So I’ve been using the Framework Laptop since Batch 2 (12/2021) - 11th gen intel processor and have been maintaining my “FreeBSD on the Framework Laptop” post for that time. It’s been a fun ride and very frustrating. Of course, I love and support the company and the mission, which is why I’m still using the machine despite these issues. I also understand that Framework is a start up, and FreeBSD is also lacking developers to provide the hardware support needed for all hardware. At the same time, Framework has done an amazing job of making a modular and repairable machine, and that everything is a work in progress. We’ve all progressed a lot together over the past few years, and we now even have AMD Ryzen chips ;).
Since the Intel AX210 (I’m using the no vPro version, but probably doesn’t matter), lacks good support on FreeBSD (they have no production ready iwlwifi driver), this not only means that we are limited to b/g speeds, but the driver will crash the system during operation, and when resuming from sleep. I just now found this post by another framework user that got “flawless” support for wifi on FreeBSD using an Atheros AR9462 M.2 (NGFF) 2230 E Key Card. They purchased it from Think Penguin. Given their success, and the modular/repairable nature of the laptop, and also given that it was difficult for me to find a good working (non USB/dongle) wifi card, I took the plunge and just ordered one for a $71 dollars, which is pretty expensive (for most people), especially when you think about all of the other cheaper cards you can get. I understand that there are probably cost benefits that framework can receive by ordering the current Intel AX210s in bulk, I think it may be beneficial to investigate if its possible for their to be more wifi card diversity that can work on not just Windows/Linux. Another benefit to using this chipset is that there are no firmware blobs and the cards are well supported on Linux (and probably Windows). To quote Think Penguin
The Wireless N M.2, or Next Generation Form Factor (NGFF) Card is the best supported M.2 wireless card on the market. With the full set of sources available under a free software license the card is uniquely supported across all distributions. Actively maintained by the community there are no third parties to get in the way of bug fixes, driver/firmware improvements, or continued support. If you have a recent distribution simply install the card and boot. It’ll work right “out of the box”.
This has the benefit of increasing security at the Operating System level.
@Second_Coming It’s definitely not the most modern card but it’s a good foundation that has good compatibility across all the OSes. With compatibility (and relatively modern features) being prioritize over the latest features (but lower compatibility), this could make a good candidate. However, overall, I think the idea is that given that the Framework Laptop is a modular and open system, and continues to become more and more open, this would be a great opportunity to possibly negotiate with some vendors to provide modern cards and chipsets, that do have WPA3/Wifi 6E, etc, and make them open source with no firmware blobs, while supporting Windows/Linux/FreeBSD (the more OSes the better). Even if they don’t provide the drivers themselves, providing open specs and references allows everyone implement those drivers.
All of this increases the value proposition for Framework as a company and of its hardware, and allows it to essentially become a reference point / machine for all OSes.
The one I posted is a good candidate (Atheros AR9462), although that’s the card I ordered so that I can actually get a stable working card on FreeBSD in general, but long term it would be better for Framework to use their market power to negotiate with vendors to provide a modern chip that’s compatible with the Framework laptop that also has open specifications (and no firmware blobs). I primarily also mentioned it in case their are other struggling Framework FreeBSD users. It’s difficult to find a good working wifi card for this OS, unfortunately.
A single laptop-only company can’t drive / steer / negotiate chipset development out of thin air. (Not that I don’t want to see it…but Framework is not HP/Dell/Lenovo scale)
You need to have a chipset that’s already has “WPA3/Wifi 6E, etc, and make them open source with no firmware blobs”.
Do you have a chipset in mind that already fulfills ‘modern’ wifi specifications AND with no firmware blobs?
@Second_Coming That’s all I got. The card is modern enough for me and there are plenty of users out there that would be happy enough with this chip.
As for stimulating chip development, someone “big enough” needs to take the first step. I feel that due to Framework’s vision and already built reputation, there could be a vendor out there willing to work something out.
@fearedbliss@Second_Coming There are none (which I suspect you both knew) and the only thing to do is to take an existing chipset and reverse-engineer it, which is no small matter.
Yeah…probably not. Qualcomm certainly not (makers of the atheros card you are using actually) and Intel probably doesn’t care enough and I bet neither does Realtek. I use the Atheros card myself and while yes it is sufficient for my use, it isn’t modern and shipping an old chipset to appease FOSS purists will hurt Framework.
@fearedbliss You are using an OS that doesn’t even have the market share of Linux which itself doesn’t come close to the market share of Windows, I appreciate you advocating for your chosen OS but be realistic, especially when Framework made no claim towards compatibility.
Making cards like these are a bit of a boutique thing, they are made of chips that aren’t made anymore and were made before m.2 was a thing, you can get the mini pcie version and usb version for <10$ but there is a good reason the m.2 version is that expensive.
Having a modern fully open source wifi chipset would be really nice, ath9k was a nice blip in history but so far has not been repeated. Currently mediatek looks like they are warming up to the idea a bit but I have no idea if we will ever get anything as open as those 9000 series atheros ones again.
Putting a usb version in an expansion card may be a cheaper options for those hardcore enough to run freebsd but not hardcore enough to get a boutique wifi card.
I’m not saying it’s going to be easy, but there has to be a way to be able to use our current market position to find at least one vendor that can make a compatible wifi chipset, even if it means:
Not necessarily having the latest wireless technologies (even n would be good enough for most people).
Is open source with no blobs (although I’m willing to compromise and still accept blobs as long as the chipset itself has properly documented specs, so that OS driver writers can easily implement said driver).
Framework can still offer the AX210 or any other modern wifi chipset, but it could also offer just one additional chipset that is focused more on wider compatibility rather than latest features. I’m also not saying Framework has to offer full compatibility for FreeBSD, but I am saying that there could possibly be some low hanging fruit, or a conscious effort to keep wider compatibility in mind when selecting hardware components / chipsets. In this post I was only requesting for the wifi chipset to be considered, but ultimately the whole machine should be considered. I’ll take what I can get at the moment.
Making an expansion card version wouldn’t be a good option, there are other issues that arise when running components as expansion cards, especially when the OS doesn’t properly or fully support thunderbolt / USB 4. Additionally, this would mean that I would still have an unused M.2 2230 slot inside the machine, and would need to sacrifice an expansion slot just for this, which doesn’t make sense.
That’s pretty much all I have to say on the topic, I’m happy to have opened up this thread to at least bring the conversation forward.
You seem to be underestimating the companies that make those things, they tell much bigger companies to f off. Especially with the fcc and co encouraging them to lock that stuff down even more.
Using an ancient wifi chipset that needs to be scavenged from old parts or new old stock that is found somewhere isn’t what I’d call a low hanging fruit.
Since the usb version of that chipset are usb2 there is no usb4 involved (The solution to make usb3+ backwards compatible with 2 was to just tape a usb2 interface to the new one, they are entirely separate interfaces)
It makes sense considering the alternative, if you want to pay less you use the expansion card one (though even that would probably not be cheap ceap), if you want to use the internal slot you use the more expensive boutique m.2 one.
There is also the option of putting the usb version on an m.2 card, which could potentially be cheaper but idk how much. I don’t think the 71$ for the one you got is that overpriced, sure it’s not high end but it’s niche.
I would personally love to have a modern day version of the ath9k but I kinda doubt that’ll happen. Some small blob mediatec thing is probably the closest thing we’ll get anytime soon. Especially with wifi6e and stuff opening more conditional bands the regulators really don’t want you to be able to mess with the radio and since locking everything down tends to be the lowest effort “solution” to that it’s what is usually done.
As someone who has been in the Linux community for almost 20 years at this point (started using it in 2006 with Fedora Core 5/6, Ubuntu 6.06 Dapper Drake), already went through all of these similar issues back then, was a Gentoo Developer (retired), and more, I’m well aware of the obstacles, market conditions, company shenanigans, and other problems. I’m just choosing not to be pessimistic and negative about certain issues.
Just because there are obstacles, doesn’t mean we should give up and not even try to take a single step forward.
Given that Framework ships the Intel laptops with Intel Wifi and the AMD laptops with AMD Wifi when technically a single model would probably work on all laptops, and would have simpler logistics, there’s probably a good reason for that. Maybe compatibility, maybe rebates, maybe contractual requirements with AMD and/or Intel.
To be honest, the amount of possible BSD users probably doesn’t justify using a different, possibly more expensive and possibly slower Wifi module. Not trying to be rude, but you can always buy your favourite module and sell the default one and the money you lose doing this is probably negligible, compared to the money it would cost framework to add logistics for a whole new part.
@Jonathan_Haas Definitely. That’s what I’ve been doing haha. But I figured if I and others are doing this for different components, maybe there is a conversation to be had if that’s the correct approach or if there is a better way.
For example, Many people in the community and myself were advocating for having matte displays as a framework option. In the meantime, we all purchased matte screen protectors from PhotoDon and other providers, fast forward some time, and now Framework will be providing matte displays as the default option for all framework laptops. The old glossy ones will still be available (I guess until the stock runs out). This is a situation where Framework took customer feedback into consideration, understood what customers were doing in those situations, and figured out a way to make it happen in a way that still makes sense to the company. I’m extremely happy about that.
I’m also really happy that the machine is modular and repairable, which of course regardless of what happens, it will allow me to replace a good amount of parts on the system to components that may be more compatible for my use case without having to push that burden onto the company. However, given that Framework’s main goal is modularity and repairability, and they have been pushing for more open and transparent sources (whether it’s schematics or actual open source code), there is an argument to be made here about what components and chipsets we select.
Yeah I guess I am a bit pessimistic about this though I got a major white pill today with amd going for coreboot support by default by 2026. So maybe one of the radio chipset manufacturers will have a bright moment at some point, might be mediatek, might be atheros coming back from the grave or someone entirely different. Even then I doubt we are ever going to get as much freedom as with the ath9k because of regulations.
Also I love that someone actually made an m.2 module out of the old atheros chips that were made back when m.2 wasn’t even a concept. Small scale manufacturing really enables some neat stuff.
@Adrian_Joachim That’s awesome! I’m happy AMD is going with coreboot. That’s a bit step forward :).
At this point I don’t mind which company steps forward and wants to work with us, anyone willing can join the family.
I didn’t know that those Think Penguin M2 chipsets were built from old interface modules. That’s awesome and explains why FreeBSD (and Linux, etc) has good support for it even though it’s a newer form factor. Now things are making more sense. With that knowledge, the $71 price tag makes a lot more sense now and that’s actually a really good price given the labor required to convert an old module to a newer form factor. Let alone testing, and additional labor.
The history on why the atheros 9000 series has so open drivers is pretty wild, there were a lot of factors that came together at the right time to make that happen.
Can’t find the talk about the history of it but there is another interesting one, we have not got anywhere near as much access to the hardware since then.
I do like the idea of supporting hardware that has open drivers. Given the practical challenges that seem to be discussed here, is it possible for community members to make a listing(wiki) of open hardware that is compatible with the Framework laptop? One of the reasons I’m not on BSD as mentioned in the other thread is the lack of or workaround required to get WiFi working on FreeBSD.
