device

Indeed, the website and esp. the pre-ordering process lets one think it's really dedicated to that use...

Why not have another page with a more general positioning, this could be used for anything!

Yeah - tell that to our investors :p (j/k I love them but they felt it important to keep focused on the marketing for a while longer).

Definitely need to add some marketing pages though, one for 'generic knob' examples, and then I actually think it would be best to split up the minisplit/window ac bits from central heating and cooling.

Do you have a mailing list where we can sign up to know when you open it up? I almost immediately bought this, but thought "I should read up on the API first..." Reading the FAQ [https://flair.co/pages/faq] I see: "Can I write apps for Puck? Not yet. But soon."

I'll buy it the moment it lets me program it to suit my needs. It looks incredibly awesome and off the bat I can think of 4 places I'd love to use it (but none of them, unfortunately, have anything to do with heating or cooling).

I wish I saw this sooner (HN news needs notifications!). You can sign up for our email list to start and you can also start a thread on our forum (https://forum.flair.co). Excited to hear all the cool things that you want to use it for fyi :)

Is there at least one button in addition to the knob? How do I get my hands on a dev kit? I have been casually looking for an internet-connected dial + button + display for a long time. I have lots of different productivity uses I want to try out. Badly.

Edit: I see now that it does indeed have "click" listed in the specs.

Yes - there is a click and rotate. No dev kits just yet - we are getting close to shipping the first set of preorders and pretty heads down on that but it has a nice bootloader so you will be able to reprogram it over the air as well as via some exposed pads on the back. Would love to hear some of the things you are looking to use it for!

Time tracking, virtual desktop/spaces navigation, audio controls... scaling server resources? I don't know, there are a lot of things you can do with rotate + click + a little screen to let you know what you're doing.

I'm glad they accepted that people were already hacking [1] their cheap Dash buttons and decided to roll with it.

It's 4 times more expensive than their hackable Dash buttons ($20 vs $5), but apparently more powerful (and secure?) from the get-go.

[1] For example, see all the entries: https://hackaday.com/tag/amazon-dash/

> It's 4 times more expensive than their hackable Dash buttons ($20 vs $5)…

That's interesting. A new product like this (no matter how seemingly small) takes some work, so to me this suggests that people hacking the "subsidized" buttons were a bigger use case then they'd planned for.

It's a button. Let's not use the same word, i. e. 'powerful', reserved for THE button.

I also think it's quite ugly. I could imagine some repurposed giant old fire alarm for, say, code deploys. But not something that's reminicent of made in china ±10 years ago.

Would also assume (even if not from a BoM perspective) that the standard Dash buttons are cheaper as they're subsidised by the sale of the products?

Just the number of points of failure in this pipeline relegates this to toy status at best.

WIFI interruption? Button doesn't work.

Internet connection down? Button doesn't work.

Bad network congestion? Button doesn't work.

Amazon suffering DDOS? Button doesn't work.

Battery dies? Button doesn't work.

Too many presses? Button doesn't work.

Misconfigured server? Button doesn't work.

Certificate expiry? Button doesn't work.

"The more they overthink the plumbing, the easier it is to stop up the drain." - Montgomery Scott

What were you planning on doing with this? Launch a shuttle?

FWIW I tried following the instructions with a normal branded dash, and it is clear that the IoT button has a different firmware.

Differences: * Amazon ConfigureMe network uses WPA2 for the IoT button. No encryption for the branded button. * IoT button has extra fields (according to the documentation) that the branded does not. They are for a certificate, private key, REST endpoint, region, and a checkbox to agree to TOC.

wwwwhaa.... what IS this?

> For example, you can click the button to unlock or start a car, open your garage door, call a cab, call your spouse or a customer service representative

I'm so confused. So now, to open the garage door, I buy an internet device, and when i click it, it sends a message to a server, charges me 0.02$, runs some kind of "cloud" script that executes some kind of "rules", registers something in a database, and sends a message back to my garage door opener?

The future is weird.

Maybe the future they envision for you is to be able to open your garage door while you are not in its proximity. Or to let your car open the garage door on its own, after you summon it. Who knows, soon enough, the cars will have a life of their own, going on nightly strolls and self mechanic check-ups. :)

There's no need to involve remote servers for your self-driving car to open your garage door. It can do it the same way you currently do, with an RF signal.

I'm sure we can concoct some contrived examples where one needs to open a garage door without being present, but I really think this is more about the big tech companies trying to get vendor lock-in.

The future is trackable by them.

It does seem a little odd. Why not a button on your phone. Or your Alexa. Or why have a button at all. The only possible reason why they are promoting this is to get people to understand the code and architecture and program it into other things or for hobbyist projects.

I use https://www.facebook.com/clickotrigger/ which is bluetooth, has internet connectivity as an option. Sadly the company appears to be dead now. Was cheaper than buying a replacement garage door key fob

"This item, sold by Amazon.com, is currently reserved exclusively for Prime members."

So my price just went up from $20 to $120. There goes that.

Something similar: https://bt.tn/

Better Alternatives ( from comments and searching on google):

Amazon Dash button 5 $ - for shopping on amazon, hacks found for ifttt integration ( integrated with buying on amazon) : http://www.amazon.com/b?node=10667898011

https://www.particle.io/button 20 $ , integrated with ifttt

Product: ESP8266 => DIY 7$ IFTTT ( https://www.hackster.io/noelportugal/esp8266-ifttt-easy-butt... , https://hackaday.io/project/5673-esp8266-dash-button , http://www.instructables.com/id/Emergency-Button/ , https://github.com/sam1am/IFTTTButton)

also for 20$ You can buy a basic android phone with infinitely many more buttons & use cases

>https://www.particle.io/button 20 $

Since this was mentioned elsewhere in the thread, this is actually $49. With a little work, the $19 Photon could certainly do the same.

A lot of complaining about the $0.02/press (no recharge/replace on battery) but all the other buttons I've seen cost significantly more and this feels a little more plug&play. I'll probably wait for it to get cheaper or get a replaceable battery as I don't have a great use case for it at this time but it's at the top of my list if I need something like this.

I know it's not as convenient as buying one of these, but it's really not very hard to make your own button, and it's a very fun project. Here's mine, which is also open source, including the PCB and things:

https://www.stavros.io/posts/emergency-food-button/

FWIW, if you don't need hardware, you can turn your phone into such a "button". Apps allow you to add a widget to your home screen and can be configured to make HTTP requests.

https://play.google.com/store/apps/details?id=ch.rmy.android...

in case anyone is interested I wrote a node.js module a while back to repurpose dash buttons

not as elegant as their official IoT button, but it works (and dash buttons are 25% of the price)

https://github.com/hortinstein/node-dash-button

edit: corrected my tired beer math

> and dash buttons are 4 times cheaper

If they were four times cheaper, they'd be -$60. I think you meant they're 25% the price.

That's not how fractions work.

It's much less polished, but instead of using pcap, I just configure dnsmasq on openwrt to call a shell script that uses wget to call an http endpoint on my computer:

https://github.com/bgw/dash-button/blob/master/openwrt.sh

Are there alternatives to this? It sadly seems like amazon has no plans to release it internationally anytime soon. I'm looking for a simple button that can just ping a URL in the simplest case.

I'm looking at a raspi + button hardware right now but something already packaged with wifi would be a lot easier to use.

ESP8266 is what you want.

Here are two DIY projects based on the ESP8266. Maybe somebody makes a similar kit as well, but I didn't find anything quickly.

https://www.hackster.io/noelportugal/esp8266-ifttt-easy-butt...

https://hackaday.io/project/5673-esp8266-dash-button

particle photon. Does input and output via GPIO just like raspi but with wifi onboard. https://store.particle.io/

The photon is way too expensive for these things. What you want is an ESP8266.

Here's mine:

https://www.stavros.io/posts/emergency-food-button/

You can also directly order a PCB, if you want, although you'll have to solder the components yourself.

Sadly, there's no affordable simple button that exists that matches what you're looking for. I've been hunting for it for 6+ months.

Competitors that I've been considering:

* https://bt.tn/

* https://www.particle.io/button

* https://flic.io/ (would have to be paired with a nearby Pi)

* http://www.dropletlife.com/ (not yet available)

My best recommendation would be build http://makeit.netflix.com/the-switch or go with the Particle option.

So it's about $0.02 per click. That is a bit expensive.

Have they updated the branded ones so you can't reuse them? otherwise why waste $20 when you can get one of the original branded versions for $5.

$20 is stupid https://www.google.com/search?q=dash+button+hack

bt.tn: 69 EUR flic: 34 USD AWS IoT Button: 19.95 USD

The branded ones are presumably subsidized by the brands they're labelled with, and getting them to do whatever you want is a fairly hacky process to my knowledge.

$20 is pretty comparable to most of the IoT "do anything when pressed" stuff I've seen on Kickstarter and elsewhere. It'd be a handy "arm my security system" sort of thing for me.

What are examples of those "do anything when pressed" Kickstarters? All I know of is Flic.

Flic, Particle, 1btn, bt.tn, etc.

My network is administered by the building, both at home and at work, so I wouldn't be able to change router config.

I really hope they haven't patented this. It doesn't seem hard to compete against it or build your own - an ESP8266 module costs about $6 and makes it super easy to hack your own "IoT button".

ESP8266 is around 2$ (even cheaper if you buy only the chip) and an nodemcu/wemos devkit around 4$.

I'd say Texas Instruments' Sensortag gives you much more flexibility. It has lots of sensors and 2 buttons.

It's a $19 button that lasts ~1000 presses and can be used to trigger events on Lambda (which are free for the first million, so no real issue). So what you're getting is a cost of $0.019 per press.

Compared to a button on a webapp (which can be pressed from your phone) that, when pressed, goes through API Gateway and then hits up Lambda? At which point you're paying $0.0000035 per press.

Decisions, decisions...

Will you be permanently mounting your phone where you want to trigger these events?

Of course. It's called my pocket. Did it ages ago.

I have already (effectively) permanently mounted my phone on me.

Some progress has been made at reverse-engineering the hardware, which is very similar to that found in the Particle Photon. At $20, the Photon is a clear winner over the IoT button. As soon as someone puts the pieces together so we can flash the Dash's firmware, that's when things get interesting.

https://community.particle.io/t/amazon-dash-anyone-hacked-it...

I wrote a blog post [1] several months ago on how to use these buttons to call Uber. Happy hacking!

[1] https://medium.com/@geoffrey___/summon-uber-with-the-new-ama...

This is a great gateway drug to learning & using AWS; smart move by Amazon.

Bummer, can't order this from Australia. Really want to get one of these for some testing. Anyone with inside info as to when these will be available outside the USA?

Sure you can: http://www.borderlinx.com/AU/en/

Could not find it in search. Also looks like it is for Amazon Prime members only.

Borderlinx chat person said: "You are welcome. We can ship the item without any issue." but also "Regarding the Amazon prime membership issue please be advised to check further with merchant."

This is the product page: http://www.amazon.com/AWS-IoT-Button-Limited-Programmable/dp...

Ask http://priceusa.com.au - I'm fairly sure their agent (in Oregon) uses Prime. Probably a ToS violation, though.

Yup, I was looking for a way to realize this: http://make-everything-ok.com/

"Yo"[1] hardware.

[1] - https://www.justyo.co/

At $20, who would by this?

Has anyone seen a DYI that's cheap?

I bought a couple as the main value for me is that it's tied directly to AWS Lambda. Yes, I could build an ESP8266, RaspberryPi, or Arduino-based solution, but then it's totally custom. Because my startup is involved in "serverless" computing, it's beneficial for me to say that this button directly calls an AWS Lambda function out of the box.

I'll replace the battery with a DC transformer and use one for demos.

Not following why exactly you've decided that being "serverless" or using Lambda makes this worth it. What exactly is IO Pipe (your startup) and what would a demo using Amazon's button do?

We're building open source tools and services for developers to build and manage the lifecycle of serverless applications. The AWS IoT Button works as an event-loop for a (simple) serverless workflow. We can use it to demonstrate and validate one of our use-cases and to provide demonstration of our tools (push the buttons and see what happens).

The concept of running our code directly from IoT devices is something we've found pretty compelling but until now it was very much a DIY affair. It's exciting to see Amazon release hardware that does this out of the box, directly running our code, as basic as it might be.

I would love a Zapier integration here

Apparently the button supports SNS, which can be integrated with Zapier.

Why US only?!

we can now have one "click" deploy's.

That was my first thought as well! How much more satisfying than clicking a link in a UI.

I read the manual, hoping to see the punchline. This isn't ... real ... is it?

People were hacking their Dash buttons to do this, and amazon was already producing the hardware and software, so why not monetize it?

Reply on news.ycombinator.com