What do you really mean with
"moving a product prototyped on an Edison to a production device",
are you trying to program a lot of Edison boards to develop your product or do you want to add an Edison module to your product.
If you want to program a lot of Edison boards, there is an interesting thread about this Bulk programming?, you could find some suggestions like;
a) Using a "JTAG hub" to program all of them at the same time
b) Create an image with all the packages needed and the script to run your project.
c) Mounting and modifying ext4 image from the Yocto image.
Now, if you want to add a slot to add an Edison compute module in to your own custom board, that it's a good feature of the Edison Compute Module because of it's size and functionality so there are a lot of options to create a prototype of something like this.
I'm not sure if these clear your doubts. If not, please give us an idea of the magnitude of your project and what other questions fo you have
I'm building a product for a startup, that will hopefully at some point go to Kickstarter. The product is basically an Edison module and a number of peripherals (I'd rather not go into more specifics on a public forum). For argument's sake, let's say my Kickstarter is wildly successful, and I get thousands of orders. Does Intel have wholesale pricing on the Edison module? Is it meant to be used in production this way (shipping embedded into other products), or is it just a prototyping platform that isn't meant to be used "in the wild?"
Example - an Arduino board isn't meant to be used as is in a production project, however I can (and have) take the Arduino reference design, repackage it on a custom circuit board and integrate into my project in a way that meets whatever manufacturing / real world requirements I have. Obviously the Edison is a lot more complex than an Arduino, but the analogy is valid.
Please let me know if I'm making sense.
This sounds like you should be talking to a sales representative at Intel. They would hopefully be able to tell you how to quantity buy.
It would be interesting to see what Intel would do should someone approach them with a large order. Chicken and the egg thing, they won't make them till you buy them and you probably won't buy them unless they commit to making them.
About the pricing of the Edison, the price is the same if you buy one or more. If you want more information about this you could contact an Intel Representative using this form. Intel Support
As you know the Edison compute module allows you to set it in your own custom board, because of it's size and functionalism you also could add or remove peripherals for your convenience in the board that you are building.
that's not really an answer to my question. Or at least an indirect one - it basically implies that the Edison isn't meant to be used in mass-produced setting, and that there doesn't appear to be a path to manufacturing. Kind of an evolutionary dead end.
Maybe we are not understanding your requirements. Is your issue with Edison a form-factor, functionality, or price issue?
The form-factor is much more compatible with medium run manufacturing than most embedded solutions: just design a custom board with the 70-pin Hirose connector, and you can easily plug in an Edison board with only the I/O you need. The wifi and Bluetooth is handled for you on the module, which minimizes the design effort if you need those components in your product.
The Edison functionality is suitable for volume manufacturing... Though I personally struggle with Yocto: it is definitely the way I would go for a production environment, as you can control precisely what components are in your production image. But if Yocto's not your thing, there's also Ubilinux which provides an environment for someone more comfortable with a debian experience.
If the price is the issue, I wouldn't classify that as an evolutionary dead-end, nor lack of a path to manufacturing -- just that it doesn't meet your particular price point. For those that need the performance of a dual core Atom, wifi and/or Bluetooth, it's my opinion (and I stress that I speak for myself here and not Intel... but I will admit to being possibly a bit biased :-), that Edison provides class leading capabilities at a price which is competitive to similarly capable solutions.
Hey @intel_jassowski and CMata_Intel,
I'm at the point where I've finished constructing and programming a prototype using Edison on the Xadow platform. I used the Xadow extension because it's form factor is much better for building a wearable than Intel's mini expansion board. Considering that I only added the Xadow breakout board in chain to the main board with the Edison, I only used the Xadow boards to solder 2 sensors and an RGB led to the I2C bus. In relation to this post by kolosy, I was hoping to find a way Intel would allow me to build a final product using a single PCB that includes all my components and the necessary components from the Intel Edison. Having access to the schematic and BOM of the Edison module would allow us to do this. I mentioned necessary components because as it stands right now, the Edison contains an MCU I don't need as I only rely on the CPU to run Yocto and my C++ code. Additionally, while the Edison module is small, it won't be suitable for me if I'm trying to build a new smart band. A band with the width of the Intel Edison would be clunky from a product design perspective in comparison to the Fitbit Flex or Jawbone's Up band. For a wearable application, I might even want to change the Edison components to fit on a flexible PCB so I'm able to build around the human form factor instead of being strictly limited to a 2 dimensional plane.
I've enjoyed developing my prototype using the Edison along with the support from this community but if I won't be able to change the dimensions of the Edison components, I'm afraid my final product design will have to switch to a different platform that better fits my design specs. Either that or I assemble my own compute module that may or may not include Intel's chips. I can understand the Edison module being a perfect fit for IoT products that typically will be much bigger than the module itself, but I had the notion that for wearables where the Edison module is the largest piece, Intel would make it easy for developers to rearrange the Edison components so that the final mass produced product would create new customers reliant on Intel chips.
I guess this would explain Intel's desire to build the Curie module which seems to be more fitting for my project's needs albeit with much less data storage capacity than the Edison. Nearly ideal although the main draw of choosing to build my prototype around Edison was the memory capacity.
Please correct me if I'm wrong about anything I've said but if it all stands, do I have any options as to completing a final product with the necessary Edison parts inside? I'm open to all suggestions and I look forward to your answers.
Let me first say that although I work for Intel, and have used the Edison board, I wasn't involved in the design, nor do I speak with any authority on that design. I think, however, it is possible to look at publicly available sources to get you an answer to your basic question: If you could source the parts and design your own board, could you make the Edison smaller?
I found some photos of the exposed parts (with handy IC labels) here:
The processor and PMIC (along with a bunch of discretes) take up most of one side. The flash memory takes up half the other side. The remaining area is consumed by wifi, antenna, and usb transceiver. Since there is no other IC's on the board, it's probably safe to assume that the MCU is integrated into one of the other components, so that won't be a possible area for optimization.
Not sure if you need all that for your application: The Processor and PMIC are obviously required, and it sounds like the flash memory is needed as well. I suppose, if you wanted to remove the Bluetooth/wifi and usb, you could get some board area reduction -- but the width of the board is not much wider than the SoC itself. However, based on your description, I'm also guessing that both Bluetooth and USB might be desirable in your application.
So, yes: the Curie does sound like it's more suited to your application, although I am not aware of what the final form factors or product capabilities will be; so I can't be much help there.
Good luck on your product!