Notkia: Linux phone in the shape of Nokia, with LoRa + WiFI + BT connectivity | Hacker News Reader

Notkia: Linux phone in the shape of Nokia, with LoRa + WiFI + BT connectivity

hackster.io

188 pointsddtaylor68 comments

What I most want is new open source hardware guts with Linux for shells like the E61 and Blackberry.

https://en.wikipedia.org/wiki/Nokia_E61

There are several blackberry keyboard projects out there. I've been on a similar journey to OP for a while. Currently I've got a phone I'm fiddling with putting a jetson xavier inside a rotary phone and another based off the featherwing with a blackberry keyboard and touch display and a feather compatible linux SOC (GiantBoard).

I recently got a ClockworkPi DevTerm and have been so pleased with it that I'm seriously considering their GameShell as a phone platform. It's modularity is great.

I'm so excited that other people are working on DIY "phones". I personally think we are going to see this scene keep growing and phones will finally be more like the custom computer market very soon. The fact that people are finally putting linux on apple devices is also very very encouraging. There are so many bricked iphones and ipads floating around with great displays and sensors that are totally useless because we can't run a reasonable OS.

https://www.adafruit.com/product/4818

https://www.crowdsupply.com/groboards/giant-board

https://www.clockworkpi.com/product-page/devterm-kit-a06-ser...

This device with both large screen and physical keyboard is still smaller than every Android device available currently on the market.

The Nokia N900 was the perfect form factor with Linux already running on it. Too bad the Neo900 stalled :(

The N900 was everything I ever wanted. :(

Same. I'd go back to it in an instant if it had a 4G network connection.

The parts are available:

https://www.aliexpress.com/wholesale?catId=0&initiative_id=S...

https://www.aliexpress.com/premium/blackberry-q10.html?d=y&o...

https://www.aliexpress.com/item/1005002813523676.html?spm=a2...

Arturo has done alot already:

https://twitter.com/arturo182

The keyboard is analysed:

https://github.com/arturo182/BBQ10KBD

A machine has been designed, but it does not have the goal of fitting inside the cases linked to above - it's on a bare board:

https://www.youtube.com/watch?v=wr3DtQQuUfI

It's available as a product including source code and design files - again, not designed to fit into the cases above.

https://www.tindie.com/products/arturo182/keyboard-featherwi...

Revision 2:

https://www.solder.party/docs/keyboard-featherwing/rev2/

And the Blackberry keyboard standalone:

https://www.solder.party/docs/bbq20kbd/

Design files:

https://github.com/solderparty/keyboard_featherwing_hw

Source code:

https://github.com/solderparty/bbq10kbd_i2c_sw

All that is left is for someone to do the same, with the aim of fitting into the cases listed above.

The connector is extremely hard to solder. They are over $1 each and so far I have destroyed 7 already.

I have tried via hot air, heated plate to hand soldering with the iron. The hot air will melt the connector so that won't work unless you heat from the bottom the the PCB. Hand soldering is also almost impossible because you can't really get to the pins and if you just slightly touch the plastic it will melt immediately. I also could not get it to work with solder paste and a hot plate. The pins are so small they don't want to adhere to the PCB. I may need to get gold plates pcbs and a stencil to place the exact amount of paste I need to get this to work. A QFN package is easy to solder in comparison.

You almost definitely would benefit from getting a stencil. Instead of a hot plate you could do a toaster oven. I got an "air frier" convection toaster oven at walmart for $50 open box. It's perfect for reflow. Also take one that you've already ruined and test the melt temp on the plastic them put a thermocouple probe on the plastic.

Is there any particular reason you can't use the featherwing package that adafruit sells? I've seen a few other folks on tindie selling other board and screen combos but personally, for my money, the feather format is pretty good for almost anything I could want to do in this form factor.

Isn't the connector meant to go into a socket?

Sorry, I mean the socket.

Yes please! I miss real buttons and the tiny blackberry form factor so much

I do too, but I really don't miss the tiny, non-tactile screen. A modern Linux smartphone with the form factor of the Nokia E90 Communicator or the N900 would make me ecstatic, though.

A long, long time ago, I did this:

https://www.flickr.com/photos/ruicarmo/16931940010/in/album-...

We tested phones to the limit and hacked them when we could, so I just casually left it on a table and a couple of colleagues nearly had a fit until they realized it was… just a GIF with a blinking cursor.

"This design is only patented in US and AU, and will expire on 2023-06-23. After it's expired, we can produce brand new plastic cases based on this design, assuming the fundraising collected enough money."

I guess it's only a year away, but I'm curious why one wouldn't just design a case too at this point. Would fix this issue: "...240x320 IPS LCD, and it's a bit larger than the original one. So, some pixels are hidden by the plastic borders, yielding a visible space of ~220x280 pixels." (Edit: 20% of the screen is hidden even though it fits in the case!)

As well as the issue with the charge port not being quite right.

Wouldn't they also potentially have copyright issues exactly replicating the original design?

I can't say for sure, but one might think so!

Then there's WiPhone [0] which was previously talked about a few years ago [1]. A LoRa module is sold separately as an addon too.

[0] https://wiphone.io/

[1] https://news.ycombinator.com/item?id=19241484

hackster link: https://www.hackster.io/reimunotmoe/notkia-f6e772

Ok, we've changed to that from https://gadgettendency.com/an-enthusiast-turned-a-nokia-1680.... Thanks!

> Note: This is NOT a custom ROM for Nokia phones. It is a freshly designed PCB with exactly the same dimensions as the original PCB of the Nokia phone, so it can be put in the Nokia's shell.

Thanks. The headline was a bit misleading.

Indeed. This way you can run anything 'on' an existing device by... replacing the entire device and just using the shell. Kinda useless from a repurposing perspective.

I like this idea. I have been recently worried about the hacker unfriendly nature of the phones as well. It is frustrating to go through a series of painful processes like unlocking the bootloader, installing a recovery image, if you are lucky enough to find a compatible one, hoping that the phone will not be bricked or end up in a bootloop, etc. Having said that, I would have liked a powerful phone where we can use the desktop version of banking websites, YouTube etc.

Please someone, sell me a feature phone with an open source OS that can run Signal. I don't need a web browser, but at least I want to be able to communicate securely with family and friends. Almost all of them have switched so Signal.

The Signal project doesn't want non-official clients to connect to their network. I doubt you'll get them to release an official client for your bespoke dumbphone OS, so I'm afraid you'll have to accept unstable hackery as a way to run signal.

You may be able to create your own dumbphone by flashing something like PostmarketOS to a compatible second hand phone combined with a custom Signal client. Just apk remove the browser and any other utilities you don't need and even the librem phone might be fast enough for your use case.

> The Signal project doesn't want non-official clients to connect to their network.

They don't encourage it but they don't ban it either. Non-official clients absolutely exist for the network, some of which use signald, a backend service and abstraction layer for the signal protocol which is neither unstable nor a hack:

https://signald.org/

They have shut down third party clients, and reserve the right to continue that. Both use f the name signal, and connecting to their backend without it being their official client are the reasons given, and those seem to apply to signald just as much.

https://github.com/LibreSignal/LibreSignal/issues/37#issueco...

Given that happened six years ago and multiple unofficial clients continue to exist without threat of ban, I'd suggest this one example cannot be extrapolated.

It is true that Whisper Systems runs the network and can gatekeep it as they see fit.

But your claim that Signal, as a matter of course, bans unofficial clients is objectively not true.

Moxie has been pretty consistent on this point. He even gave a talk at C3 a couple years ago that was basically an hour long explanation for why he continued to think this way and that he hadn't changed his mind at all. https://media.ccc.de/v/36c3-11086-the_ecosystem_is_moving

Additionally, they're more or less required to go after trademark violations if they get too big.

> He even gave a talk at C3 a couple years ago that was basically an hour long explanation for why he continued to think this way and that he hadn't changed his mind at all.

In that talk he explains why they don't decentralize their protocol or the ecosystem.

His concern is that a decentralized ecosystem means actively supporting third party implementations and having to achieve consensus when making changes, which slows down (or flat out stops) their ability to evolve the protocol and turn out new features.

That's a totally orthogonal issue to banning unofficial, unsupported clients.

> Additionally, they're more or less required to go after trademark violations if they get too big.

That's a trivial problem to solve: don't put Signal in your client name.

The unofficial clients are part of the ecosystem. Centralizing the ecosystem means centralizing the back ends and the clients.

It's not orthogonal at all, and he's talked at length about how he views every client connected to their servers as under his control, and his dislike for unofficial clients, and his willingness to squash them if they get too big.

Before signal he was head of security at Twitter which has a similar 'squash unofficial clients if they get too big' policy.

> That's a trivial problem to solve: don't put Signal in your client name.

Signald has signal in it's client name.

Unless there’s an official statement about unofficial clients being allowed, then that single example can and must be extrapolated: it’s the only factual evidence.

oh wow HN is parroting this bullshit again.

Moxie and Signal didn't do anything other than asking LibreSignal politely to stop using their servers. They didn't "shut down" anything, and they have shown no interest in shutting down signal-cli/signald.

MobileCoin, the cryptocurrency that Signal incorporates and essentially sanctions, uses signald for their bot[0].

the Signal Matrix bridge has hundreds if not thousands of open source users, plus Element One and Beeper users, all using signald and not having any repercussions for years. The only 3rd party Signal client that's caught any flak from Signal that I'm aware of is LibreSignal.

also see: last time this came up[1]

[0] https://github.com/mobilecoinofficial/mobot/blob/dev/RUN.md

[1] https://news.ycombinator.com/item?id=27682490#27683374

Literally in that thread you cite as [1] is an additional example of a third party client being removed both from google play and github.

Would the donor behind the Signal Foundation, Brian Acton, one of the WhatsApp founders who disagreed with the Mark Zuckerberg model of the internet, support the use of his donation to make threats or engage in legal proceedings against "non-official clients". Are there any statements from him about "non-official clients". Mike Benham ("Moxie Marlinspike") is no longer running "Signal Messenger LLC".1 Signal Foundation and its subsidiary Signal Messenger LLC appear to be the entities that control the Signal trademark. For example, the US service mark "Signal" is now registered to "Signal Technology Foundation", not "Whisper Systems LLC".

1. https://www.theregister.com/2022/01/11/signal_ceo_moxie_marl...

Download link:

https://gitlab.com/signald/signald/-/archive/main/signald-ma...

or https://web.archive.org/web/20210204055922if_/https://gitlab...

I wouldn't call it unstable hackery. There's a library called signal-cli that's being used to develop a native desktop app for signal (as opposed to the absolutely horrible electron web app available officially). The server software is open source already, so it isn't hard to develop a to spec working client that only spoofs identifiers to prevent detection.

I never trusted signal.

No unofficial clients + us-based? Smells fishy.

I’d rather run Telegram and/or Matrix (ideally Matrix).

Punkt MP02 might meet your needs. https://www.punkt.ch/en/products/mp02-4g-mobile-phone/

I like the idea but that's a lot of money for very limited features. I understand the lack of features is the point, but a Nokia 6300 4G KaiOS phone is only $70. No Signal, but it has WhatsApp.

Do you still seriously need to give out your phone number to use it? Seems shady.

Can you just disagree with the design choice rather than implying, without evidence, nefarious motivations?

Ongoing work is being done to enable Signal accounts without requiring a phone number (there's been a number of changes to the protocol to support this in the last 6-12 months), it's just not done yet.

Design choice should be "no need to trust". If not, nefarious motivations should be assumed, unless proven otherwise.

I have evidence. “Moxie” is simply not a person I would trust, for various reasons.

How would you identify your phone otherwise?

MAC address, randomised by the pocket-sized computer owner as needed. Perhaps laypersons are less able to distinguish computer networks from telephone networks than they were 20-30 years ago. Signal relies on a computer network. Signal runs on "personal computers" that are bundled with "phones", but not on phones that lack personal computers. Perhaps one day we will reach the point where "VoIP" becomes meaningless because all voice will be carried over IP.

Why do you need to identify your phone to use an online messaging service?

By a random ID as others [1] do?

[1] Threema

With a username and password

Same but replace Signal with XMPP, which is far more flexible, has been around for a longer time and it's federated.

But yeah, I agree. We need a feature phone with IM support. That would be perfect.

Mutida is open source but with no build instructions or information on how to hack on the device. Read the source, but don't modify it.

Since they don't implement apps, there's no way to get a secure messaging system in there as SMS will be you only option. I suppose you could set up OTR or PGP, but you'd still have no built-in ability to encrypt and decrypt.

Buy a Pixel 5a and install GrapheneOS on it. Install signal and disable the rest of the apps. Done.

If it can run Signal, then it is pretty much smartphone not feature phone.

There were a bunch of feature phones around 2014 that had an unauthorized (and poorly working) WhatsApp client. Signal of today is a bit trickier to run than WhatsApp of 2014, but it could be done. WhatsApp also ran officially on Nokia Series 40, which was a feature phone platform with apps.

There's also KaiOS which is running on phones marketted as feature phones, although it's a fork of FirefoxOS which is Android base plus an HTML+js app model (or something).

I'm not sure what you're basing this on. Java runtimes have been on feature phones for far longer than we've had smartphones. The only reason why we didn't get more capable feature phones around the time we got 3G networks is because it's the same time we got smartphones with big screens and simple touch-based UIs.

The OP seems to ask for only signal, not the possibility to install any other app. I would call that very much a feature phone.

Depends on how you use those terms. If we're talking about processing power and how sophisticated the underlying tech has to be, then yes I'd tend to agree that the ability to run modern-ish internet-connected apps is at the equivalent of a smart phone. On the other hand, if "feature phone" is about the way the user interacts - a phone that ships with a give set of features and is not intended to run arbitrary apps - then it doesn't matter how sophisticated it is under the hood, only that it bakes in certain apps/features (phone, SMS, Signal) and that's it.

The signal protocol itself should run on a microcontroller. Showing images or videos would use more compute power.

I really like the choice of LoRa instead of 3GPP stuff.

I wonder why they went with MIPS instead of ARM...

though one can use any old Android that does Wifi to do SIP, it has its charms to use less capable hardware for this with less software complexity.

I don't know that many projects of this category: the "ZeroPhone" was based on a rpi-zero and active 2017-2019, the "WiPhone", a ESP32/Arduino based device, is still ongoing.

Suggested title change: fitting a linux computer into a Nokia 1680 shell. @dang

Yes, this looked very interesting but it falls a bit short:

> “The device can be used as a computer, walkie-talkie or modem. But Notkia does not know how to make phone calls. The fact is that Remu NotMoe could not find a sufficiently compact 4G LTE module.”

It’s still neat, but it’s not a phone.

https://www.aliexpress.com/item/1005003357030675.html looks pretty compact.

It can easilly be a voip phone.

Mail the mods if you want something changed since there are no @dangotifications

> To do this, he created a special printed circuit board with a processor, memory and other components. But the keyboard is completely original.

So, it is just a carcass?

Reply on news.ycombinator.com