Thanks for the responses. But will those 2 use the same configuration scripts that is in this forum documentation? Thanks.
I haven’t tested 8262 (E or A) but it should work. Also, I haven’t tested any of the redcaps, which may not function similarly to 8202g and may not even fit the L5 (physical size matters, antenna connections matter, USB/PCI matters, M.2 keying matters etc.). It would be nice to get more of these, but for other modems, please start a separate thread.
(Also note that all the testing and reporting I’ve done should be considered a PoC, proof of concept - it has worked, it still works, but it’s not “primetime” and may need manual steps as Byzantium or Crimson haven’t included any scripts to handle any other modems than the BM818. I got confirmation last summer that Purism is aware if this modem and that someone may have made some internal ticket but no other info)
I doubt that that is correct.
The kind of initiative shown by the OP is most likely encouraged by Purism. Of course from a Purism Support point of view, if you have a problem with cellular comms, they might ask you to restore the original modem and then reproduce the problem.
The fact that Purism hasn’t officially commented on this modem as an option more likely reflects the fact that they haven’t had the resources to test it out to their satisfaction and accommodate it in software and configuration if required, taking into account that any testing would doubtless require testing on multiple continents with multiple mobile network providers.
I believe that Purism is looking at other modem options and they may be looking at modems other than the one in this topic and/or they may be looking at this modem.
Edit: to add citation for previous para: T-Mobile 4G (!) shutdown beginning in 2026 - #5 by JCS (since @Privacy2 expressed concerns about “speculation”).
I thought “speculation” if frowned upon. Isn’t this where one might ping @JCS ?
One thing that people may consider is whether a 5G modem would require a different antenna for optimal reception. Most 5G cellular modems have 4 antennas … whereas I believe the default cellular modem in the Librem 5 has two antennas.
I definitely believe that they are testing 5g stuff otherwise will be a new design or phone out with 5g. Modem here is at the end of support anyway, but I still want to test it just to have 5g haha. Maybe I can wait a bit longer.
8202g is designed to work with only two antennas and I can confirm that it works in 5G with the ones L5 has. Are they the most optimal for that is another question, but at least the combo seems work adequately. 5G using those special frequencies for max data speeds is not available with this modem (to my understanding they aren’t implemented in many devices, but I could be mistaken on that) and neither could L5 handle those speeds (limited to USB transfer speed - the modem (at least) and antennas should be able to give it more than it can handle, if network and subscription limits allow).
Not to mention … not implemented in many mobile networks or in many countries.
and
I knew that my phone (Pixel 7a; supporting 5G UC t-mobile connection) is frequently on 5G. And given your discussion, I decided to test the speed (I rarely use it). From inside my house I just did a test (using Ookla Speedtest) of the cellular modem. I just got 780Mbps download speed. It didn’t do an upload test for some reason.
I had no idea it was that fast. It seems insane.
Yep. 5G can be very fast. But even that is subject to network congestion and it’s not about the top speed, it’s the average that counts. L5 cant do ludicurous speed but it can be faster than what BM818 gives. Theoretical max (as per USB) would be 480Mbps (so, in real world maybe half that?), if my recall is right.
But to be clear, 5G comes in two frequency ranges. The lower frequency range (FR1) is nominally called “sub-6 GHz”, and the frequency is often not a lot different from the frequencies used by 4G and speeds potentially not a lot different from 4G either. The upper frequency range (FR2) where you will get serious speed most likely do require a new antenna but are not widely supported in either devices or networks.
For example, in my area 5G is provided on frequencies anywhere from 700 MHz to 3500 MHz, and at the low end frequencies of that range are exactly the same as frequencies that provide 4G.
The main difference between 4G and 5G is in the possible channel bandwidth.
Can you confirm the frequency and the channel bandwidth?
However I feel like, with the congestion in my area, today I would never achieve close to 1 Gbit/sec anyway.
Of course it will also depend on whether your plan imposes any kind of specific or de facto limit on the total amount downloaded. If you can use up your month’s allocation in half a day then perhaps it is truly insane. ![]()
I tend to do the big downloads on the household’s fixed internet connection, which has a much more generous monthly allocation.
Yes, for USB 2.0, but the actual implementing chip is much slower than that. So it’s going to need Librem 5 v2 in order to get full value out of either 4G or 5G.
No. I don’t know how manually, so I think it would require an app from the Play Store. And even though such apps are isolated in their own profile/container … I try hard not to use the Google Play Store.
I don’t have a limit on my data … but I pay for every bit of data (none of it is “Free”). That single 10 second test cost me $10 of data. Before I did the test, I thought it would cost about $1. Insane. ![]()
I rest my case. Seriously though I guess if you actually relied on this mobile data service for significant use, you would be on a better plan.
(Due to masses of bad PR for ISPs and mobile network providers, the norm here these days is more that you get “shaped” when you exceed your monthly allocation, rather than paying extortionate rates for “excess data”. So you could drop down from 800 Mbit/sec for 1 minute to … 1.5 Mbit/sec for the rest of the month.
)
Definitely. When I’m not traveling I average around $30/month. When I am traveling domestically (doing e-mail, maps, etc.) it’s more like $50/month. And if I’m traveling internationally, I get a “vacation e-Sim” ( it’s about $1/GB ; I can route all data through that 2nd sim … my phone has a nano-sim and an eSim). When I’m backpacking I average around $20/month. So it’s pretty reasonable for me.
DigiKey is my component go-to and has the SIM8202G-M2 in stock, but it’s kind of pricey at qty 1.
let me see if they still have it. I keep getting denied by the bot on that site.
What do you mean by that? Aren’t you confusing this with the hub’s built-in card reader?
A reference to your post but I might have misinterpreted what you wrote. That is, the speed you quote may have referred to just the built-in card reader or it may have referred to the chip (USB2642) as a whole. I took your comment as referring to the chip as a whole.
But I just pulled the datasheet and I think I did misinterpret what you wrote. From the datasheet:
The flash media interface can support sustained transfer rates exceeding 35 MB/s.
and
The flash media interface can support sustained transfer rates exceeding 35 MB/s if the media and host support those rates.
It’s still true that we will need Librem 5 v2 in order to get full value out of either 4G or 5G.
Maybe someone can give some actual speedtest results with either the BM818 or the modem of this topic, as connected via this chip (and with WiFi killed to be sure that data is going via the cellular modem) showing faster than 35 MB/s (280 Mbit/s). Of course, the limiting factor can be anywhere in the path, not just this one chip - and that includes factors external to the phone such as congestion on the user’s tower and/or congestion on the backhaul and/or the user’s data plan.
The BM818 only claims “150 Mbit/s” (download) so we would probably have to lower the bar for that modem and just ask that anyone can get an actual speedtest result close to that figure with that modem.
“Exceeding 35MB/s” is used there because that’s close to the maximum possible data rate you can get with USB 2.0 BULK transfers. It effectively means “it should be possible to max out your host controller’s performance with this chip”. Remember that your data is not the only thing that goes through the cable, it’s packed into 8b/10b encoded periodic microframes, and since a USB device can never speak unprompted, you use up some of these frames on asking your devices for data, leaving the bandwidth not fully realized.
I thought we resolved this and the bottleneck was the SDCard Reader and it was two facts:
- USB2642 does not support UHS-I and UHS-II bus interfaces as the spec version is not compliant and the bus speeds are above the bus speed of USB2642 (35MB/s)
- When a card reader does not support UHS … it defaults to “standard speed” which, I think, is either 12.5MB/sec (but might possibly be 25MB/sec) … which is what they were getting.