DIY cellphone | Hacker News Reader

DIY cellphone

hlt.media.mit.edu

183 pointsryanmolden56 comments

"Cellular connectivity is provided by the SM5100B GSM Module...". Meh. No way a DYI. It's like you assemble your desktop PC from pieces like motherboard and power supply and say that you built a DYI PC.

You did in that case. It'd be ridiculous to try and build a cellphone, the not use the all-in-one chip solutions that are out there.

Actually it is not ridiculous. There is a project which is much more a DYI then this one. Enjoy: http://wush.net/trac/rangepublic (OpenBTS).

Same with osmocom.org (their site is down atm. though), but they've among many things built a GSM base band stack, running on real phones.

While certainly not intended for every day use, it's very handy to atleast have something for research and learning where you can access all the protocol layers. Just imagine the TCP/IP world if no one could ever send/receive any packet/frame/bits/protocol below the application layer.(i.e. TCP/UDP/IP/Ethernet was just a black box on a silicon chip).

That's basically the GSM world, at least until the osmocom guys made their own stack, running on cheap and old nokia phones.

The all in one chip virtually is the cell phone. Its like buying a house, fitting the doors and windows and claiming you built the house.

I cant see this as DIY at all, but equally, its not a bought phone. More kit phone than DIY phone.

Sure, but it would be a cool project to build up a nice phone environment using this module.

If anything, this would be more akin to saying "I'm using hardware features on my microprocs to do this neat thing". Not amazing engineering.

On a related note, does anyone know if we can get these modules to do data? Or if there's a cheaper version of these modules?

I did a quick peek at their electronics schematics. Well, believe it or not, but it looks like designing a laser-cut custom wooden case for this phone took much more effort than the electronics part. So, returning to my analogy: you are building a DYI PC because you laser-cut a custom case for the off-the-shelf parts you purchased. If this is the level of DYI to be amazed of on this site, I'm completely confused.

I posted this not so much because I was amazed by the thing itself or thought it represented some pinnacle of "true" DIY, but more so because I like the idea of making things that were once far out of reach of anyone but serious hackers into something that a broader collection of people could feasibly tackle. It isn't quite there yet, but it is due to people like this that it will be, one day. You could nitpick all day about how most anything isn't truly DIY unless you mine/refine all your own materials and create your electronics from first principles, but that starts to get a bit silly.

More generally, I am a believer in the idea I first read from Clay Shirky that as more people get involved in an area average quality goes down but innovation goes up a lot as you get lots of people trying lots of new things, which in the end is only goodness. Except, perhaps, for those that prefered it remain elite so they could feel special (not saying this is you, but I see this attitude in a lot of pioneers in areas that later become more mainstream).

The GSM module supports GPRS and TCP/IP via AT commands[1] so it should be possible to get some data going.

[1]https://www.sparkfun.com/datasheets/Cellular%20Modules/CEL-0...

Totally possible. I've built telemetry systems out of this family of devices. However rather than using the AT TCP stack, you're much better using the standard GPRS PPP mechanism that most USB cell modems use so that you can use your chip's or OS's TCP stack instead.

For example device, see http://www.windowsfordevices.com/c/a/News/Opensource-hardwar...

For PPP info, see http://ernstagn.home.xs4all.nl/GPRS-HOWTO/

[EDIT:] It's been a while, but Telit was also pushing their embedded python interpreter pretty hard back when I was into all of this. I had need for an embedded micro for other parts of my application so I never bothered too much with it, but it looked pretty interesting.

I would suggest the ADH8066 GSM module instead, which also includes a SIM socket on the back, and makes things a lot easier. And it's cheaper too.

Edit: And you can easily use it for data.

I don't think this is about an individual's amazing engineering effort so much as it's about the ability for an individual to leverage the amazing engineering efforts of huge teams in the form of low-cost, low-quantity integrated modules.

At least, that's what I find impressive about it.

Since it seems to be backed by MIT, I also kinda hoped to see an independent GSM stack and not just some COTS baseband attached to a COTS MCU.

To use it legally on real networks, that independent stack would have to be validated by the FCC, which is a big obstacle for a DIY project.

I'm a former DIY Cellphone hacker, and builder of several commercial GSM telemetry systems... Even using an integrated module you'll still run into FCC certification issues. Then there's PTCRB certs, CTIA membership, and carrier certifications. Of course even if you have the money you'll realize that all of these things have been done to the integrated module (which you haven't modified) before it ever hit the market.

OpenCell - my 2008 entry into the DIY cell phone game: http://web.archive.org/web/20080709054225/http://www.widgetr...

Fortunately, someone stopped them from re-inventing wheels, axles, GSM stacks, etc.

GSM is the definition of phone. Why not just call it "DIY handset" if that is what you are really doing yourself?

True, but PC assembly vs black box computers spawned an entire industry and and have kept component prices low even for end consumers. I had to replace a hard disk recently and am still bemused by my ability to purchase a new server-grade 2TB drive for $125.

DIY

As the person who made the phone, I think this is an interesting question.

Do we say that Apple doesn't build computers themselves because they buy the CPU from Intel? Or because they use Qualcomm transceivers (and lots of other off-the-shelf components) in their phones? On the other hand, we probably wouldn't say someone made a phone if they just put it into a different enclosure.

I can say that I spent a lot of time selecting all the components, putting together the schematic, routing the board, and writing the UI software. (All of that took a lot longer than laser-cutting the case.) Of course, it probably took a lot more work than that to create the GSM module. But that's also true of the microcontroller I'm using, the LCD, even the battery. Just about everything is built on top of the work of others (see the Toaster Project for an example of how hard it is to do otherwise).

Interestingly, this critique hasn't really come up from the people I've worked with to build the phone. I explain the function of the GSM module (and the rest of the components) but after putting it all together, people don't seem to think of the phone as just a box for the module -- in part, I'd argue, because they realize how many other things are in there too.

Again, though, there's a fundamental question here that I'm trying to explore with my research: what does it mean to do it for yourself?

Don't listen to these people. This argument is old, and tired.

It's the same as:

"I don't code python, that isn't a real language, it does everything for you, unlike C!"

Really? Does everything for you? Good to know that you built your own transistors from scratch via materials that you mined out of the ground, while naked, using only tools that you scavenged out of the forest, and that you came up with the entire process yourself.

Seriously, this is cool.

Well said.

Who cares, phones are interesting to people because they enable communication with others. Phone projects like this are interesting because they enable people who wouldn't ordinarily be able to build a cell phone, to build a cell phone.

"Wait, you said you programmed this yourself, but I noticed that you're using c and some c libraries. That's not DIY! And you're using a compiler. That's also cheating! You should really be flicking toggle switches and pushing a button to enter the hex values directly."

BTW, the latest files are here:

Hardware: https://github.com/damellis/cellphone2hw Software: https://github.com/damellis/cellphone2

How's the reception and microphone quality? (i.e. how well does the sound of your voice on the phone compare to say an iPhone or Nokia-era candybar phone?)

Also, great work! I looked into building a cell phone before, and it's an awesome amount of work.

It apparently sounds pretty good, according to the people I've called. The reception seems pretty standard, although I don't actually display signal strength in the interface yet so I'm not totally sure.

This is what I love about living in the future (and HN). I live just north of Seattle, I read a story last night about this project in a magazine from the UK (New Scientist). I posted it to a website in California and the project creator, from Massachusetts, responded and it turns out he has had an account here for over 4 years.

I know PG wrote an essay about how certain cities act as magnets for smart people looking to change the world, and how it is advantageous for companies to set up in those cities. That always struck me as a rather antiquated system (though I don't deny it is currently true) and I always wondered how long it will be before that changes. The idea that to properly collaborate we have to arrange such that all of our molecules inhabit a relatively small geographic area just seems silly. Things like this really make me appreciate the power of the internet to connect people and propagate interesting ideas (instead of just memes). Makes me think that before I die the idea of necessary co-location in order for people to collaborate effectively will be a bygone notion. I would welcome that change.

I'm surprised that nobody mentioned OpenMoko so far: The attempt to create a truly open sourced mobile phone: http://en.wikipedia.org/wiki/Openmoko

Sadly, it failed and was abandoned in 2009.

Or Silicon Valley Homebrew Mobile Phone Club in 2006. Or TuxPhone or OpenCell (my contribution).

[Edit: Or SparkFun's original claim to fame, the Port-O-Rotary: https://www.sparkfun.com/categories/96]

Personally I like the look of cardboard: http://bit.ly/10K7Qi0 (my first DIY cell phone)

Actually, there is still a small but healthy community working on a few distributions and new hardware [1].

[1] http://projects.goldelico.com/p/openphoenux/

Didn't a marketing company recently put live messages into a magazine page using a single chip gsm solution?

Yeah here it was: http://creativity-online.com/work/cw-live-twitter-feed-in-pr...

I vaguely remember people changing out the gsm sim or making calls on it.

Ah nevermind, it was a fullblown android phone, not a single chip solution: http://mashable.com/2012/10/02/ew-has-smartphone-inside/

My grandmother has pretty bad arthritis. No major carrier sells a handset that is easy for my grandmother to manage. It may be time to put one together for her.

In the UK they sell a range of mobile phones with big buttons, particularly for the elderly. Basic features on it only of course, but that's probably all they will use anyway.

http://easiphone.co.uk/index.php?main_page=index&cPath=1

Basic features on it only of course, but that's probably all they will use anyway.

That's a rather self-fulfilling attitude, and I hope that we as an industry try harder really soon.

My parents are in their late 60s, and are quite good with computers. I'm working on a side project with a woman in her eighties who started programming during the Eisenhower administration. And I know a number of tech folks in their 40s or 50s who have one form or another of muscular or nervous problem. None of them should have to compromise on features just because they want a more accessible device.

I didn't mean it in that sense, my parent are both pretty old but have no idea about tech. For them this phone was perfect, since all they wanted to do was make calls, a green button to make the call and a red button to finish the call.

With a phone like this, there has to be a compromise somewhere. By basic features I meant by today's smartphone standard, you are never going to get that to that level with an LCD screen and button taking up all the phone real estate. It is really only going to be possible for calls and SMS messages.

Of course, if they need a smartphone with big "buttons" and a big screen then you could always buy them the Samsung Galaxy Fucking Note

http://techcrunch.com/2012/10/26/behold-the-best-samsung-gal...

There's a company called MatobMobile who sell a variety of "accessible" handsets from a variety of makers. Doro is one well known brand.

(http://www.matobmobile.co.uk/) EDIT: This website is bafflingly awful considering they're aiming at customers with accessibility needs.

But putting one together is a great idea, especially with 3d printing.

How to shrink a basestation into a Raspberry Pi. http://www.youtube.com/watch?v=GCcKgrzbix4 It's not the handset, but rather the basestation. But the concept is similar. I saw this working. It works. And there are open-source versions of the handset software too, although not as mature as OpenBTS.

Actually what these guys don't mention is that the Raspberry Pi is plugged into a relatively huge USRP, because that kind of blows their whole "Look how tiny it is" PR out of the water.

+5. You can run the whole stack on a router. It's basically doing call control for a couple of channels. The heavy lifting is done in USRP.

The host part of OpenBTS is not just call control(Freeswitch or Asterisk provides most of the call control part).

OpenBTS is a software defined radio and really does most of the processing in software. The USRP provides the radio interface, up/down conversion between baseband and carrier frequencies and the analog/digital conversion.

The USRP just sends/receives those samples without caring or knowing about GSM. Samples are processed in software on the host, where the GSM stack is implemented.

There's other hacks(http://www.youtube.com/watch?v=xFjVcxMpA6c) where you can just use a couple of phones as the transceiver instead of the USRP.

Right, call control with asterisk, and GSM stack in software.

As far as I was aware, the GSM processing isn't much more intensive than equivalent: Ethernet packet assembly + TCP/IP + HTTPS?

I thought all the DSP happened in USRP. So out of that, you get a bit stream, that is sliced into frames, and then processed like any network stack?

No, OpeBTS does the GSM related DSP. You get a stream of I/Q samples from the USRP, which is more or less just the digitalization of the baseband. OpenBTS carves the time slots and "frames" out of this, does the GMSK de-modulation and similar stuff.

Note that the lower layers of GSM is very different from e.g. ethernet. Once you get past all the very gory stuff up to layer 2, the concepts are pretty much the same though.

Can anyone share any details about one of those laser wood cutters? Costs/Materials/Brands/Issues etc. for a hobbyist?

If you're really into DIY, you can make your own CNC laser cutter http://www.instructables.com/id/Laser-cutter-start-slicing-s...

Otherwise, you're talking $5k and up. However these guys are cheap and fast if you want to farm it out: http://www.pololu.com/catalog/product/749 (I don't work for them, and I'm not an affiliate, just a happy customer)

I've heard good things about the Lasersaur, a DIY laser cutter: http://labs.nortd.com/lasersaur/

Also, Ponoko is a decent option for a laser-cutting service: https://www.ponoko.com

I bought a 40 watt laser cutter/engraver from China via ebay. It cost a little less than $1800. It cuts wood and some thin plastics (acrylic) like this quite well.

Unfortunately, its quality was so terrible, it was little more than a box of parts. I had to build a new stepper driver and run it from MACH3 to get any results out of it. I also had to replace some "greased rail" style bearings with linear ball bearings to get decent accuracy.

The one they've got at MIT probably says "Epilog" on the side and costs somewhere between your car and your house.

Edit: I generally cut "luan" plywood 1/8" or thinner. It seems like about the thickest I can handle. 1/16" veneer gives me the best results.

thanks a lot to everyone who replied! Sounds like the 3d printer that i have no use for may come before the laser cutter i have no use for :/

Take a look at FSLaser (http://fslaser.com). Minimum cost for an entry-level system is $2350+shipping.

Disclaimer: I'm an employee.

With all these open cell phone stacks: what do people do about IMEI? Isn't modifying IMEIs illegal in many countries? Or is the IMEI already programmed as part of some module from the factory and thus there's no need to change it?

EDIT: Whoops. I did mean IMEI and not ESN. I've edited this post.

On GSM phones, ESNs don't apply. IMEIs are the equivalent, however. In that case one of the value-adds of an integrated module is that it comes with an IMEI. If you're going into large scale production you can pay CTIA for your own block of IMEIs and reprogram your modules, but typically that's not worth it. However if memory serves (it's been about 5 years) they force you into doing this if you get above a certain production volume, otherwise it becomes difficult tor track which chunk of the traffic is yours, and whether or not you're shipping a ton of devices which haven't passed carrier certification, etc.

I'd imagine it works similarly with ESNs, but I have no experience there.

http://hlt.media.mit.edu/?p=27

This has more possibilities to me. More exhibition with interaction from viewers.

Finally, a karma free phone. I wonder if people will start designing cases made using a 3D printer for the next generations of these DIY cell phones.

I, Pencil: My Family Tree as told to Leonard E. Read:

http://www.econlib.org/library/Essays/rdPncl1.html

It's like Raspberry Pi for cell phones...

Combine this with the Raspberry Pi base station, and there's HackerNet :) http://www.newelectronics.co.uk/electronics-news/basestation...

Reply on news.ycombinator.com