Page 1 of 2

A step forward: LTE

Posted: Fri Jan 25, 2019 1:15 am
by Jackr
Some of the next models of rpi could use SOC with integrated cellular modem, to access LTE connection and phone features. Socs like snapdragon or mediatek. That would be a big step forward for rpi, considering also increasing competitors.
If I was ingenuous, I'd say that it's very strange that no arm boards with integrated modem and Linux support have been made(correct me if im wrong). The Snapdragon 410E is supported by debian but... surprise ! has no modem. :twisted:
I hope that rpi will take this direction, If g00gIe will allow that...

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 1:21 am
by topguy
What do Google have to do with it ??

Not that I believe it has any chance of happening. ( It will only benefit a very few people, would be expensive and its very easy to add 3G/4G otherwise. )

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 2:33 am
by W. H. Heydt
It is difficult to integrate digital logic (such as a GPU or CPU) and analog (radio) on the same chip. In this case "difficult" can be equated to "expensive", in both the effort to do it and the final cost of the chips.

The next question is...how many people would this benefit? keep in mind that the expense would be added to *every* such Pi made (they're selling on the order of 5 million Pi3 boards per year) and probably only benefit a small fraction of those that buy Pis.

Bear in mind that to use a cellular connection, you have to be paying for cellular service, a much more expensive proposition, in general, than paying for internet service. Ever notice that devices that can do both will use a WiFi connection, where available, rather than the cell signal?

So...while an interesting idea on a theoretical level, not a practical--or even necessarily desirable--one in practice.

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 2:53 am
by Andyroo
That’s what HATs are for...

Build a flexible base
Make it expandable
Allow the market to make add-ons for specific needs

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 4:56 am
by Jackr
@topguy oh ur sweet... raspberrypi.org/about/supporters (if this topic gets blocked you know why)

@topguy and @Andyroo Not so easy nor cheap, and the efficency ? And size ?

@W. H. Heydt
The next question is...how many people would this benefit? keep in mind that the expense would be added to *every* such Pi made (they're selling on the order of 5 million Pi3 boards per year) and probably only benefit a small fraction of those that buy Pis.Bear in mind that to use a cellular connection, you have to be paying for cellular service, a much more expensive proposition, in general, than paying for internet service. Ever notice that devices that can do both will use a WiFi connection, where available, rather than the cell signal?So...while an interesting idea on a theoretical level, not a practical--or even necessarily desirable--one in practice
Yeah yeah, but everybody have a smartphone in theyr pants!

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 6:04 am
by W. H. Heydt
Jackr wrote:
Fri Jan 25, 2019 4:56 am
@topguy oh ur sweet... raspberrypi.org/about/supporters (if this topic gets blocked you know why)

@topguy and @Andyroo Not so easy nor cheap, and the efficency ? And size ?
As every engineer knows...efficiency depends on what you want to effish. A big factor in the design of the Pi is *cost* efficiency.
@W. H. Heydt
The next question is...how many people would this benefit? keep in mind that the expense would be added to *every* such Pi made (they're selling on the order of 5 million Pi3 boards per year) and probably only benefit a small fraction of those that buy Pis.Bear in mind that to use a cellular connection, you have to be paying for cellular service, a much more expensive proposition, in general, than paying for internet service. Ever notice that devices that can do both will use a WiFi connection, where available, rather than the cell signal?So...while an interesting idea on a theoretical level, not a practical--or even necessarily desirable--one in practice
Yeah yeah, but everybody have a smartphone in theyr pants!
Who is this "everybody" of which you speak?

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 7:13 am
by fruitoftheloom
Jackr wrote:
Fri Jan 25, 2019 1:15 am
Some of the next models of rpi could use SOC with integrated cellular modem, to access LTE connection and phone features. Socs like snapdragon or mediatek. That would be a big step forward for rpi, considering also increasing competitors.
If I was ingenuous, I'd say that it's very strange that no arm boards with integrated modem and Linux support have been made(correct me if im wrong). The Snapdragon 410E is supported by debian but... surprise ! has no modem. :twisted:
I hope that rpi will take this direction, If g00gIe will allow that...

There is a roadmap of future products, though not publicly available, therefore it is very likely that you request would not be considered, though it may of been already, for several years in the future.

Therefore it would be a backward step because technology would of moved on....


I do not understand your reference to Alphabet ( Google ) please enlighten as like a good conspiracy theory, not.

:roll:

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 7:17 am
by fruitoftheloom
Jackr wrote:
Fri Jan 25, 2019 1:15 am
Some of the next models of rpi could use SOC with integrated cellular modem, to access LTE connection and phone features. Socs like snapdragon or mediatek. That would be a big step forward for rpi, considering also increasing competitors.
If I was ingenuous, I'd say that it's very strange that no arm boards with integrated modem and Linux support have been made(correct me if im wrong). The Snapdragon 410E is supported by debian but... surprise ! has no modem. :twisted:
I hope that rpi will take this direction, If g00gIe will allow that...

Regards alternative SBC there is a few that offer 4G LTE and run Android OS and last time I checked Android is based on a Linux Kernel.

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 10:14 am
by jamesh
Jackr wrote:
Fri Jan 25, 2019 4:56 am
@topguy oh ur sweet... raspberrypi.org/about/supporters (if this topic gets blocked you know why)

@topguy and @Andyroo Not so easy nor cheap, and the efficency ? And size ?

@W. H. Heydt
The next question is...how many people would this benefit? keep in mind that the expense would be added to *every* such Pi made (they're selling on the order of 5 million Pi3 boards per year) and probably only benefit a small fraction of those that buy Pis.Bear in mind that to use a cellular connection, you have to be paying for cellular service, a much more expensive proposition, in general, than paying for internet service. Ever notice that devices that can do both will use a WiFi connection, where available, rather than the cell signal?So...while an interesting idea on a theoretical level, not a practical--or even necessarily desirable--one in practice
Yeah yeah, but everybody have a smartphone in theyr pants!
If everyone has a smartphone in their pants, then no-one needs LTE on their Pi, just tether to the smartphone.

This is a tiny market, and would require a major hike in Pi price to implement, to the detriment of millions of users who wouldn't need it.

Not going to happen, much better done using a USB dongle for those that need it.

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 11:47 am
by topguy
@topguy oh ur sweet... raspberrypi.org/about/supporters (if this topic gets blocked you know why)
I've been here long enough to:
- Not really seen any direct Google influence on the Foundations choices of hardware. I'm curious to know why you would think that.
- Know that when these "New Pi must have this!!!" threads get blocked ( this thread will be too), it is because they quickly degenerate into people calling each other names for not trusting the Foundation to do whats is right for the Pi etc.. etc..

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 12:19 pm
by 6by9
topguy wrote:
Fri Jan 25, 2019 11:47 am
@topguy oh ur sweet... raspberrypi.org/about/supporters (if this topic gets blocked you know why)
I've been here long enough to:
- Not really seen any direct Google influence on the Foundations choices of hardware. I'm curious to know why you would think that.
- Know that when these "New Pi must have this!!!" threads get blocked ( this thread will be too), it is because they quickly degenerate into people calling each other names for not trusting the Foundation to do whats is right for the Pi etc.. etc..
Google support the Raspberry Pi Foundation - the educational charity side.

All the Pi development is done by Raspberry Pi (Trading) Ltd. Whilst wholy owned by Raspberry Pi Foundation, there is very minimal interference from RPF to RPT, and therefore Google as a supporter have even less influence.

But still, there is no need to integrate LTE into the Pi itself. Those who need LTE connectivity can achieve the same through USB or possibly a HAT, and therefore there is no need to increase the cost of the base Pi. The fact that Wifi isn't integrated on many of the Pi models should be an example that there isn't the need for everything to be built in.

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 12:52 pm
by Will5455
@W. H. Heydt
The next question is...how many people would this benefit? keep in mind that the expense would be added to *every* such Pi made (they're selling on the order of 5 million Pi3 boards per year) and probably only benefit a small fraction of those that buy Pis.Bear in mind that to use a cellular connection, you have to be paying for cellular service, a much more expensive proposition, in general, than paying for internet service. Ever notice that devices that can do both will use a WiFi connection, where available, rather than the cell signal?So...while an interesting idea on a theoretical level, not a practical--or even necessarily desirable--one in practice
Yeah yeah, but everybody have a smartphone in theyr pants!
I don't!

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 12:56 pm
by jamesh
Jackr wrote:
Fri Jan 25, 2019 4:56 am
@topguy oh ur sweet... raspberrypi.org/about/supporters (if this topic gets blocked you know why)
I'd be interested to know what you are thinking here. Are you saying that because of the content of this thread, that Google might want it locked for some reason? And that they would actually think I would lock it for them?

The reason this thread will get blocked is because feature request threads are repetitive, end up as slanging matches, and the people who start them are on the whole unable to see the reasons why their pet idea might not be "The next big thing".

I've explained above why built in LTE is a non-starter. Do you still think it would be a good idea if it added $5 to the price of every Pi with it on, even if the vast majority of people never actually used it?

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 1:12 pm
by chwe
jamesh wrote:
Fri Jan 25, 2019 12:56 pm
I've explained above why built in LTE is a non-starter. Do you still think it would be a good idea if it added $5 to the price of every Pi with it on, even if the vast majority of people never actually used it?
if you find a 4g capable SoC with decent mainline Linux support (we're talking here about linux not some crappy android bsp kernel) which only adds 5$ to a pi.. definitively yes! :lol:

Problem is you wont find such an SoC.. Cellular SoCs are made for android.. they stick to their bsp kernel for forever.. Drivers for the cellular are mostly blobs that's why you never see an major kernelupdate on your cell phone.. Have fun to sort out such a bsp kernel mess to bring up some decent linux support.. ;)
jamesh wrote:
Fri Jan 25, 2019 10:14 am
If everyone has a smartphone in their pants, then no-one needs LTE on their Pi, just tether to the smartphone.
well, if that's the only use-case for 4g cellular you folks here can imagine.. then it's really useless to bring it to an SBC.. :mrgreen:

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 1:15 pm
by fruitoftheloom
chwe wrote:
Fri Jan 25, 2019 1:12 pm
jamesh wrote:
Fri Jan 25, 2019 12:56 pm
I've explained above why built in LTE is a non-starter. Do you still think it would be a good idea if it added $5 to the price of every Pi with it on, even if the vast majority of people never actually used it?
if you find a 4g capable SoC with decent mainline Linux support (we're talking here about linux not some crappy android bsp kernel) which only adds 5$ to a pi.. definitively yes! :lol:

Problem is you wont find such an SoC.. Cellular SoCs are made for android.. they stick to their bsp kernel for forever.. Drivers for the cellular are mostly blobs that's why you never see an major kernelupdate on your cell phone.. Have fun to sort out such a bsp kernel mess to bring up some decent linux support.. ;)
jamesh wrote:
Fri Jan 25, 2019 10:14 am
If everyone has a smartphone in their pants, then no-one needs LTE on their Pi, just tether to the smartphone.
well, if that's the only use-case for 4g cellular you folks here can imagine.. then it's really useless to bring it to an SBC.. :mrgreen:

http://www.orangepi.org/Orange%20Pi%204G-IOT/

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 1:45 pm
by jamesh
chwe wrote:
Fri Jan 25, 2019 1:12 pm
jamesh wrote:
Fri Jan 25, 2019 12:56 pm
I've explained above why built in LTE is a non-starter. Do you still think it would be a good idea if it added $5 to the price of every Pi with it on, even if the vast majority of people never actually used it?
if you find a 4g capable SoC with decent mainline Linux support (we're talking here about linux not some crappy android bsp kernel) which only adds 5$ to a pi.. definitively yes! :lol:

Problem is you wont find such an SoC.. Cellular SoCs are made for android.. they stick to their bsp kernel for forever.. Drivers for the cellular are mostly blobs that's why you never see an major kernelupdate on your cell phone.. Have fun to sort out such a bsp kernel mess to bring up some decent linux support.. ;)
jamesh wrote:
Fri Jan 25, 2019 10:14 am
If everyone has a smartphone in their pants, then no-one needs LTE on their Pi, just tether to the smartphone.
well, if that's the only use-case for 4g cellular you folks here can imagine.. then it's really useless to bring it to an SBC.. :mrgreen:
Of course there are loads of use cases. The point was that the OP said everyone has a smartphone in their pants. This was a direct resposne to that. Every single use case would be possible simply by adding a 4G USB dongle.

Remember, that in order to remain backward compatible, we would need to find a SoC that does everything the current ones does, but with 4G added. There is no such chip, and there will never be one. So the two options are: Add a separate chip (interfaced...how?) or use a 4G dongle.

Adding a seperate chip is expensive, and that high cost would be passed on to every user. Which would be an UNBELIEVABLY STUPID THING TO DO because our sales would crash and it could even put us out of business.

I really don't understand why people cannot see this. It's basic business sense, but I guess that why so many kickstarters fail - the people behind then have no business sense whatsoever, and simple do not think things through. You don't put expensive features in a product unless you make it back in sales. This won't make it back in sales revenue, and would certainly result in a decrease in sales due to the higher cost of the product.

Re: A step forward: LTE

Posted: Fri Jan 25, 2019 1:51 pm
by chwe
fruitoftheloom wrote:
Fri Jan 25, 2019 1:15 pm
chwe wrote:
Fri Jan 25, 2019 1:12 pm

if you find a 4g capable SoC with decent mainline Linux support (we're talking here about linux not some crappy android bsp kernel) which only adds 5$ to a pi.. definitively yes! :lol:

http://www.orangepi.org/Orange%20Pi%204G-IOT/
..
I spend quite a few hours with the sources for the 4G-IoT, kernel and bootloader... We're far away from decent kernelsupport.. We're talking here about an outdated 3.18 kernel (without git history) glued with an LK bootloader.. bloated in an messy android buildscript.. :lol:

Xunlong makes some great SBCs, unfortunately their cellular ones aren't.

And for the kickstarter stuff, the SBCs I followed on kickstarte did all good (e.g. librecomputer)..

Re: A step forward: LTE

Posted: Sat Jan 26, 2019 2:17 am
by Jackr
Thanks for the attencion guys!
@jamesh I'd spend 35$ more for that. And yes It would require quite an development effort... And it would be a special model.

@chwe
Problem is you wont find such an SoC.. Cellular SoCs are made for android.. they stick to their bsp kernel for forever.. Drivers for the cellular are mostly blobs that's why you never see an major kernelupdate on your cell phone.. Have fun to sort out such a bsp kernel mess to bring up some decent linux support.. ;)
Thats exactly what I was talking about ! There are a lot of cellular SOCs but (almost) none can be installed a linux! Also, no dev boards. Why ? Why would a chip maker develop the support for android and not linux ??? Qualcom website is extreamly closed. Find support for a product is a tresure hunt!
Ok the ARM architecture is aso to blame, in fact a dev can't simply support the CPU or the SOC, you must support a board! cos the boot process is funky (tell me if im wrong, and maybe explain...). I just thought that the raspberrypi team coul do the magic... :roll:
@fruitoftheloom
Thanks fruit, I knew that. Im sure it uses a separate chip like simcom, that is almost the same as to use a hat.
However orangepi is one of the competitors I mentioned :(

Re: A step forward: LTE

Posted: Sat Jan 26, 2019 3:49 am
by W. H. Heydt
Jackr wrote:
Sat Jan 26, 2019 2:17 am
Thanks for the attencion guys!
@jamesh I'd spend 35$ more for that. And yes It would require quite an development effort... And it would be a special model.
For special hardware for short runs, expect to add at least one zero to the end of that price.

Re: A step forward: LTE

Posted: Sat Jan 26, 2019 4:13 am
by rpdom
W. H. Heydt wrote:
Sat Jan 26, 2019 3:49 am
Jackr wrote:
Sat Jan 26, 2019 2:17 am
Thanks for the attencion guys!
@jamesh I'd spend 35$ more for that. And yes It would require quite an development effort... And it would be a special model.
For special hardware for short runs, expect to add at least one zero to the end of that price.
Depending on the size of the run, I'd say 3 to 5 zeroes.

Re: A step forward: LTE

Posted: Sat Jan 26, 2019 6:29 am
by Heater
Jackr,
Why would a chip maker develop the support for android and not linux ???
Presumably because most such SoCs are designed for mobile phones, tablets and such where Android rules the roost. They are basically designed specifically to run Android.

Given the strong ties between the Raspberry Pi Foundation and Broadcom I don't see them adopting a different vendors SoC any time soon.

Re: A step forward: LTE

Posted: Sat Jan 26, 2019 4:46 pm
by chwe
Jackr wrote:
Sat Jan 26, 2019 2:17 am
Thats exactly what I was talking about ! There are a lot of cellular SOCs but (almost) none can be installed a linux! Also, no dev boards. Why ? Why would a chip maker develop the support for android and not linux ??? Qualcom website is extreamly closed. Find support for a product is a tresure hunt!
Ok the ARM architecture is aso to blame, in fact a dev can't simply support the CPU or the SOC, you must support a board! cos the boot process is funky (tell me if im wrong, and maybe explain...). I just thought that the raspberrypi team coul do the magic... :roll:
well let's elaborate this a little bit. We shouldn't distinguish between 'Linux' and 'Android', probably more a distinguish between 'debian' and 'android'. Both are based on some sort of a linux kernel with some differences. Mostly SoC-makers build their BSP-kernel based on an normal linux LTS kernel with their own drivers to support the SoC. The CPU isn't the issue here.. but other IPs inside the SoC may be more troublesome. The code quality (compared to mainline) is often lower. The code has just to work good enough to run android once achieved there's not much to do. Compared to the million phones with android they sell.. those few debian based boards simply doesn't matter.. why should you prepare a kernel to run flawlessly other sorts of linux than android.. there's simply no market for it.. Drivers for cellulars are often delivered as blobs.. if you look into the 4G-IoT sources.. you'll find a lot of blobs there.. so porting such an SoC to mainline would need stable APIs and/or a dedicated driver to 'talk' cellular blob whereas documentation for such blobs is notoriously non existing..
For the bootpart. that's quite often a mess through different SoCs. E.g. the RPi wouldn't refuse to do anything without the threadX blobs.. Amlogic SoCs have a cortex-M3 for handling a lot of clocks etc. (also blob controlled) RAM init is often also done by blobs (e.g. rockchip).. it's hard to be sure what happens there.. Allwinners can mostly be used without any blobs (except the few bytes in brom).. You can build an u-boot with SPL.. Default in 'android world' to boot up a board is LittleKernel (https://github.com/littlekernel/lk/wiki/Introduction) whereas most ARM SBCs uses u-boot, the Snapdragon 410 mentioned above uses LK as well.. IMO it's a weird mixup of trying to bring up some sort of a 'debian'. LK fires up u-boot, u-boot fires up an kernel.. It's on a lowlevel side for sure not as fun to deal with such a setup. Why should someone ever go down such a rabbit hole just to bring up an SBC with a cellphone SoC. Use a better upstream supported SoC combined with a 4G modem with some sort of a serial interface and you're done.. you don't have to deal with the rabbit hole nor a BSP kernel of questionable code quality...

And for the end-user.. if you can't wait until someone comes up with something like that. Look for an USB 4g stick.. they are often subsidized by ISPs with sim lock.. doesn't matter as long as you don't plan to change your ISP.. Make sure that you find drivers for it and you're done... Some other SBCs have everything needed to bring up 4G. If this is your need go for such a board or bring up a stick.

Re: A step forward: LTE

Posted: Mon Jan 28, 2019 5:38 am
by Jackr
thanks chwe. The only soc I found is a mediatek of the gemini pda where you can install kali, they say. But im not sure what kernel they uses

Re: A step forward: LTE

Posted: Mon Jan 28, 2019 3:48 pm
by chwe
what's the goal. having Kali Linux with an SBC which has a cellular? Find a cellular stick which works and then debootstrap your own Kali.. armhf/arm64 packages are there - it's just a sort of debian so getting kali to run on random SBC isn't much an issue as long as someone got a debian running on it.. I would rather prefer to mess with an cellular usb stick as to mess with a crappy kernel and have the cellular 'on board'.. It's one driver to deal with compared to a bunch of different ones for a whole SoC.. :P
Or start from a Debian and install the needed packages to 'kali-ize' your debian - probably the 'easier' way to begin with.. But if you want to deal with low level stuff.. there are a bunch of cheap tv boxes to have fun with bootloader/kernel on arm.. :P You'll learn a lot.. :P

Re: A step forward: LTE

Posted: Mon Jan 28, 2019 11:51 pm
by Jackr
chwe wrote:
Mon Jan 28, 2019 3:48 pm
what's the goal. having Kali Linux with an SBC...
I don't know if it uses a linux kernel with a couple of unofficial drivers or many unofficial drivers or a android kernel ?!
The case 1 is ok, the case 2 i dont know, the case 3 NO!
Then there is another problem, that soc (MT6797X) is a bit old and power hungry...