jars121
Posts: 123
Joined: Tue Jun 25, 2013 8:35 pm

CM3 vs. other commercial platforms

Tue Apr 11, 2017 9:09 am

G'day,

I'm hoping some commercial embedded system engineers/designers can offer some advice as to the validity of the CM3 platform when compared to some of the established SoM and CoM platforms. I'm currently building a prototype embedded system, which I plan on testing on a standard RPi (for the time being), but I want to make sure the CM3 (with custom carrier board) is going to be the best fit for my application before committing to the RPi platform.

Are there any hidden (or otherwise) pitfalls or limitations of the CM3 of which an experienced embedded system professional would be aware, that would steer them towards an established commercial platform (Toradex, Olimex, etc.)?

The requirements for my application are as follows:

- Linux-capable
- Support for capacitive touchscreen
- 10+ analogue inputs
- 10+ digital inputs
- Python-capable, with libraries for sensor input/output
- eMMC
- SD card support (not a 'must have' requirement)
- I'm not attached to a particular architecture, but the solution will need decent graphics performance and multithreading
- WiFI/BT (not a 'must have' requirement)
- USB

The CM3 (with custom carrier board) certainly satisfies each of the above requirements, but I want to assess what other industry-standard embedded system platforms are worth considering.

Thanks!

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 23352
Joined: Sat Jul 30, 2011 7:41 pm

Re: CM3 vs. other commercial platforms

Tue Apr 11, 2017 9:18 am

You already made a comment on this, but for clarity

- Linux-capable - YES
- Support for capacitive touchscreen - YES, the RPF sells a touchscreen, but if you want to interface a differenet one yourself it will need HDMI input or a convertor.
- 10+ analogue inputs -NO, no analogue inputs on the Pi. Would need to be in the carrier, then interfaced perhaps via SPI?
- 10+ digital inputs - YES
- Python-capable, with libraries for sensor input/output - Python yes, sensor libraries up to the sensor vendor.
- eMMC - Built in 4GB
- SD card support (not a 'must have' requirement) - YES
- I'm not attached to a particular architecture, but the solution will need decent graphics performance and multithreading. - YES
- WiFI/BT (not a 'must have' requirement) - This would be down to the carrier board./
- USB - ONE port as standard, more would require a hub on the carrier.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
"My grief counseller just died, luckily, he was so good, I didn't care."

jars121
Posts: 123
Joined: Tue Jun 25, 2013 8:35 pm

Re: CM3 vs. other commercial platforms

Tue Apr 11, 2017 10:16 am

Thank you James, your input is much appreciated as always :)

My current prototype uses a number of MCP3208 ADCs to manage the analog inputs, which would then be managed on the carrier board as suggested.

I've got a RPF 7" screen, so I'll be interested to see how the prototype performs, as the RPF display was one of the major attractions to this platform for prototyping.

dkondrad
Posts: 4
Joined: Thu May 25, 2017 2:43 pm

Re: CM3 vs. other commercial platforms

Thu May 25, 2017 3:03 pm

Greetings,

I have also been considering using CM3 for an industrial project; however, I am running into very strong resistance from our senior hardware engineers. Much of this resistance is due to the "image" of Raspberry Pi: A hobby project for educating and teaching electronics. It doesn't seem to matter that there is little if any physical difference between the CM3 and any other SoM out there. Are there any resources that the Pi Foundation or the manufacturers of the CM3 provide to overcome this?

Specifically, our organization looks for:
  • * Manufacturing Quality Practices
    * Purchasing Channels
    * Reliability Testing (a blog post on the max temperature rating doesn't suffice)
    * Warranty & Maintenance
    * Existing Case
    * Design/Production Commitment (I've seen 2023 quoted in a few places, what happens after that?)
I think it would be a huge win for our organization (and Raspberry Pi in general) if we were able to surmount the qualification hurdles and be able to use the module in our products. A prepared whitepaper or qualification packet that engineers can use to lobby for the integration of Compute Modules is something that really needs to exist in order to get any serious uptake in industrial spaces.

Best regards,
David

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 23352
Joined: Sat Jul 30, 2011 7:41 pm

Re: CM3 vs. other commercial platforms

Fri May 26, 2017 9:31 am

I'm not sure we have anything specific. I can try and answer some of your questions..

* Manufacturing Quality Practices

RS and Farnell make the Pi under licence at the Sony Factory in Wales. This factory is used for lots of other stuff. The quality is high and the Pi boards themselves have a very low reject rate.

* Purchasing Channels

RS and Farnell - both very well know suppliers.

* Reliability Testing (a blog post on the max temperature rating doesn't suffice)

Not sure what would be required here. Making a CM and carrier board would require it own reliability testing. Anecdotally, Pi's themselves are very reliable.

* Warranty & Maintenance

Standard warranty depending on location sold. Maintenance - we don't do maintenance contracts, but probably have one of the best maintained Linux kernels on SoC devices in the world. Support is via the forums or the Raspberry Pi github issues page.

* Existing Case

?

* Design/Production Commitment (I've seen 2023 quoted in a few places, what happens after that?)

I don't know the official line. We intend to keep selling Pi's for as long as people want to buy them. We have a internal product roadmap for the next 5 years at least. Of course, roadmaps can change according to the market.



As for Pi being a hobby project, I think your HW engineers need to have a bit of a reality check. We have sold 13 million Raspberry Pi's, and are now the worlds third largest 'home' computer manufacturer outside of PC and Mac (recently overtook Commodore IIRC). Not bad for a hobby project. Although our charitable aims are well known, a lot of those 13M are sold outside of the educational sector and many many are sold in to industrial sectors. Our development team are also world class, and run in a separate subsidiary to the Foundation (Raspberry Pi Trading Ltd)

I've worked on home grown embedded systems in the past (industrial ink jet printing mainly). I wish we had had access to the Pi - it would have saved so much time and money.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
"My grief counseller just died, luckily, he was so good, I didn't care."

yaheath
Posts: 15
Joined: Tue Mar 28, 2017 9:58 pm

Re: CM3 vs. other commercial platforms

Fri May 26, 2017 4:10 pm

jamesh wrote: Standard warranty depending on location sold. Maintenance - we don't do maintenance contracts, but probably have one of the best maintained Linux kernels on SoC devices in the world. Support is via the forums or the Raspberry Pi github issues page.
I feel compelled to jump in here and comment on the importance of how good the Pi's Linux kernel support is.

For example, I've been maintaining a product that uses an OMAP4. It's stuck on Linux kernel 3.2 because the proprietary drivers don't work on anything newer, and TI has basically abandoned development on the OMAP platform. Sure, you can boot a newer kernel on OMAP4 but you can't use the DSP anymore without the proprietary stuff. So it's essentially dead, at least dead to me as I will refuse to build anything new that's stuck on an outdated linux kernel.

If there's ever going to be a next-generation version of our product it'll be based on the CM3.

dkondrad
Posts: 4
Joined: Thu May 25, 2017 2:43 pm

Re: CM3 vs. other commercial platforms

Tue May 30, 2017 2:34 pm

Greetings,

Thank you, jamesh, for your insight and input.

I've uncovered most of what you wrote, as you say, anecdotally from various media outlets and information available elsewhere on Pi Foundation pages. I was hoping to attract the attention of a marketing/PR type person as well.

Engineer to engineer, we both know the strength and benefits that the platform provides; however, it is frequently not the project engineer that makes final determination of approved parts on a design. We have to satisfy other parts of the business with regards to sourcing, stability, etc. which are mostly vetted by non-engineering types that rely mostly on the available documentation with some investigative reporting thrown in. It also really helps if distribution and sourcing agents throw in their support of the product.

I'll give another example of how perception and the position of the person relaying the information has a huge impact. When I was pitching the idea to use this CM3, we had a visit from an Avnet field applications engineer to discuss options. I brought up the idea of using the module and the FAE immediately shot it down in front of management with similar reasons I mentioned previously. This had a huge impact because: 1) they are a big supplier; 2) They (Avnet) own Premiere Farnell, which distributes the boards. No amount of engineering reasons can overcome simply not having the support of the company that distributes the product.

It looks like I'll have to dedicate some serious time to creating a whitepaper for my organization that pulls all the facts together and makes a persuasive argument toward using these modules in designs. I do, however, strongly believe it would help the cause if someone from the foundation with official capacity to speak on these matters would do this for the general public consumption; not from an engineering point of view, but from a marketing/purchasing point of view.

Thanks again for all the feedback!
I'll report back if I've had any success!

gsh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 1423
Joined: Sat Sep 10, 2011 11:43 am

Re: CM3 vs. other commercial platforms

Tue May 30, 2017 3:16 pm

Hi,

Can you provide an example competitor to the CM3 who provide this type of information? We try to provide as much information as possible for the CM3 device and the datasheet was our attempt at providing this information (since we offered even less with CM).

It would be useful to understand how a competitor to the Raspberry Pi CM3 approaches this problem

Thanks

Gordon
--
Gordon Hollingworth PhD
Raspberry Pi - Director of Software Engineering

dkondrad
Posts: 4
Joined: Thu May 25, 2017 2:43 pm

Re: CM3 vs. other commercial platforms

Tue May 30, 2017 4:36 pm

Hi Gordon,

I think the best apples-to-apples comparison would be Variscite, which offers SoMs in the $40-50 USD range. I'll link some examples of marketing material below, but basically their entire website is a targeted marketing campaign, from the "About Us" video on the main page, to the blurbs in the technical datasheets. It's pretty safe to assume very little of the content of the site was written by engineers.

http://www.variscite.com/
http://www.variscite.com/images/stories ... er2016.pdf
http://www.variscite.com/products/syste ... cale-imx-7

Now, I can't vouch for how much meat is on their marketing bones since I have not used any of their products in designs, but you can't deny that they sure do spin a good story behind picking them as a vendor.

Lastly, I want to mention, in case the message gets lost in all the politics of engineering, that the R-Pi3 delivers, both in hardware and software (and I assume the same is for CM3). I never had something "just work" out of the box for anything more than simple contrived examples. I went from nothing to prototype in about 2 weeks of rapid design (and that was with learning new frameworks). I've had less then stellar experiences with companies that talked a bigger talk but fell short when it came to kernel and software support (yes, I've lived through that TI hell as well).

Thanks for all the hard work that has been put into this product!

Regards,
David

W. H. Heydt
Posts: 10762
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: CM3 vs. other commercial platforms

Tue May 30, 2017 7:05 pm

Is it worth noting that NEC selected the CM3 for the "intelligence" in their new line of commercial displays? Including doing whatever it took to be able to make a 16GB CM3?

Kind of funny in a way...the *cheapest* of the new displays (V404) lists for $1050...but you could use a $30 CM3 to give it progammable capabilities.

dkondrad
Posts: 4
Joined: Thu May 25, 2017 2:43 pm

Re: CM3 vs. other commercial platforms

Thu Jun 01, 2017 4:57 pm

W. H. Heydt wrote:Is it worth noting that NEC selected the CM3 for the "intelligence" in their new line of commercial displays? Including doing whatever it took to be able to make a 16GB CM3?

Kind of funny in a way...the *cheapest* of the new displays (V404) lists for $1050...but you could use a $30 CM3 to give it progammable capabilities.
Yeah, that was my first line of defense.

The response I got was something along the lines of "NEC is a large company, they probably bought the IP rights and are manufacturing the module to fit their specifications in their own facilities." Of course it doesn't help the counter-argument that you see an NEC version of the module for purchase.

There are those in the organization that are sympathetic to my cause... we'll see if there are enough to rally around a design once I lay out all the facts.

I know it seems strange to install a $30 module in $1000+ monitor, but looking at their value proposition, it makes perfect sense. They are replacing about $1000 dollars of customer provided equipment and wiring that would have to be used if a non-NEC monitor is used; that right there is a heck a sales pitch because I could spend an extra $500 on a bigger/brighter/etc monitor and still be keeping $500 an install in my pocket.

As I mentioned before, the other key to CM3 in a NEC monitor is that the software just works. Given that you already have the content created, an IT intern can throw a few scripts on the module to play it back or whatever. NEC has even gone as far as pre-installing the codec licenses for most common video formats.

Regards,
David

W. H. Heydt
Posts: 10762
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: CM3 vs. other commercial platforms

Thu Jun 01, 2017 5:50 pm

dkondrad wrote:
W. H. Heydt wrote:Is it worth noting that NEC selected the CM3 for the "intelligence" in their new line of commercial displays? Including doing whatever it took to be able to make a 16GB CM3?

Kind of funny in a way...the *cheapest* of the new displays (V404) lists for $1050...but you could use a $30 CM3 to give it progammable capabilities.
Yeah, that was my first line of defense.

The response I got was something along the lines of "NEC is a large company, they probably bought the IP rights and are manufacturing the module to fit their specifications in their own facilities." Of course it doesn't help the counter-argument that you see an NEC version of the module for purchase.
I'm working in a desultory sort of manner on finding out how much the CM3-16 NEC module sells for. NEC did actually state that their modules would be available for purchase independent of buying their monitors....so I am rather curious about just what they want for the 16GB version.
There are those in the organization that are sympathetic to my cause... we'll see if there are enough to rally around a design once I lay out all the facts.

I know it seems strange to install a $30 module in $1000+ monitor, but looking at their value proposition, it makes perfect sense. They are replacing about $1000 dollars of customer provided equipment and wiring that would have to be used if a non-NEC monitor is used; that right there is a heck a sales pitch because I could spend an extra $500 on a bigger/brighter/etc monitor and still be keeping $500 an install in my pocket.

As I mentioned before, the other key to CM3 in a NEC monitor is that the software just works. Given that you already have the content created, an IT intern can throw a few scripts on the module to play it back or whatever. NEC has even gone as far as pre-installing the codec licenses for most common video formats.

Regards,
David
I wish you all the luck in the world in pitching the CM3 to your Powers That Be.

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 23352
Joined: Sat Jul 30, 2011 7:41 pm

Re: CM3 vs. other commercial platforms

Fri Jun 02, 2017 9:54 am

dkondrad wrote:
W. H. Heydt wrote:Is it worth noting that NEC selected the CM3 for the "intelligence" in their new line of commercial displays? Including doing whatever it took to be able to make a 16GB CM3?

Kind of funny in a way...the *cheapest* of the new displays (V404) lists for $1050...but you could use a $30 CM3 to give it progammable capabilities.
Yeah, that was my first line of defense.

The response I got was something along the lines of "NEC is a large company, they probably bought the IP rights and are manufacturing the module to fit their specifications in their own facilities." Of course it doesn't help the counter-argument that you see an NEC version of the module for purchase.

There are those in the organization that are sympathetic to my cause... we'll see if there are enough to rally around a design once I lay out all the facts.

I know it seems strange to install a $30 module in $1000+ monitor, but looking at their value proposition, it makes perfect sense. They are replacing about $1000 dollars of customer provided equipment and wiring that would have to be used if a non-NEC monitor is used; that right there is a heck a sales pitch because I could spend an extra $500 on a bigger/brighter/etc monitor and still be keeping $500 an install in my pocket.

As I mentioned before, the other key to CM3 in a NEC monitor is that the software just works. Given that you already have the content created, an IT intern can throw a few scripts on the module to play it back or whatever. NEC has even gone as far as pre-installing the codec licenses for most common video formats.

Regards,
David
NEC do have a custom board, but the only difference is the extra flash memory, required for their particular use case. All the rest is the same. So a NEC CM board will work in a standard CM carrier board.

There was no IP transfer, NEC use the already in place Farnell/E14 board customisation facility, and simply asked for one to be made with 16GB of flash rather than 4. See here https://www.element14.com/community/doc ... on-service

So anyone, not just large manufacturers like NEC, can have custom devices made. Not hugely expensive either.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
"My grief counseller just died, luckily, he was so good, I didn't care."

Return to “Compute Module”