New compute module?


25 posts
by machinabio » Tue Feb 03, 2015 1:07 am
Since the new Pi 2 was announced, I'm curious what the plan is on revising the compute module? The new ARMv7 architecture would make a world of difference to my application. We had to do a lot of extra work to get a NodeJS/Meteor app running on a Pi. It works fine for a demo, but for a production application we were considering switching to a beagle bone specifically because of the ARMv7 architecture. If we had that on a compute module, we could get everything running smoothly pretty easily.
Posts: 2
Joined: Thu Nov 20, 2014 6:44 am
by Briarios » Tue Feb 03, 2015 2:31 am
machinabio wrote:Since the new Pi 2 was announced, I'm curious what the plan is on revising the compute module? The new ARMv7 architecture would make a world of difference to my application. We had to do a lot of extra work to get a NodeJS/Meteor app running on a Pi. It works fine for a demo, but for a production application we were considering switching to a beagle bone specifically because of the ARMv7 architecture. If we had that on a compute module, we could get everything running smoothly pretty easily.



I would have to say is to watch this space, they will probably eventually update the Compute Module, but they will update to the PI2B+ and PI2A+ first, then move to the Compute Module.
http://rpi.pmfg.me -- Test board Eagle Cad Files.
Posts: 117
Joined: Thu Feb 21, 2013 11:35 am
Location: Perth, Western Australia
by GadgetUK » Wed Feb 04, 2015 1:51 pm
If you take a look on the latest the OpenPI kickstarter comment below

https://www.kickstarter.com/projects/wi ... d/comments

Q. When the version 2 Compute Module becomes available will we able to do a straight swap from the version 1 module?
A. That's what I've been told


Looks like compute v2 will be compatible.
Posts: 61
Joined: Thu Jan 19, 2012 6:02 pm
by DougieLawson » Wed Feb 04, 2015 2:03 pm
Despite the fact the the Compute Module2 isn't announced so this is speculation. Wasn't the whole reason for using a SODIMM connector exactly that a newer, sparkier board would simply work as a drop-in replacement with no external changes needed to the board that hosts the CM. That means the folks building on the CM platform get long term hardware support with no re-tooling needed when a new version (if there is one) appears.
Microprocessor, Raspberry Pi & Arduino Hacker
Mainframe database troubleshooter
MQTT Evangelist
Twitter: @DougieLawson

Since 2012: 1B*5, 2B*2, B+, A+, Zero*2, 3B*3

Please post ALL technical questions on the forum. Do not send private messages.
User avatar
Posts: 28099
Joined: Sun Jun 16, 2013 11:19 pm
Location: Basingstoke, UK
by RaTTuS » Wed Feb 04, 2015 2:11 pm
there will be a PI2 CM but it will not be done soon.

it will be a plug in replacement

[gleamed from Ebans statements - I don't have links sorry]
How To ask Questions :- http://www.catb.org/esr/faqs/smart-questions.html
WARNING - some parts of this post may be erroneous YMMV

1QC43qbL5FySu2Pi51vGqKqxy3UiJgukSX
Covfefe
User avatar
Posts: 9181
Joined: Tue Nov 29, 2011 11:12 am
Location: North West UK
by gsh » Wed Feb 04, 2015 2:21 pm
Official statement:

We don't specifically guarantee to be backward compatible because it may not be possible in future to support a drop in replacement (let say for example we needed a different power supply, it simply wouldn't be available)

But we'll do everything in our power to try as hard as we can!

But to put your mind at rest yes a CM2 design will work in a CM platform without changes. Only caveat is the power supply requirements (Pi 2 uses more power than Pi)

Gordon
--
Gordon Hollingworth PhD
Raspberry Pi - Director of Software Engineering
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 1152
Joined: Sat Sep 10, 2011 11:43 am
by wirelesstings » Sat Feb 07, 2015 10:19 pm
Hi Gordon,

Does that mean the regulator is different on the PI2 than the Compute base board. We might have time before the OpenPi kickstarter ends to revise if necessary.

Cheers

Miles
Posts: 3
Joined: Wed Jan 28, 2015 1:16 pm
by PrescottJoule » Mon Feb 09, 2015 10:31 pm
I think it will probably be a good while before they make a new CM because the 2 is brand new. Also, to many the extra processing power is unnecessary and would be considered overkill.
Posts: 16
Joined: Thu Feb 05, 2015 11:22 pm
by mikemoy » Fri Feb 13, 2015 3:24 pm
Tiny bit disappointed that it will be backward compatible, though I understand why.
I would have really liked to see you guys throw on the LAN9514 onto the CM.
Posts: 50
Joined: Sat Sep 06, 2014 12:38 am
by gsh » Fri Feb 13, 2015 3:35 pm
Except that makes it significantly more expensive for people who don't want a 9514... There are significant people who just want a single USB device (a WiFi dongle for example)
--
Gordon Hollingworth PhD
Raspberry Pi - Director of Software Engineering
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 1152
Joined: Sat Sep 10, 2011 11:43 am
by plugwash » Fri Feb 13, 2015 4:05 pm
mikemoy wrote:Tiny bit disappointed that it will be backward compatible, though I understand why.
I would have really liked to see you guys throw on the LAN9514 onto the CM.

why?

The lan9514 is readilly available and in a package of comparable difficulty to the sodimm connector. So if you want one it should be easy enough to put it on your own board. If you don't want one you can easilly leave it out.

wirelesstings wrote:Does that mean the regulator is different on the PI2 than the Compute base board. We might have time before the OpenPi kickstarter ends to revise if necessary.

On the BCM2835 the core voltage is provided by an on-chip adjustable switching regulator supplied through the "VBAT" pin of the SoC. On the Pi1 "VBAT" is connected to the 5V line. On CM1 "VBAT" is brought out to the edge connector and the hardware design guide says you can supply it with anywhere from 2.3V to 5V. On the compute module IO board VBAT is connected to 3.3V (dunno why they went for 3.3V and not 5V)

On the BCM2836 the onchip switching regulator is gone. On the Pi2 there is an external adjustable switching regulator to power the core. Based on what GSH has just said I would expect that CM2 will have a similar core voltage regulator on the module with it's input supplied from the VBAT pin on the module edge connector.

So the main thing is to make sure whatever you are using to supply VBAT has some spare capacity. Personally for a 5V powered design I'd just hook it up to the incoming 5V supply.
Forum Moderator
Forum Moderator
Posts: 3217
Joined: Wed Dec 28, 2011 11:45 pm
by jasonsmedley » Mon Mar 23, 2015 5:22 am
Do you have any rough timeline of releasing an updated compute module. We would prefer to launch with a CM2.
Posts: 1
Joined: Mon Mar 23, 2015 5:17 am
by jamesh » Mon Mar 23, 2015 9:45 am
Probably at least the end of this year. I don't think the design has been started yet.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Please direct all questions to the forum, I do not do support via PM.
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 17202
Joined: Sat Jul 30, 2011 7:41 pm
by RaTTuS » Mon Mar 23, 2015 10:03 am
as the RAM is not POP now it will be more a a redesign on the package than a simple replacement chip
How To ask Questions :- http://www.catb.org/esr/faqs/smart-questions.html
WARNING - some parts of this post may be erroneous YMMV

1QC43qbL5FySu2Pi51vGqKqxy3UiJgukSX
Covfefe
User avatar
Posts: 9181
Joined: Tue Nov 29, 2011 11:12 am
Location: North West UK
by Jim Manley » Tue Mar 24, 2015 9:08 pm
RaTTuS wrote:as the RAM is not POP now it will be more a a redesign on the package than a simple replacement chip

Given that the SoC connections should be identical to those on the Pi 2 B (which for all intents and purposes _is_ a Pi 2 B+ from what I can tell - 4 USBs, B+ voltage regulators, "non-mounting alignment" holes, etc.), the PCB layout for that portion of the CM2 is already done. Since the flash RAM for the CM2 will likely be the same 4 GB on the CM1 that part of the layout is also done (lots of OS features and app packages fit well within 4 GBs of space). Most of what needs to be done is routing Pi 2 unique connections (if any) to the SODIMM board edge connectors, and the second 90% of what needs to be done is the FCC/CE compliance testing and mods (the first 90% of a project takes 90% of the allotted time, and the last 10% takes another 90% of the allotted time ;) ).
Last edited by Jim Manley on Tue Mar 24, 2015 9:20 pm, edited 2 times in total.
The best things in life aren't things ... but, a Pi comes pretty darned close! :D
"Education is not the filling of a pail, but the lighting of a fire." -- W.B. Yeats
In theory, theory & practice are the same - in practice, they aren't!!!
User avatar
Posts: 1593
Joined: Thu Feb 23, 2012 8:41 pm
Location: SillyCon Valley, California, and Powell, Wyoming, USA, plus The Universe
by 10robinho » Tue Mar 24, 2015 9:10 pm
I hope we will see CM2 soon, I'm sure there are many industrial customers (like us) that already have big plans with this little thing :)
Posts: 44
Joined: Thu Feb 28, 2013 7:42 pm
by Jim Manley » Tue Mar 24, 2015 9:31 pm
10robinho wrote:I hope we will see CM2 soon, I'm sure there are many industrial customers (like us) that already have big plans with this little thing :)

I've been disappointed, but not surprised, by the slow uptake on the CM, at least in production volumes. I wonder how many dev kits were sold, as that would give a clue as to how much interest there really was. My take was that netbook and tablet component manufacturers were some of the larger anticipated customers and given that we see new Android tablet designs every week (each of which can't possibly be making any money on their own, what with their paltry individual model sales numbers) it is a bit puzzling why we haven't at least seen a basic netbook based on the CM(1) (and the performance and screen size would put such a device firmly in the netbook, and not laptop, category). The Slice doesn't really count since its form-factor is anything but modest in size, composition ... or especially price - I have no idea why they used the CM(1), particularly in light of Gordon's statement where drop-in compatibility between a potential CM2 and the CM(1) isn't guaranteed, AIUI.

[edit] OTOH, I can see where a very convincing educational tablet could be based on the CM(1) and the forthcoming Foundation touchscreen (next month? Please? Pretty please?? Pretty please with sugar sprinkled on top??? Pretty please with sugar sprinkled on top, wheels, and a sandwich???? :D ). You read it here first, folks! No, I do not have any inside track on this and any wagering done based on this bald-faced (who, _me_?) speculation is a fool's errand wrapped in an Enigma Machine, and surrounded by a Puzzle Palace! ;)
The best things in life aren't things ... but, a Pi comes pretty darned close! :D
"Education is not the filling of a pail, but the lighting of a fire." -- W.B. Yeats
In theory, theory & practice are the same - in practice, they aren't!!!
User avatar
Posts: 1593
Joined: Thu Feb 23, 2012 8:41 pm
Location: SillyCon Valley, California, and Powell, Wyoming, USA, plus The Universe
by jamesh » Wed Mar 25, 2015 10:04 am
Why do people think the CM hasn't been selling? Figures have never been released.

Consumer items (like Slice) are only one part of the CM's abilities, there are lots of industrial applications that could use them that the huge majority of people would never hear about.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Please direct all questions to the forum, I do not do support via PM.
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 17202
Joined: Sat Jul 30, 2011 7:41 pm
by plugwash » Wed Mar 25, 2015 11:29 am
Jim Manley wrote:
RaTTuS wrote:as the RAM is not POP now it will be more a a redesign on the package than a simple replacement chip

Given that the SoC connections should be identical to those on the Pi 2 B (which for all intents and purposes _is_ a Pi 2 B+ from what I can tell - 4 USBs, B+ voltage regulators, "non-mounting alignment" holes, etc.), the PCB layout for that portion of the CM2 is already done. Since the flash RAM for the CM2 will likely be the same 4 GB on the CM1 that part of the layout is also done (lots of OS features and app packages fit well within 4 GBs of space). Most of what needs to be done is routing Pi 2 unique connections (if any) to the SODIMM board edge connectors

The compute module has every IO pin hooked up, the B series doesn't. So the layout under the processor will need work to get those connections out. I don't know how difficult said work will be (it depends a lot on how the ball arrangemnets of the 2836 match up with where the signals need to go).
Forum Moderator
Forum Moderator
Posts: 3217
Joined: Wed Dec 28, 2011 11:45 pm
by 10robinho » Wed Mar 25, 2015 11:36 am
jamesh wrote:Why do people think the CM hasn't been selling? Figures have never been released.

Consumer items (like Slice) are only one part of the CM's abilities, there are lots of industrial applications that could use them that the huge majority of people would never hear about.


I agree with that. As I said couple of posts earlier, I can't wait to get CM2 released.
Companies like ours are going to buy it (and are buying CM1) in 1000s, so sales figures shouldn't be bad, while sellers margins should be higher because of lower packing and logistics expenses.

Anyways, keep up good work!
Posts: 44
Joined: Thu Feb 28, 2013 7:42 pm
by TangoMikeBravo » Wed Mar 25, 2015 3:04 pm
We're interested in using the compute module to replace PC104 Intel processors for subsea applications where the processor has to run at seabed pressures of up to 400 bar. The main problem with any of the Intel-based PC104 units is that there are usually two or three crystals on the board (processor, Etherent and sometimes another, not sure what it's used for). These are pretty much the only pressure-intolerant components and we have to make special provision to protect them from pressure. On the simpler stuff like Ethernet switches, we do that by ripping the crystal off the board and replacing it with a MEMS oscillator, which we have tested and works nicely at the highest pressures we can test (400bar). We can't do that with the PC104 processors though because they're quite densely packed on the PCB making it quite difficult to rework and there's usually not space to fit the MEMS. That means we have to bodge them with a metal cap and epoxy to isolate them from the pressure environment, which works well enough up to 100bar but is unreliable above that.

The RP2 though, and the first generation compute module, are nicely laid out which makes removing the crystal and replacing it a doddle, and I'm about to start a round of pressure testing on them to see if there are any other gotchas, but I suspect they'll be fine, particularly the CM which we would prefer to use because we can then choose known good components for stuff like Ethernet magnetics and the like on a motherboard we can make ourselves.

We're not quite sure that the original CM is powerful enough for us though, but the RP2 certainly is and we're looking forward to the Rev 2 CM (assuming everything else passes the presssure qualification of course). Not that our volumes will make a difference to the Foundation's bottom line, at a few 10s a year though :)
Posts: 2
Joined: Sun Feb 08, 2015 7:56 pm
by 10robinho » Mon Apr 13, 2015 8:27 pm
Any news on this matter?
Posts: 44
Joined: Thu Feb 28, 2013 7:42 pm
by ShiftPlusOne » Mon Apr 13, 2015 8:39 pm
When there are news, they'll be announced in the appropriate places (not on a random forum thread because it was bumped enough times). Please don't do that.
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 4349
Joined: Fri Jul 29, 2011 5:36 pm
Location: The unfashionable end of the western spiral arm of the Galaxy
by jdb » Tue Apr 14, 2015 6:48 pm
TangoMikeBravo wrote:We're interested in using the compute module to replace PC104 Intel processors for subsea applications where the processor has to run at seabed pressures of up to 400 bar. The main problem with any of the Intel-based PC104 units is that there are usually two or three crystals on the board (processor, Etherent and sometimes another, not sure what it's used for). These are pretty much the only pressure-intolerant components and we have to make special provision to protect them from pressure. On the simpler stuff like Ethernet switches, we do that by ripping the crystal off the board and replacing it with a MEMS oscillator, which we have tested and works nicely at the highest pressures we can test (400bar). We can't do that with the PC104 processors though because they're quite densely packed on the PCB making it quite difficult to rework and there's usually not space to fit the MEMS. That means we have to bodge them with a metal cap and epoxy to isolate them from the pressure environment, which works well enough up to 100bar but is unreliable above that.

The RP2 though, and the first generation compute module, are nicely laid out which makes removing the crystal and replacing it a doddle, and I'm about to start a round of pressure testing on them to see if there are any other gotchas, but I suspect they'll be fine, particularly the CM which we would prefer to use because we can then choose known good components for stuff like Ethernet magnetics and the like on a motherboard we can make ourselves.

We're not quite sure that the original CM is powerful enough for us though, but the RP2 certainly is and we're looking forward to the Rev 2 CM (assuming everything else passes the presssure qualification of course). Not that our volumes will make a difference to the Foundation's bottom line, at a few 10s a year though :)


I was about to ask if operations were on the UK Continental Shelf but there you get a paltry 12 atmospheres at depth... You have to be on the Abyssal Plain depths to get 400 bar...
Rockets are loud.
https://astro-pi.org
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 1656
Joined: Thu Jul 11, 2013 2:37 pm
by TangoMikeBravo » Thu Apr 16, 2015 7:25 am
jdb wrote:I was about to ask if operations were on the UK Continental Shelf but there you get a paltry 12 atmospheres at depth... You have to be on the Abyssal Plain depths to get 400 bar...


Arr, no, it's one bar per 10m of seawater depth. The UK and Norwegian oil and gas lines currently operate up to about 1000m or 100bar and are being expanded to areas over 1300m or 130bar. There are fjords in Norway with depths of over 1000m, for example Sognefjord where we have tested some stuff has a maximum depth of about 1300m. Some wells overseas are drilled at about 3000m below the sea surface, hence the 400 bar requirement (to give a safety overhead).
Posts: 2
Joined: Sun Feb 08, 2015 7:56 pm