Hacker News new | past | comments | ask | show | jobs | submit login
Purism Librem 5 Hardware Schematics (puri.sm)
104 points by blendergeek 2 days ago | hide | past | web | favorite | 49 comments

(Not a criticism of purism but of how little is required to say "Open Hardware" in general). I have to say how disappointing it is what counts as open hardware. A simple schematic and some xrays? Why not the Kicad(I'm pretty sure that's what they said they used to make it) files, or maybe an openboardview [1] .brd file that would actually help for device repair and hacking on the phone at a hardware level.

As a comparison to FOSS this is the equivalent of releasing a binary and some header files.

[1] https://openboardview.org/

>I'm pretty sure that's what they said they used to make it

That was said about the devkit.

Gerber files are supposed to be released in a few years, after recouping development costs.

I sure hope this is true, and it would be a show of good faith to give it to anyone who asks that also signs an NDA or something until the development costs are recouped so people can at least repair them easily.

I don’t know what EDA software was used to produce the schematics, but it certainly wasn’t KiCad.

Also from the overall style it looks to me like the schematics actually were created by some east asian ODM and not by Puri.sm themselves.

How can you tell? I'm curious as to which specific details you're referring to.

Pretty interesting, I hope they make something similar for the laptops. I use an x200 with Libreboot to avoid the proprietary software in new Intel chips and the bios. It's still fast since I use Arch with dwm, but at some point in the future I would like to switch.

> I hope they make something similar for the laptops

From https://puri.sm/faq/: "We cannot share the schematics nor design currently because it is copyright encumbered from Intel reference designs. In the future, on future versions we plan to be able to release those."

> x200 with Libreboot

How safe are those now that the Intel vulnerabilities are known? I doubt hardware that old is getting microcode updates.

You can disable hyperthreading to mitigate a lot of Intel vulnerabilities. Qubes, which is a popular OS choice within the coreboot crowd, does this by default.

But not all of them, right? Some of the vulnerabilities do require upstream microcode updates, and without them, I feel like the risk introduced outweighs any benefit provided by Libreboot.

Hi Luke M Smith, how's it going? I love your channel.

So far I have heard of maybe one end user with a phone IRL. Are there any examples of non-promoters / paid reviewers / internal employees (like plain ole' end users) showing off the phone?

Can confirm, I am not a promoter / paid reviewer / internal employee.

You don't report the weight...

My garbage kitchen scale says 231 g.

Thank you, good enough!

What I find strange that they still don't have specs for the dimensions and weight... That is something they have been very coy about...


I suspect this has to do with the current scheme of shipping out the devices in iterative "batches." Presumably they are assuming the early batches will be heavy and clunky, with size and weight of each batch decreasing over time as they learn more about manufacturing.

I find this more troubling than the idea of early batches shipping with pre-beta quality software. At least improvements to software can be pushed out to older devices. If you buy an early batch and get a phone the size and weight of a brick, though, that phone is always going to be the size and weight of a brick.

You have to opt in to the early batches and don't they post photos showing how ugly it is? It seems like customers are informed about what they're getting.

review of an early phone: "The phone is chonky. I measure it at 15.6 mm thick, 150.2 mm tall, and 75.2 mm wide."


I assume that the specs will change batch to batch.

for comparison:

iPhone XS: 7.5mm x 143.6 x 70.9

iPhone XS max: 7.7mm x 157.5 x 77.4

Also from the same person, Azdle, further up in this thread:

> My garbage kitchen scale says 231 g.

Nice move from puri.sm

At page 15 and 16 there are some M.2 connectors. One has an USB line named USB2_WIFI_D* and the other USB2_4G_D* so I guess these are used for 2 external radio modules (WIFI/BT and 4G).

I wonder what these modules are. Will it ship with the same from the devkit (RedPine WiFi/BT M.2 module on SDIO, SIMCom SIM7100E)? Pretty important from a security/privacy standpoint.

The WiFi/BT module is the same. The modem module is BroadMobi BM818; at later batches there will be an option to get Cinterion PLS8 instead too.

Was the title changed to get around this exact story being posted yesterday?[0] This title is wrong and is not the schematics but a blog post announcing the release of the Birch schematics. The correct title is “A Different Kind of Transparency”

[0] https://news.ycombinator.com/item?id=21690269

It doesn't count as a dupe if the story hasn't had significant attention yet. This is in the FAQ: https://news.ycombinator.com/newsfaq.html

Ah, I had forgotten about this. Thanks!

Calls don't work yet it seems

The reviews said the speaker isn't wired up. I wonder if you can call with the headphone jack?

It's purely software. An update with audio routing will be out likely this week (it's in testing, and you can already get it if you compile stuff yourself).

My understanding is that the microphone (& speakers?) isn't wired directly to the modem, but needs the CPU to replay the audio to the modem so that the modem (aka, baseband) doesn't have access to the microphone audio except when it actually needs it. So, I would think that the headphone mic wouldn't work either.

This is good for privacy, but can this way of not wiring the microphone directly would also be helpful to do recordings and to send/receive faxes, or does that not work?

My understanding is that, in general, it's not possible to send a fax via cell phone audio because the digital compression applied does not pass enough information to sustain a fax modem connection. It may be possible with a higher-bandwidth thing like voice-over-LTE, though? (of course I'd love to be wrong and learn that this is easy :-)

CSD data calls are perfectly possible in GSM - your carrier can disallow them though. However, you usually feed and receive data to/from the modem in a different way than via its audio interface ;) Other than that, you're right. It's possible to come up with a way to pass data through GSM codec, but it definitely won't be a regular FAX connection then.

Normal phones can record phone calls, right? I assume that this wouldn't make anything possible that isn't already possible. But, again, I really don't know what I'm talking about here, this is just my understanding from what I've read.

Before I bought a Xiaomi it was impossible to record any call (Samsung, HTC) and it seems it was by design. The only way was installing an app that somehow routed the recording through their servers (no, thanks) otherwise you could only record outgoing (or was it incoming?) sound.

Usually phones have their modems tightly integrated within SoCs, so audio is both directly connected to the modem and accessible to the CPU.

Why has it become so acceptable to launch half-baked products that don't actually work? Twenty years ago, you launched a product and it had to work right out of the box. What's gone wrong here?

It's a crowdfunded project where some of the backers opted in to receive a device from an early batch. What's so unacceptable about that?

And they've been pretty clear that this batch is expected to have issues resolvable in software. I'm sure the audio issue for calls will be fixed in the next few weeks, along with some improvements to battery life and whatnot.

If you don't want a pre-production product, don't opt in for it and wait for Evergreen.

Some of the customers for this phone are developers who want to contribute to its development. And some customers are so interested in it that they would prefer to receive the half-baked version instead of waiting.


Why would they bother?

Arduino, Trezor, and 3D printers (I forget the name) have all been cloned. It's a real concern for companies that are trying to fund open hardware R&D by marking up the hardware.

Currently I doubt there's enough demand for the Librem phone to justify cloning though.

Well Prusa, the creator of the most cloned and world most popular 3D printer design is happy about all the clones. And they occupy different spaces in the market. Clones are $250 and the Prusa is $1000. And yet business is booming. As an open hardware creator myself my hope is always that someone else better at manufacturing will clone my work so I can source from them and focus on whatever reasons I had for making the hardware.

Also, Librem 5 schematics are GPL, so it would be sort of weird to make copying into something negative, while using the copyleft license.

Don't get my hopes up.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact