The Long-Awaited Batch Update Post


#21

This announcement gives people a strong reason to wait till the Dogwood batch to get improved cooling.

There was no mention of heat pipes in Birch in the announcement, so I presume that Purism has decided to scrap that idea and instead use a metal sheet that will connect directly to the metal frame to dissipate heat. That will be a lot cheaper than heat pipes, which I understand are quite expensive to implement, because they have to be customized to the phone, so they don’t really make much sense for low-volume production.

I assume what happened is that Purism realized that it needed better cooling, and then it started asking suppliers what it would cost and how long it would take. After hearing the cost and the time to implement heat pipes, it started searching for a cheaper solution. This means redesigning the board to move the i.MX8MQ to the other side, so there will be many changes.

The fact that a graphite sheet isn’t enough cooling shows how hard it is to use a processor like the i.MX8MQ that wasn’t designed for mobile phones. The kernel bugs that Purism is having to solve also show how hard it is to use a new SoC which doesn’t yet have good mainline Linux support.

All this work makes me appreciate just how hard it is to design a phone that is able to run on 100% free software. It would have been much easier for Purism to use a chip that already had good kernel support like Allwinner A64 or Rockchip RK3328.

The other question is whether it would have been better for Purism to announce immediately that it had found problems in its Aspen batch and leave people hanging as to what would happen next, or to wait until it figured out what it was going to do before making an formal announcement. If it had done the former, would people have freaked out even more than they did and would more people have canceled their orders?

Purism caused a lot of speculation when nobody received phones, but there also would have been a lot of speculation if Purism had made an announcement on Sept 24 that it had encountered severe problems and hadn’t decided what to do about it. Purism might also have thought that it could solve the problems within a month, so it could still ship Aspen on time.


#22

True. This was a tough call to make. I hope, it ends up being a better choice.


#23

How bad are the thermal issues going to be for pre-Dogwood batches?

Aspen can clearly run simple apps, but what happens when you push it? Does it overheat and crash? Does it damage the phone? Are the Birch and Chestnut phones just going to be configured with a 20% underclock and work fine?


#24

The connectors in the photos suggest the ability to swap out components?


#25

Or throttle? Throttle may be better than underclock.

I sometimes see “thermal throttling” messages in syslog on my x86 laptop (not a Purism laptop).


#26

I’d really rather they focus on making the phones.


#27

I have had two computers that overheat. I don’t want such a phone! When it says CPU location is not ideal for thermal dissipation… sorry, put me in Dogwood.


#28

If they say there are development issues I think: OK so I wait longer, NO REAL PROBLEM HERE!

If they say Aspen is being delivered and it really isn’t, then everybody thinks: Is this development issue?.. Or is this financial / disgruntled workers? POSSIBLE BIG PROBLEM!

Bottom line: just tell it like it is!


#29

Or just release it and tell people that there is no issue with signal-they are just “holding the phone wrong” :slight_smile: As the Great Messiah suggested w initial iPhone 4 batch
Kidding aside, yes it was a messed up choice with that [in-house] shipping and all.
Also, not sure , who would want to deal with overheating issues. I get the enthusiasm, but ppl plan on keeping this phone way beyond this ridiculous 1-2 yr cycle and I don’t see it age well with heavy use. So, I admire the courage of those who still want the Birch and Chestnut.
It’s possible that these gonna be very small batches as well. As a company, you want as many as possible of polished products in consumer hands, not partially crippled.


#30

That was a great post and I really appreciated hearing about the problems they’ve had. Although I cannot wait to see this phone in person, with the updated batch information, I may like to wait until at least Dogwood (if I were eligible).

I would like them to detail and show images of what each batch will have, as they are ready, so that anyone who is offered that batch, can make an informed decision. Things may still change as they go and if Birch or Chestnut do have heat pipes or something else to dissipate heat (as suggested in one of the videos), it would be good to see some details before making a decision.

I have to agree, it would have been a tough decision to make.


#31

It may be somewhat sound if it will not be connected rigidly (so for example via a thermally conductive rubber pad). Another danger is that cases typically allow for some elastic deformation, even if it’s very small, and also materials expand when heated up, so improper connection between case and SoC may cause trouble due to thermal cycling or during rough handling.


#32

I don’t get it. Am I really the only one actually being deeply disappointed by this Update?

Before they said “People have been using the Librem 5…” - now they say: “oh we intentionally just ‘delivered’ to Todd and a handful of his employees, because this device is not fit for usage”. It’s yet another delay. Which would have been fine by me if they had communicated it like this. But they didn’t. I find this highly deceiptive.

What’s even worse is that they still don’t adress the overheating issue. They want to tackle it with cooling. But this is not how it should be done. This is a mobile, a battery powered device. Heat is electrochemical energy from the battery transformed. The way to tackle this should be to reduce the generation of crazy amounts of heat, which will drain the battery in no time. Instead they let the battery drain and are happy with cooling the wasted energy away so the device doesn’t die.
Thermal design / low-current design is the most integral part of a mobile device and should have been solved from the planning phase on. Now they claim they are ready to deliver, but still don’t seem to have a professional energy design.

And the worst part is the sheer lack of communication or rather the attempt to put the spotlight away from the serious problem (just like zlatan said in his interview). We didn’t even get the chance to help them with the issues, because they kept them under the rug.
And don’t give me the “i’d rather they concentrate on developement, rather than posting updates”. That’s not how a modern company is structured. They alread have marketing/PR-guys who do not participate in developement. And internal reports are surely made, so the PR-guys could easily adjust those reports for public consumption, instead of posting more pictures of kill-switches and coffee. No ressources diverted from developement.


#33

Which is exactly the reason I want chestnut.


#34

I’m a little disappointed. I would have liked a number of units distributed. I would have liked an explanation of why the UBports devs didn’t receive a phone, since they changed Aspen to more of a final prototype. And more clarity on expected batch sizes would have been nice.

Overall though, I’m content with it. Is the delay of Birch shipping a bummer? Sure, but it shows that they can adjust during production (if limited). With the iterative schedule, it looks like fixes for Aspen problems can be incorporated into Chesnut, Birch into Dogwood, so on and so forth. And for PR, there’s more than one way for a “modern company” to operate. Like @amosbatto said, it was really a choice of two bad options. Announce the problems as they arise, or do a summary batch update. Personally, I prefer having an update at the end saying “This is what happened in this batch, this is what we decided, this is how we’re going to fix these issues, and here’s the update roadmap going forward” rather than play-by-play updates.

For me, I’m sticking with Chestnut. I never intended for this to be a drop in replacement for my android, and it will take time for me to adjust and get everything (matrix bridges to different apps, outright replacements for others) to daily driver state. I had always planned on having two phones, and then buying a newer, updated version later.


#35

I am currently backing three different crowdfunding projects, all of which should have been delivered 9 months ago or more. This is part of the game when you invest in many of these projects (small companies, limited resources).

I can understand the frustration about the Librem 5 not being shipped to the masses already (I feel it too) and yes, recent communication about shipping leaves a lot to be desired.

However, I think there is too much focus in the forum about how bad everything is, or speculations about if there is ever going to be any phones shipped etc. etc.

Unfortunately long and ranting criticism isn’t very constructive. I am happy that there are actually companies and communities that try to offer an alternative to Apple & Google in this space. We all want this, so let’s practice a bit more patience. Even if Purism doesn’t succeed in the end, I still think they have done a valuable job in trying that should benefit future projects.

And remember, the excitement of longing for something often beats the excitement of actually having it…


#36

Quite possibly. Look at it this way: what do you lose if it’s delayed further? There might not be an improvement in your life, but there won’t exactly be a regression.

My view is that of detached apathy. I’ve set it in motion (my preorder), now I can just forget about it until one day I get an e-mail asking about my modem choice and then some other day in the future I sign for a delivery. Then again, I’ve never cared for smartphones (or dumbphones, for that matter).


#37

Ahh, nonono, guys, you didn’t get me. I’m not disappointed that there is a delay. That’s fine.

I’m disappointed by the lack of honesty (yet again) and more importantly by the lack of concepts presented to solve serious, elemental problems.

It’s good that they seek for a good cooling solution (although it is way too late in the developement timeline to do so). But it’s much, much more important that they search for ways to reduce the generation of heat. Again: I’m not sure all people here realise that the heat consists of 100% energy drawn from the battery.


#38

You are assuming that Purism knew on Sept 24 that it wouldn’t be shipping to any customers. A month ago, they might have still been thinking that they could fix a few last minute bugs and still ship by October 23. Then, they realized there were so many problems that they couldn’t ship. I think that is more likely what happened, from what we know, than the interpretation that you are giving.

If Purism planned to deceive its customers, the CEO wouldn’t give an interview where he says:

  • This phone that I’m showing you is the first one produced. No company wants to publicly admits that it hasn’t yet done any internal test batches and its customers are basically getting rough prototypes that haven’t yet been tested, but that is what Todd told us.
  • The phone sucks a lot of energy and needs to be charged twice per day.
  • It runs a very basic 2D game at 10 frames per second.
  • It has overheating problems, and will need to add heat pipes in the next batch.
  • The camera, Bluetooth and video out don’t yet work.
  • Admit your lack of technical knowledge, by saying that you don’t know whether it supports Vulkan or not.

If you plan on not releasing the phone and you plan to deceive the public, then you don’t say any of these things publicly and you don’t do your first interview with a guy like Gardiner Bryant who you know is going to ask you difficult questions about Linux gaming where your device frankly sucks.

If you have ever been involved in a product release in a small company, then you would know that all sort of unexpected things happen and you have to scramble to make all sorts of last minute decisions. Purism’s decision to not ship the Aspen batch to customers strikes me as that sort of decision.

I chalk a lot of this up to Purism being a small company which has never done this before and it has a CEO and a whole crew of new employees who don’t have a lot of experience, so they thought that they could release without having first done an internal test batch and they thought that they could trust the spec sheet for the processor and wouldn’t need to add extra cooling.

A more experienced company wouldn’t make these sorts of mistakes, but more experienced companies don’t attempt impossible tasks like:

  • trying to create a third mobile operating system that has to compete with the Google and Apple duopoly,
  • making a device that runs on 100% free software, which makes everything more expensive and creates all sorts of engineering challenges,
  • creating a web services platform that respects users rights, and has to try to equal the services provided by Gmail and Twitter,
  • trying to lead a movement for user digital rights, by educating the public about the problems in the tech industry and testifying about the need for the government to regulate the tech industry to protect users’ rights.
  • trying to design a phone that won’t have planned obsolescence, which reduces its sales and harms its bottom line.

In order to do what Purism is trying to do, you need dreamers who aren’t so jaded that they are willing to try and do what conventional wisdom says is impossible, so we have to expect some hiccups along the way.


#39

I think the main power/heat issue is the CPU. It is not a mobile CPU. It is a CPU that works with Purism’s stringent freedom requirements, and that fits in a mobile form factor, but it’s power draw is not optimized with mobile in mind.

This was known from the start, and there have been posts about it, but to that degree, there is only so much you can do to mitigate heat when the CPU is what it is. If the CPU is going to suck battery and generate heat no matter what, then it is better to focus on cooling to reduce thermal throttling.

As @amosbatto said, and as I think you know, this is a monumental project, and obviously the first release (even Evergreen batch) won’t be perfect, but you have to start somewhere.

The Librem 5 v2 is planned to have an improved CPU (14 nm vs 28 nm, which should help with power draw) - they debated including it in the v1, but it is a relatively recent release, and there wasn’t enough test data/documentation to safely choose it by the time the decision had to be made.


#40

Additionally I think they work on reducing the power consumption / heat generation. Maybe even thats what Todd meant when he wrote about the issue of “thermal throttling”. An example is the work on this issue. In general a way to reduce the power consumption is to optimize the software stack (like making posh more efficient).