I have to ask this... There are so many features of the mbed that make it very attractive for it to be an actual part in a production solution. I especially like that the code can be compiled into a BIN file and simply dragged into the memory of the device.

It having two built in CAN controllers, it's size and speed are also attractive. Like an arduino a custom shield could be created and the mbed simply plugged into it and your good to go.

So why would this not be a great device for an actual production level solution? It's easy to use and has great features...

The only downside I see is not having a local compiler... But even if that went away (mbed folds) the devices already out there would still run...

Maybe mbed needs to develop a branch for folks that want to use this as a solution and not just for prototyping?

Personally I'm more interested in the solution rather then becoming an expert on microcontrollers. Many other businesses are built on this same premise (make it simple for anyone to develop via a GUI)...

Thoughts?

--

Hi Timothy,

You are right, there is no reason why mbed couldn't be used for a production solution. However the main reason why the mbed (or other prototyping tools) are not used for production is cost.

It costs around £40 for an mbed but it is a relatively simple process to take the LPC1768 micro (the micro-controller the mbed is built around) and use it in place of the mbed once the design is proven. The cost of the LCP1768 is around £5. Hence by using this method you save £35 production cost per unit.

That's very true Martin. But if using an mbed directly joined to a board with more descrete parts can serve as a solution, and the cost margin is right, I suppose it would be a fast and easire way to get buisness going... Then maybe version 2 or 3 of product X could be built the way you describe?

Any other reason why this might be a bad Idea?

--

Hi Timothy,

I think your observations are true that mbed can be used in small volume products if applicable, and in fact we already know of some people using it in their systems. As Martin says, there are often cheaper ways to build the same thing, but for small volumes you are probably right that often it could work out well. A lot of it comes down to overall costs and what you value, especially your time and NREs and how that all adds up.

The main reason we aren't actively supporting/promoting it that way is because we feel it is better to be good at one thing first than ok at multiple things. So we are currently focused on helping you get the proof of concept built. Hopefully we're doing a good job already, but we think we can do even better. If we can be successful in that, then I hope there will be opportunities to help more people in different ways, but we won't get that opportunity if we don't get the first part right!

I think the same argument even holds up for building prototypes. For example, we're not trying to be the cheapest hardware out there. But hopefully you get more than just hardware and if we can work on doing things that save you a few hours on your prototype development or let you try something that you perhaps wouldn't have gotten around to, hopefully we've paid for ourselves better than worrying about being the cheapest.

So if it makes sense, go for it, and keep the questions/suggestions coming! It is really useful to get this sort of feedback to make sure we're making the best decisions for everyone.

Happy prototyping!

Simon

Thanks Simon... BTW: This s a great value and very cool little bugger... I just love it.

And there's always this :)

IMO, an issue with low volume product based on the mbed is that if you use the libraries sans source code, and your customers report vague problems, possibly due to end case bugs in the libraries, what quick recourse do you have?

A solution is to not use these, and instead write it all (yuck) or find a commercial supplier to whom you do have recourse, and still use the mbed device.

To make this scenario stronger, an mbed with a new-program downloader via ethernet instead of USB would be needed, for use by a customer who cannot be expected to deal with the USB download method.

In either case, use of the on-line compiler isn't essential. But there may be some terms and conditions hiding in the mbed hardware purchase agreement.

<<quote>>...mbed can be used in small volume products if applicable, and in fact we already know of some people using it in their systems<</quote>>

Do you have any examples of this? I'm trying to justify using mbed at my company as a platform for production test equipment, but I'm getting some push back from the engineers who prefer to use Windows + Labview + Expensive hardware.

Thanks, Jason

Well, yes in fact we use it to make a few information kiosk with LED matrix display. There is also the LPCXpresso that use the same processor & can be put on same base board. The LPCXpresso is cheaper than the mbed & comes with an offline compiler.

mbed would be perfect to use in production test equipment provided you have reliable testing software running on it. I understand why the engineers want Labview & expensive hardware. It is because they trust the results that come out of it. The mbed has yet to prove it. You'll need to actually create a test bed & compare results to prove the point to convince them.