Librem 5 batch Fir/v2 design progress?

Hardware Librem 5
1

When the Librem 5 shipping was announced it was stated that Batch Fir would start shipping Q4 2020. Now I’m not interested in the debate about shipping delays due to coronavirus and the ensuing parts shortages. I’m much more interested in what design progress has been made as these delays should not have impacted design just as they should not have impacted software development.

Being as the Librem 5 is currently being built as soon as parts are available, I see no reason why the design process shouldn’t move forward and why once complete newer shipments shouldn’t be that fir/v2 configuration.

If the design was supposed to be done in time for Q4 2020 shipping and the design process was mostly complete for the L5 Evergreen by that point I’m not sure I can accept ignorance as a valid reason for the design estimate being this far off.

I don’t expect Purism to be transparent about this, as there is a history of lack of transparency on topics like this, but I can hope someone shares some insight as to whether or not this design work is being done and if there is an estimate on when the design should be nearing completion.

6 Likes
2

I think right now the problem is the same as back in 2017 - should the design be based on existing (improved) iteration of the imx8 (imx6 then) or rather imx9 (imx8 then).

1 Like
3

I doubt that’s how backorders with Purism’s suppliers work. I don’t know any details of Purism’s supply chain, but my guess is that, for example, if Purism has 5,000 CPUs/SOCs on backorder with suppliers, that Purism couldn’t just convert those orders to a newer version of CPU without any changes to wait times. More likely, they’d have to go to the back of the line for the newer chips, which would undermine the priority of getting devices to people who have waited for them a long time.

2 Likes
4

Even accepting the logic that the existing orders couldn’t be negotiated with existing suppliers (whom generally are pretty receptive to alternate part numbers when they don’t have stock/manufacturing capacity for the original part numbers) that doesn’t impact the ability to perform the design work.

5

Except that decision should have been made years ago at this point. Even if supply shortages impact the ability to manufacture at full capacity it doesn’t prevent designs moving forward and manufacturing with whatever capacity is available.

6

I agree that working on the next design would be possible, regardless of the delays fulfilling orders for the current hardware version. However, if Purism gives high priority to fulfilling L5 backorders, I could see how releasing design details for the next version might be premature. For example, a reasonable policy could be that Purism doesn’t take any Fir/v2.0 orders until after all Evergreen backorders have been fulfilled. Imagine, for example, Purism fulfills all backorders within the next 12-18 months. In that situation, newer, better hardware (CPU, battery, touchscreen/glass, etc) may be available by the time they start taking orders for the Fir/v2.0 devices. That would possibly render obsolete any design details that Purism publishes now.

Another factor is feedback from customers. Tripling (or more) the number of people using an L5 Evergreen will allow much more feedback on the limitations of Evergreen’s hardware design, leading, eventually, to a superior redesign. Implementing a new hardware design will be very expensive, so delaying the redesign and getting that additional feedback could be very valuable. I do think this could go too far, however, because redesigning sooner would potentially decrease the number of people who receive faulty devices, so it’s a balancing act.

3 Likes
7

If Purism hadn’t offered for people whose ordered already to choose batch fir in their shipping email what your saying would have more weight, by presenting that option to begun with there is an implication that it is an option that doesn’t hurt the business for people to choose it.

I’m also a little dismayed that current purchases are over a year lead time and are implied to be Evergreen only as that shipping email went out to previous pre-orders but Fir/v2 doesn’t appear to be an option for these new orders implying the design that was supposed to be completed for shipping in 2020 isn’t even expected to be complete for shipping until at earliest 2023. The available information seems to imply that the estimated timeline for the design, which should have been reasonably easy to accurately estimate after having just completed the previous design, was off by several years and isn’t expected to be done anytime soon.

Sure, it is also possible the design is in progress and mostly done, but there doesn’t appear to be evidence available in favor of that possibility.

2 Likes
8

Not quite, the supply delays are affecting overall business model. The expectation was to deliver phones in 2019, receive feedback, fix urgent things and start new design considering money flow from steady sales (and customer feedback, constructive one, not ranting). As of now there’s no steady sales as many people gave up on this project and don’t consider it interesting anymore, partially due to price hike but mainly due to inability to understand “when am i going to get it”. So right now there’s no money, limited feedback, uncertainty with the end price of the product (hence inability to calculate business model of the Fir Phone project).

5 Likes
9

Considering that NXP only posted the data sheet for the i.MX 8M Plus on Sept 8, 2021, I think it is very unlikely that Purism has started working on Fir. (You need to register with a company email to download the i.MX 8M Plus documentation.) The i.MX 8M Plus currently has a 52 week lead time and none of the chip suppliers have it in stock, so even if Purism orders now, it will be a year before the company can get the chip.

I have updated the community FAQ with updated info on the Plus: https://source.puri.sm/Librem5/community-wiki/-/wikis/Frequently-Asked-Questions#71-what-will-be-the-changes-in-the-fir-batch-and-should-i-wait-for-it

As I point out in the community FAQ, the Plus is not going to be a drop-in replacement for the Quad, because it doesn’t support DisplayPort, so Purism is going to have to include an additional converter chip to go from HDMI, LVDS or MIPI DSI to DisplayPort in order to support video out.

The second thing to keep in mind is that the design of the Librem 5 is outsourced to an external company and doing board prototyping is expensive, so it is a significant expense that I doubt Purism is doing right now.

PS: I have long thought that Purism should make Fir with the existing Quad and simply increase the RAM to 4GB (the max possible) and the eMMC to 128 or 256 GB, because there are too many disadvantages to the Plus in my opinion. Of course, what I really want is for Purism to make Fir based on the RK3588 (if a free graphics driver is available), so it has better specs than the PinePhone 2, which will use the RK3566.

4 Likes
10

If you have an outstanding order for Fir then you should ask Purism rather than listen to customer speculation.

With finite developer resources, I would much rather that Purism makes Evergreen a better phone, via software changes, on the existing hardware, rather than make a second phone model, which would inevitably have some differences and in any case would tie up resources in various parts of the company.

Surely Purism’s credibility is on the line to deliver Evergreen rather than divert resources into Fir?

2 Likes
11

If the shipping for V2 was supposed to start in 2020 then this CPU wouldn’t be relevant as it couldn’t have been planned for a design that would be shipping in 2020.

Surely Purisms credibility is on the line to deliver fir over a year after it was supposed to be shipping :upside_down_face:

Also being as the offer of fir as an option for existing orders was made, it seems more reasonable to me that unfulfilled evergreen orders would be converted to fir and would not be negatively impacted by continued design progress.

I’m not trying to bash Purism, I’m asking about something that appears to me to have been overlooked/forgotten about.

Sure, I could get an official response that way, or I could have a conversation. Both are valid approaches.

Except by Purisms timeliness that should have been largely complete long before any shipping delays or any of these other delays and in turn I’m not convinced now has much impact on that unless the implication is that the design was mostly complete then abandoned because of issues with shipping V1 which doesn’t make sense to me.

1 Like
12

Or, if having significantly underestimated A-E, Fir is now on the distant horizon and all Fir orders should be converted back to Evergreen?

Some of the supply issues will be the same, so if Evergreen is “lead time 52 weeks”, you would have to think that Fir would be no less than 52 weeks.

I would take “52 weeks” as a placeholder estimate anyway.

1 Like
13

In my view Evergreen would be the completion of the commitment that began with the crowdfunding campaign - while Fir is a future option that has been informally offered. (I don’t think you can actually order Fir. You can only order Evergreen and then contact support and ask for Fir, allegedly.) Credibility may well be on the line for either but in my view Evergreen is a commitment to which more credibility is attached.

At this stage I don’t think there’s any reason for a customer to convert a Fir order to an Evergreen order. If you have a Fir order, leave it as it is - because there is no real certainty about when either would be delivered.

2 Likes
14

Look, I get why you are encouraging Purism to work on Fir, since the company did make a commitment to customers when it started accepting orders for Fir. For most people who aren’t using the L5 as a convergence device or doing much graphics processing, Fir with the Plus probably will be a better phone than Evergreen with the Quad.

However, I suspect that Purism made its projection based on a whole bunch of conditions/assumptions that are no longer valid, which is why I am speculating that Purism isn’t yet working on Fir.

The Plus was supposed to start volume production in Q3 2020, which is why Purism anticipated that Fir would be able to ship in Q4 2020. However, I suspect that Purism thought that NXP would release a chip that would mostly be a 14nm die shrink of the existing 28nm Quad, which is why Purism thought that it would only need 3 months after the release of the chip to start shipping phones with it. It turned out that the Plus is an entirely different chip from the Quad, which is going to require a significant redesign of the board, and it introduces all sorts of new components (ISP, DSP, NPU, 3D audio, secure key storage, hardware video encoder, etc.) which may not have free/open source drivers or may need to spend years getting them supported in mainline Linux. (Since I can’t download NXP’s software for the Plus, I can’t investigate the driver situation, so I would like to hear from someone who has more info.).

The other thing is that I suspect that Purism thought that NXP would quickly support the i.MX 8M processors in mainline Linux. It turned out that NXP added support for many of the features in the i.MX 8M Quad very slowly to mainline Linux and Purism had to do a lot of its own kernel work. The second thing is that we know that Purism didn’t set its Q4 2020 projection for Fir based on a 52 week lead time to get the Plus processors and Purism may not have known how bad the lead time would be until NXP started shipping the Plus.

As @ruff pointed out above, Purism probably expected to have shipped all the Purism preorders by the end of 2019 and have a year of taking preorders for Fir to finance the development costs of Fir. Purism is stuck in a horrible limbo where it can’t ship many phones and is getting very few orders, so it isn’t getting the kind of funding that it expected. Maybe the two investment rounds ($2.5M + $7.3M) will offset the lack of funding from new orders, but I don’t think that is certain. MrChromebox on Reddit has said that Purism is working on the design of the Librem 16 laptop, which makes a lot more sense than investing in Fir.

The other consideration is that Purism probably made the announcement for Fir, thinking that it would have the high-end Linux phone market sowed up, but PinePhone’s announcement of the PinePhone Pro and PinePhone 2 have taken away that market, and I suspect that Purism is going to have a hard time getting enough preorders for Fir to justify its development costs.

OK, sorry to be a Debbie Downer. Clearly I’m speculating here, so don’t take anything I’m saying in this post as fact, but I think it unlikely that Purism has started working on Fir at this point.

Like @irvinewade, I think the company should be focusing its efforts on Evergreen, which is why I advocate that Fir be an upgraded Evergreen with more RAM and a larger eMMC, rather than go for a new design with a different processor. That would allow Purism to deliver Fir within a reasonable time frame (assuming that Purism orders the parts early to account for the global chip shortage).

5 Likes
15

I see the Linux phone market has having 2 subgroups:

Group consumable: The phone can use a wider and cheaper selection of hardware without regard for the distribution problems that come with binary blobs. This group is great for people who just want Linux and do not mind having their Linux distribution selection limited, maintained by a few, and also the support window limited in duration.

Group freedom: The phone is limited to a smaller selection of hardware where the binary blobs stay on the hardware and all hardware support is submitted upstream, so nearly any Linux distribution can support the hardware, and it is each distribution’s own support policy that determines how long the device is supported.

Purism has also added logical and physical isolation of the insecure radio CPUs and an option for reduced supply chain risk to further add value.

Unfortunately, Purism has had to put a lot of labor into supporting the existing hardware. I do not think that it makes sense to upgrade to something that requires nearly as much effort, unless there is something significant to gain.

Personally, I will not be buying any new devices that do not currently support the AV1 codec and, for portable devices, WiFi 6 with WPA3 and OFDMA on the 2.4 Ghz band, unless they add some special value (like not being iOS/Android). Although battery powered devices should look at supporting Wi-Fi 6 Release 2. The CPU RK3588 looks great with AV1, however, for Purism, open drivers are non-negotiable for their market segment, and I am not sure about the RK chips from a supply chain risk point of view, compared to NXP.

I am glad that the consumable market exists to provide a Linux phone with a lower cost, but I think that given Purism’s mission, its upgrade path will be a little different.

4 Likes
16

Could you link the thread mentioning the Librem 16 please? Very curious about that.

17

Or better still … start a new topic.

2 Likes
18

https://teddit.net/r/Purism/comments/s35dzw/retraction_the_chip_for_the_purism_14_is_actually/

1 Like
19

I would say that there are 5 subgroups buying Linux phones:

  • Tech enthusiasts: they want up-to-date and cool hardware and don’t mind paying extra, but they want that hardware to work as advertized and are willing to use an Android kernel and drivers through libhybris and accept proprietary blobs. They buy the F(x)tec Pro1-X and Planet Computers Cosmo Communicator and Astro Slide.
  • Budget utility: they just want a basic Linux phone that works, and they dont mind using an Android kernel and drivers and some proprietary blobs. Right now their options are the Volla Phone or Xperia with Sailfish OS from jolla-devices.com or installing on their own.
  • community development: they want to participate in community development of a pure Linux phone, so are willing to experiment and tolerate things not always working, but they don’t want a big markup over the cost of the hardware. They want open source, but are willing to tolerate some blobs while the community works on freeing them. They buy the PinePhone and PinePhone Pro.
  • free software diehards: they want 100% free software and are willing to accept any inconvenience including old hardware and limited functionality. They buy the L5, because Purism seeks the FSF’s endorsement. The ones on a budget who understand the tech details may buy the PinePhone, but not the PinePhone Pro since it requires blobs to boot.
  • security/privacy diehards: they want free/open source because they are convinced of the security and privacy benefits. They are generally willing to pay extra and don’t demand the most powerful hardware, but they generally want the tech to work for secure communications. Some have ordered the L5, but many are on the fence waiting on its software development before they buy the L5. Others have bought the Volla Phone, since they need something more functional today and Ubuntu Touch is the most secure of the existing Linux distros with secure boot and app sandboxing, but it has maintenance problems. Most have decided to install an AOSP derivative on their Android phones while they wait for mobile Linux development.
5 Likes