News from the embedded world


As mentioned, we're planning to put the CPU together with the memory and storage on a pluggable daughterboard.

This saves money in production (since the mainboard will be two-layer and only the small CPU board needs to be more complex), and it adds the possibility to upgrade the Pyra in the future with whatever SoC / memory / storage will become available in the future.
I guess I missed out on hearing about the plugable daughterboard, but that is just awesome. Is there any known limits to this sort of setup? IE could, in theory, 10 years down the line have an update that has a 6 core 5ghz cpu in it (if a small enough size/power sipping one was released)? Or would that be to much for the rest of the system?
 
As mentioned, we're planning to put the CPU together with the memory and storage on a pluggable daughterboard.

This saves money in production (since the mainboard will be two-layer and only the small CPU board needs to be more complex), and it adds the possibility to upgrade the Pyra in the future with whatever SoC / memory / storage will become available in the future.
I guess I missed out on hearing about the plugable daughterboard, but that is just awesome. Is there any known limits to this sort of setup? IE could, in theory, 10 years down the line have an update that has a 6 core 5ghz cpu in it (if a small enough size/power sipping one was released)? Or would that be to much for the rest of the system?
This kind of update has limits anyway. It's good to do incremental changes but over time, other non-replacable parts of the system will change (usb ports, newer video out standards) and you'll have to modify the whole machine again. I'm not sure how much there is to gain with such modularity vs opportunity cost of not releasing a fully/up to date machine every 3-4 years. 
 
Wow, (1) means multiboot without having to use one of the front SD Card Slots. Thats awesome! Now I know where I'll put Debian... oh, wait ;)

Pyra is going to be awesome!
 
As mentioned, we're planning to put the CPU together with the memory and storage on a pluggable daughterboard.
Just Great!
2. UMTS and a hardware switch

However, we found another, pretty cheap but effective solution:

We plan to include a small power measurement chip into the hardware, that is hardwired to a small LED.

Whenever the UMTS module needs power (which it does when it will be activated), the LED will light up.

This hardwiring cannot be changed in software, so you will KNOW when your UMTS module will be active!
Whats the general plan on wireless indicators ? Seperate indicators for WLAN,BT and xG (similiar to the Pandora) ? What about a software controlled LED and a small dipswitch (accessible via the battery compartment) ?
 
  • Like
Reactions: szr
That was very interesting. Thanks for the link.

I found this chart to be most interesting:

arm_performance.jpg


If I'm reading this chart correctly, (and if I'm not, I trust Exophase to correct me) this is saying that at maximum clock, the A7 will still use less power than the A15 at minimum clock. If so, that is really impressive, and really shows how useful the A7s would be in a device like the Pyra.

-God Ginrai
 
While MicroSD Cards / eMMC usually use about 0,3W, the nanoSSD can use up to 0,9W
 
storage is not a CPU or GPU, it's not like someone starts writing and reading more because the storage is faster.

What I'd like to see is actual numbers of data size per joule (MB/J) and idle power consumption.

Peak power consumption is misleading there.

2. UMTS and a hardware switch
 
What about including software power switch that works independently of modem?

Say, a chip that measures power could also cut power to modem.

That way we can remove our concerns about firmware and only direct pyra OS hacks could re-enable modem without user's consent.

But hacked pyra could use another means to collect and transfer user information so such hacks are not modem-specific.

While it's possible to buy the latest BayTrail without any NDAs and in low numbers, Intel suggests to not buy it directly but get it on a module from one of their partners. Getting the SoC directly to work is a bit more complicated than an ARM SoC, since it will also need a BIOS...
It's not big news that intel processors need BIOS to initialize and it may turn out good in the end due to Coreboot advancements. My guess is that Atom SoCs will have better opensource support by the end of this year so Pyra86 may be easier to create by the time Pyra is out then it is now.

I also visited Qualcomm and got another contact for getting the Snapdragon. The 805 will be too power hungry for mobile devices though (it's mostly aimed at tablets), so it won't be usable for us.
Well, current pandora is a 8'' tablet (with small screen) and pyra will be no different. If you convert pandora's slice component layout to planar one you will get area similar to 7''-8'' tablets so pyra should handle such power. If Qualcomm representative meant >=10'' tablets than it's another story.

Like I said earlier, power capping options are much more important for us then maximal power consumption as long as chassis can dissipate it
 
Additionally, we will try to add the MicroSD/SIM-Card Combo-slot below the battery - and include a small chip that will be able to either connect the eMMC or the MicroSD Slot to the SoC.
This is bloody genius, this is. Woot!
 
  • Like
Reactions: szr
As mentioned, we're planning to put the CPU together with the memory and storage on a pluggable daughterboard.

This saves money in production (since the mainboard will be two-layer and only the small CPU board needs to be more complex), and it adds the possibility to upgrade the Pyra in the future with whatever SoC / memory / storage will become available in the future.
I guess I missed out on hearing about the plugable daughterboard, but that is just awesome. Is there any known limits to this sort of setup? IE could, in theory, 10 years down the line have an update that has a 6 core 5ghz cpu in it (if a small enough size/power sipping one was released)? Or would that be to much for the rest of the system?
This kind of update has limits anyway. It's good to do incremental changes but over time, other non-replacable parts of the system will change (usb ports, newer video out standards) and you'll have to modify the whole machine again. I'm not sure how much there is to gain with such modularity vs opportunity cost of not releasing a fully/up to date machine every 3-4 years. 
And yet I am currently using XBMC via a Pi on my old TV with RCA with mouse/keyboard plugged in via USB2. Just because new standards come out doesn't mean people stop using the old ones. So while it would be of use to have updates to the whole piece as well, that doesn't mean that the old system isn't any good. I think of it much like Xbox originals. It only has 32mb ram, a crappy CPU, no HD, and yet people are still using it all the time. It was my main media device till I got my Pi. That is well over 13 years. If it had the ability to upgrade the cpu/ram then I would still be using it.


That is not to say new standards are not nice, but 10 years from now I will have HDMI TV's just sitting around I could use, as well as tons of USB2/3 devices. The only thing that /might/ have a problem, for me, would be wifi. This is solvable though, as the Pi supports USB devices. So I would find updates to the SOC quite usefull.
 
  • Like
Reactions: szr
UMTS Hardware switch:

just to verify that: But we still power on/off the module from the kernel via GPIO, right?

To all rest: very well done...

Thanks a lot.

@gta04: i hope the CPU board will be small. This will give us the the oportunity to not only upgrade the Pyra but also to build other devices on top of the cpu board....
 
Optional means the module is expensive (the actual hardware), so if you don't pay for it you don't get it.


I don't think we really need such a warning light. If you're paranoid about it, don't get the module.
If you could cut the module in half and just give me the GPS part of the die, I wouldn't have to choose. I just don't want my GPS to come at the cost of the Pyra having an additional avenue of attack above and beyond the Pandora.

As a Canadian, wireless data is too expensive for me to use. (Though, if it can be done with an indicator LED, having the option in case the situation improves is nice too.)
 
Is the GPS a "true" GPS or A-GPS? If it's a true GPS just don't put in your SIM card. A-GPS needs to use cell towers so you would need a SIM. Without a SIM the chances of being attacked are extremely unlikely at best. I would dare say virtually impossible, they would have an easier chance of getting to you through your wifi internet connection behind multiple firewalls.

I haven't read the discussion on if an attack through cellular modem is possible or not. I know for a fact it is possible, don't ask me how I know, but the reality of what is possible with an active cellular connection is pretty scary. The PROBABILITY of you being attacked or snooped on are however extremely extremely unlikely unless you are an international person of interest. I agree if your concern of the ability to take control of your device or read your data don't buy the cellular module. I've become comfortable with knowing what is possible but knowing I'm not a person of interest to anyone that if someone is really looking for information on me, they can have it.
 
Last edited by a moderator:
It's a true GPS.

I as well would like a GPS but don't want to pay for the modem if I can't get the PLS8, so I'll just survive without until EvilDragon starts providing upgrades to the 4G modem.
 
I don't know much about that new UHS-II or something standard, but would it be possible to add the additional lines it needs to at least some of the SD ports? If that was easily possible I think it could be worth it as it would make the base board very future proof.

And I think maybe it would be a good idea to put the eMMC on the CPU module because that's something you are going to want to upgrade earlier than the rest of the base board (SD slots, especially with the above suggestions, USB3, controls - takes a lot longer to outdate).
 
Last edited by a moderator:
Well, if you're really worried about those, you can make sure to reserve enough pins on the connector for that, and possibly terminate them on the mainboard that does not support them. Assuming they have the pins to burn.
 
Well, if you're really worried about those, you can make sure to reserve enough pins on the connector for that, and possibly terminate them on the mainboard that does not support them. Assuming they have the pins to burn.
What I wondered was whether we should actually support UHSII on the mainboard (and SD connectors) for future SoC modules. If the cost was like <$5 I think it could be worth it. But are those UHS II SD card cages cheap, backwards compatible, /are/ there spare pins on the connector, etc? That's what I wondered.
 
Is the module that expensive though that ED will populate according to territorial demand? ED?
I can't speak for what ED can get, but I'm seeing single chips for 57 EUR for the European/American versions, 67 EUR for the "world" version, and 105EUR for the 4G versions.
Great news on the eMMC with uSD override capability! That is AWESOME.

Personally, I would really like an option to skip over the 3G US version and go straight to 4G.

I suspect that of the audience who are willing to pay 55+EUR for mobile data (3G or 4G), a majority of them would be willing to pay the additional 50~EUR to get the 4G connection rates. To make this work they're also going to need to pay a phone/data provider - so this is just the 'parts' expense of a much more expensive endeavor. Here in the US, they charge by the phone and data cap - not by the speed of the connection. A 3G or 4G phone/data plan costs the same.

Is 4G considered to be a 'worldwide standard'? Maybe just skip the 3G options entirely?

This may leave out the people who would buy the phone chip just to get GPS though - is there an ultra-cheap module that can be put in instead that just has GPS? Make the the 'default' non-mobile data Pyra?
 
Back
Top