Hacker News Reader

Comment by Accacin | Hacker News Reader

That looks really interesting. Their funding target seems quite high so I'm not entirely convinced that they'll be able to meet it.

I will definitely try and support this project as I'd love to be able to move to a more open phone.

Link to their crowd funding page: https://puri.sm/shop/librem-5/

View on HN

I'm surprised you think that the funding target seems high. I personally believe their funding target is in the range of 10-20x too small to actually make a smart phone, especially when they plan to make all their own software and not simply reuse Android/Sailfish/Ubuntu as a starting point.

I expect that just the certifications for the radios, both for various countries/government agencies and various phone carriers, to cost them many hundreds of thousands of dollars. And that cost has nothing to do with actually designing any of the hardware or software.

There are two ways to interpret their funding goals: based on projected cost and based on projected backers. 1.5 million dollars may not be very much money for smartphone development, but it is very ambitious as a crowdfund goal. It basically requires around 2500 backers ($1.5MM / $600) to pay $600 out of pocket for a device that, at best, will be delivered in 18 months.

Given the success path of most hardware kickstarters, I'd be surprised if this fully funded. The price point is relatively high (even though it is a whole phone, I get that, it is just a lot of money). I'd also be surprised if it actually delivered in 18 months. That's basically saying: "we have the design all done and are in our final mass production engineering phase right now". I mean, if you follow any prior hardware kickstarters that include significant production parts (pebble, keyboardio to name just two), the one major lesson is: mass production takes a ton of time and money to get right.

> the one major lesson is: mass production takes a ton of time and money to get right.

True, but that said, Purism has plenty of experience with this entire process through their laptops. Hopefully the lessons they've learned there will help them a lot here.

I'm very happy with my Librem 13 laptop so far and pre-ordered a phone. I know there's some risk of failure but I'm happy to put my faith in Purism here, based on what I've seen from them so far.

It basically requires around 2500 backers ($1.5MM / $600) to pay $600 out of pocket for a device that, at best, will be delivered in 18 months.

There may also be a handful of people like me who think the idea is cool but don't necessarily want the device (or don't want it right away); I just donated $20.

JackPair is another example that's had tons of hardware issues. Stuff that a new device maker might not see coming. One that does lzptops might also get blindsided moving to mobile where there's extra challenges.

Yeah this is exactly what I meant by my comment.

Don't think the phone is 100% in house. They're using a base OEM model from someone that happens to have opensource drivers for everything (which makes me think it's an older model).

But even just for the software, it seems low.

> They're using a base OEM model from someone that happens to have opensource drivers for everything (which makes me think it's an older model).

I... don't think that's the case here.

In the specifications they state:

CPU separate from Baseband

My reading is that modem is connected to the CPU via USB. The only other project I know of that was looking at doing such a thing were the OpenMoko guys (GTA04 with an OMAP3xxx, I think? My memory is fuzzy on this).

AFIAK, there are no 3G/4G modems with an open source baseband. So the traditional route for libre/privacy conscious people has been to isolate the baseband from the application processor via USB. Normal phones have the baseband and SoC integrated on the same SoC and connected via a high speed serial bus.

On further reading, it's going to be an i.MX6 (because i.MX8 is not available yet and you don't typically use unreleased SoCs for your libre hardware) which is pretty inefficient by modern ARM SoC standards.

So this is essentially a small tablet running Linux with an integrated cellular modem.

"My reading is that modem is connected to the CPU via USB. The only other project I know of that was looking at doing such a thing were the OpenMoko guys (GTA04 with an OMAP3xxx, I think? My memory is fuzzy on this)."

First, modem connected via USB is a good thing - it's the least bad option with regard to closed basebands.

Second, it's not that rare of a configuration - I believe there are quite a few consumer smartphones that connect the baseband via a USB interface ... it's not like there's a little USB connector inside there...

"So this is essentially a small tablet running Linux with an integrated cellular modem."

I don't think that's exactly correct ...

It is indeed possible to stick a USB cellular modem onto a tablet device and even use it for telephony. The problem with this arrangement is that there are many other things (like echo cancellation and other voice enhancements that need realtime processing) the baseband processor does - and you lose those functions.

An integrated USB modem that is actually the integrated baseband of a mobile phone does not have those problems - even though the bus used is USB.

It will most likely be a USB-like connection; HSIC, I suppose. BTW, this is the condition that all Replicant phones have to meet in order to be ported to it. So quite a lot of phones already have such an architecture.

https://redmine.replicant.us/boards/33/topics/7383?r=7569

At least the Nokia N900 used the separate CPU and baseband configuration.

The osmocombb project [0] is an implementation of baseband modem functionality, but I'm not aware of anyone using it. Old TI Calypso modems also have an open source firmware implementation [1].

[0] http://osmocom.org/projects/baseband [1] http://www.freecalypso.org

> So this is essentially a small tablet running Linux with an integrated cellular modem.

This reminds me that the 3G version of the 2012 Nexus 7 tablet had the baseband processor separate from the CPU. It was generally said that this was one of the reasons it had sub-par battery life, but I don't know if that was true.

Yup.

On their site: "will work with 2G/3G/4G, GSM, UMTS, and LTE"

If it's using something like this: http://a.co/9IVPmkI -- then maybe it's using the USB pins on a mini PCIe card.

I think it would be wise to wait until they have committed to a particular cell modem. Most people will not be able to use a phone day to day if it suffers from:

• Dropped calls

• Bad call quality

• Low speed data (anything less than 4G/LTE)

> then maybe it's using the USB pins on a mini PCIe card.

Every mini-PCIe cellular modem I've seen interfaces via USB and not via native PCIe. You can even buy adapters to use a mini-PCIe modem as a USB modem. [0] I don't think this is standardized, but it seems to be the defacto standard for all modems in this form factor.

Now this may change once LTE/5G speeds regularly exceed what USB 2.0 is capable of since mini-PCIe does not provide pins for USB 3.0, the standard is too old. I would guess they'll move to NGFF as that's already the case for newer business notebooks which offer a WWAN slot.

[0] https://www.amazon.com/PCI-E-Adapter-Module-Support-connecto...

This is that link without the reference https://www.amazon.com/dp/B00NPX09QY/

They say the hardware will use an i.MX6 or i.MX8 SoC, which I assume implies they are using the Freescale/NXP SoCs. i.MX6 isn't really intended for use in phones (afaik) and I don't know of any recent phones which use i.MX6 in them. i.MX8 isn't publicly available yet, but possibly they could leverage some other vendor's hardware phone design using i.MX8.

However, based on the schedule they show in the pledge info, they already have the "PureOS I (CPU)" step mostly completed and they plan to be booting the hardware before the end of 2017, so this seems to make i.MX8 unlikely imho.

i.MX6 was used in the kindle I believe, which I would be using right now if there was an easy way to add more storage.

Some Kindles may use i.MX6, likely simply because some i.MX6 versions have an integrated e-paper drive subsystem which avoids the need to use an external e-paper controller. I believe early Kindles used the Epson e-paper drive controller.

Perhaps we'll see an e-ink smartphone other than the recently announced Yotaphone 3!

There isn't that much to write, Matchbox has enough to make X11 touchscreen friendly. Really it's just about all the hardware stuff which they've done before I'm sure.

Seeing as the screenshots are from Gnome, I'd be surprised if they used X11, considering Gnome's future is on Wayland. Not to mention the irony of shipping a "secure phone" with X11...

I still feel like being tied to X would be better than being tied to android. I use X11 on my laptop as do most people I know. Is it really that hard to just not run apps that you don't trust? There isn't that much out there that isn't built by a fairly popular open source project and is worth using.

With X11, an exploit in any GUI application can be turned into a privilege escalation vulnerability. On Wayland, to achieve the same, an exploit in a GUI application would need to exploit another application running with higher privileges to escalate.

Not running apps you don't trust is an overly simplistic security model, and completely ignores exploits.

X has been able to run rootless with KMS-based drivers for a while now.

An exploit in a Wayland app gets you just as far as an exploit in an X app connected to a modern X server.

Reply on news.ycombinator.com