It’s really too bad this isn’t a more realistic option (financially.). I think for a lot of people the desire to incorporate the G120 components (there aren’t that many) in their board directly isn’t driven by a desire to save money, but to not make it so obvious they used an OTS module. If GHI just sold the uCs with the bootloader installed, I’ll bet many would happily pay (full price of G120) - (GHIs cost for external components) for the bare uC. GHI makes the same profit for less work and customer has what he wants.
Unless it’s horribly difficult to put the bootloader on the bare chip.
If you do cost analysis of the increase in cost on your board from 2 layer to 8 layer, the cost of laying out the PCB, the cost of material on G120 module and the cost of GHI licensing this to you, you will find it much more economical to use the module at lower volumes.
We have done this with many customers in the past. It seems like a good idea till you factor in every variable and then, at 100 pcs or sometimes 1000 pcs, Using G120 as-is is the way to go. Now, if you need 100,000 pcs then you are right, licensing is usually the way to go.
If you have further questions please contact us directly so we can learn more about your project, volumes, needs and we will surly help you find the best economical way.
Just my $0.02, but I used to feel the same way, then I asked my customers.
I’ve never had an issue using OTS modules with customers. Normally they are about, in order:
Performance (to spec)
Given that the majority of embedded design time is now spent doing software (especially for applications where NETMF is an option). I can’t imagine a customer objecting to the thousands of dollars saved by not having to get an RTOS or bare metal micro up and running.
I suspect that the “customer is worried about me using an OTS module” is more accurately stated as “I, the designer, am worried that the customer will think I’m cheating”.
If you give the customer a solution to their problem they won’t care how you got there.
I agree with you when you say that robably under 100k units, it is uninteresting to rebuild the wheel, but something is also important when you start in production line : how long will the SoM be able for production, and maintained in framework evolutions ?
If it is less than 3 to 5 years, it should be a problem for startup like our that can not egange a redesign every year and must concenrate to the soft part of the product more than the hard.
Wont GHI decide to do what did with the chipworkx stand on the 4.1 version of the framework, and not upgrade it to the 4.2 ? It was my first choice for the past year, but when I saw that upgrade will not come on, I had to think about another choice…
More over that key point, my first question was more to be able to adapt the chips positionning on my board to some physical constraints I have.
@ Louis, but the decision any vendor makes on a technology platform like CWX is something you, as a downstream consumer, can’t control or predict. And in some regards, when you’ve jumped on a horse you can continue riding with that horse until there’s some compelling reason to find a new horse. CWX isn’t “dead”, sure it’s not being developed further with 4.2 and beyond, but GHI have a rock solid 4.1 platform, that contains all the “Premium” goodies you want, and that I expect they will continue to provide support for, for those commercial customers; what the limitations of the support they offer their commercial customers I can’t say, but it’s a similar model to Microsoft’s Extended Support Phase. You can’t get a bug fix done as “general maintenance” but you can still have a problem looked at to see if it’s your use of the product or the product itself. If you have a widget with a CWX in it, and it’s in 1000’s of shops/hospitals/cars/whatever, you’re probably not looking to migrate to 4.2, you’re probably in “maintenance” mode. For new products / designs you may choose not to pull out old faithful CWX and make another design with it, you might look at something that doesn’t have a documented “don’t use in new designs” tag, but to keep your widget going there’s no reason not to continue using 4.1 to it’s fullest.
Edit: Just on your point about the suggestion that G120 could go the same way with CWX; whether you’d decided to put the layout on your own board and find a way to license the firmware, or whether you’ve just used the module, a (hypothetical) decision that said G120 stops at 4.2 and will never get 4.3 will still leave you in the same position as you would have been in once that decision is made - either keep using it in current form on current framework, or look at what’s the next big thing.
I do totally get your point about layout constraints; that’s where an option like this would be killer. That’s also where Hydra probably suits, with a totally open design. Then again, it can always be worth a direct call to the GHI team who can do more for a commercial customer than they can for the community here.
OK Brett, but behind the GHI Premium, there’s also Microsoft NETMF, and how to be sure that some issues will be solved on the N-1 version when the N version is released ?
Moreover on NETMF that unfortunately, is not even promoted by Microsoft as a real Embbeded efficient way of developing IoT products, while they do a punchy marketing around Windows 8 that wont be avaiable on ARM until several years…recent news are that a WEC2013 will be sold during the meantime… fine ! :-((,
Keep products up to the last framework is in all case a pledge of durability and availibility for evolutions you may not need at first, but want to implement in future versions of the soft…
you can’t, but it’s open source and you can look at it / propose fixes yourself. But again, if you have 100’s of devices out there you probably aren’t finding things that are groundbreakingly new, so this isn’t always something that you need to be concerned about. Are there any CWX 4.1 bugs that you have not been able to work around?
netmf and WEC are totally different target products, and lets not go into discussions about things like Windows RT. While I work for Microsoft, the only thing I can tell you is that these products do not play anywhere near where netmf does, and netmf doesn’t play anywhere near where these play. They have something about “embedded” in their name is about the only thing in common.