Do you have a question on Embedded electronics and controls? Ask our Expert, Paul Clarke

Hey Paul,

I am starting out with Embedded "stuff" and I had a question. What's the deal with the various manfuactuers in terms of the chips? I want to design something that uses an ARM7 or above and I see that you can get the processor chip from Atmel, Philips, Samsung, etc.

What motiviates a circuit designer to use one brand over another?

Also, we would like to work in the .net MF environment. If you know of any procs that have a ported .mf framework, any insight would be appreciated. I looked at the other IDEs and don't really want to use assembly or an Embedded brand derivative of C for development. After like 8 languages, it gets crazy to go after more if you know what I mean.

Many Thanks!

Ed

Hey Paul,

I am starting out with Embedded "stuff" and I had a question. What's the deal with the various manfuactuers in terms of the chips? I want to design something that uses an ARM7 or above and I see that you can get the processor chip from Atmel, Philips, Samsung, etc.

What motiviates a circuit designer to use one brand over another?

Also, we would like to work in the .net MF environment. If you know of any procs that have a ported .mf framework, any insight would be appreciated. I looked at the other IDEs and don't really want to use assembly or an Embedded brand derivative of C for development. After like 8 languages, it gets crazy to go after more if you know what I mean.

Many Thanks!

Ed

Well the good thing is that your asking the question. There is no real answer to be totaly honest and you will get differnt answers for everyone. They all have good points and bad points. My worry however is that you have so far selected a chip but dont know what to design... I'm going to say you have started the wrong way round. But I understand where your comming from.

I have a tech artical on EE_Web about chip selection that should help and explain my comment above better : http://www.eeweb.com/blog/paul_clarke/the-right-processor-for-the-right-job

As for .NET I have no experiance but you will see that I suggest this should only be part of your selection process, not set in stone.

I hope that helps. And good luck!

You don't really give enough detail about the kind of work you plan to make any but the most general advice possible. Embedded covers everything from very simple (like a relay timer with 2 logic level inputs and 2 outputs) to very complex (like a network router or a Satnav).

If you come back with a bit more detail I'll try to help.

Regardless of the size of the job I would think very carefully about the .Net Micro Framework. I think it would be fair to say that this has not been the succes that MS hoped it would be (and they have "open sourced" it). There are very few ports and it looks far too resource hungry for simple applications. For applications which are more like an embedded PC I think you would probably be better off with WinCe (or Windows Embedded Compact 7 as we are meant to call it ) which would allow you to stick with Visual Studio type development tools and keep you in the mainstream of the MS eco system.

The good news (if you go that way) is that modern (Cortex) ARM micros are programmed in standard C and the use of assembler is not required.

Michael Kellett

Some information to consider:

ARM Holdings

"ARM Holdings is the world's leading semiconductor intellectual property  (IP) supplier and as such is at the heart of the development of digital  electronic products."

"The ARM business model involves the designing and licensing of IP rather  than the manufacturing and selling of actual semiconductor chips. We  license IP to a network of Partners, which includes the world's leading semiconductor and systems companies."

Take a look at the links below and the columns for Memory, Peripherals, & Embedded Interface.  These are all features in addition to the processor core.

168 Product results found for “arm7”

Also see this Wikipedia article.

The .NET Micro Framework is currently supported on ARM architecture processors (including ARM7 and ARM9)

For development tools please look here.

...With referance to my point before - engineers should be encoraged to select a chip on what they need, not bindly use a ARM or .NET just cos its sounds nice...

@Tech_Support

I think Ed is aware that ARM are a nice bunch of jolly chaps. His problem is deciding which ARM licensee (if any) to buy chips from. I think we need to know more about his application before we give too much advice.

In my case, we buy a product manufactured by others and programmed by others per our spec. The device uses an ARM5 processor. We wanted to refresh and change a bunch of different things to "do it ourselves". So, on the surface, anything greater than an ARM5 in terms of feature-set, speed, etc. is better.

Other than using ther standard IO (Analogs & Digitals; I need 12 analog inputs and 8 digital inputs with 8 digital outputs), we need to have ethernet (10/100) and at least one serial port. There is no external display requirement so we won't have to worry about that.

Ultimately, think of the application as one where we are using a serial port to read/control the analogs and digitals of the proc and an ethernet port that allows the same along with a CLI and a small web server. There is more in terms of whats connecting to the analogs and digitals but until I can get this going, that part of the concept is moot. I can verify the values with a meter on the output and short the digitals to test the inputs.

I get what you are saying about the .net CF. I see that it tends to take up a lot of overhead. My concern with going with a standard C implementation is everything has to be created from scratch. For instance, if I want a simple web server, I would have to decode the HTML and then respond back per HTML standards. With a .net implementation, it can handle the actual back and forth syntax for me.

Are there C libraries available that handle that type of stuff that you know of?

Also, if we went that route, what IDE would be used for Standard C?

If we went with a WinCe implementation, does that require Microsoft licensing?

Sorry for all the questions but I am trying to rapidly wrap my head around this.

Thanks so much!

Ed

Hello Ed,

There are several ARM Cortex micro controllers which support the kind of IO and Ethernet you describe. You need to look carefully to see that the peripherals are good enough. ST, NXP and TI (Luminary) all offer potentially suitable devices. Some of the TI Cortex M3 parts have on chip ethernet MAC and PHY which can be good, NXP and ST parts need an off chip PHY.

My experience is with the Keil(ARM) tool set which is quite pricey but works. You code in C but they offer libraries for a basic real time scheduler and very basic web support. You have spotted the gap in this approach compared with .net (or similar)  - once you get into serious web serving you are going to have  a lot to do yourself. It all comes down to volumes - if you are going to make tens of thousands it may well be worth the software effort to use a cheap micro - if the volumes are low then Wince (or similar) on a bigger processor makes more sense.

I don't know enough about your application to call it one way or annother but I feel that if that web server thing might grow you could save yourselves a lot of grief with a big OS, on the other hand the micro controllers coded in C are much easier to deal with for simple IO.

It's never easy ! (and if it was no one would pay you to do it)

Michael Kellett

@Paul,

The .net, for me, is because I already use about 6 programming languages daily. Adding one more has implications in terms of cost of the IDE, ramp up time in terms of using the IDE and the associated language requried, etc.

I have explained the choice of ARM above; in short it's because of what's in our product already. Also, from what I have seen, it's the "in fashion" proc. Having said that, since I am just starting out, I want a proc that I can scale up or down as needed.

I read your article (thanks for the link) and I agree that I don't want to go overboard on the proc if I don't need to. However, in my opinion, there are other factors than just what you are trying to physically do with the proc. For instance, if you create a design using a proc that totally fit your requirement but was in high demand and not easily available vs. one that was a bit more than you need and readlily available (at a small price difference), you may go with that, right?

Hi Ed,

As I said - you have to justify to yourself why you have selected a chips, tools etc you use. Yes you can ask about different devices and listen to people but in the end you have to make the choice not just follow bindly what you have done before, or others.

All best.

Hi ed,

In all fairness, most MCU's have the same caracteristics for general purpose applications.  Since you have chosen a processor, it is very important for you to define the device you plan to build and work out all of the processing threads, inputs and outputs.  You also need to assess your timelines for event responses.  At that point, you can look at your selected MCU to see if it meets those needs.  In some cases, you might have to use some extra hardware to make up for MCU limitations.

I have worked on many projects where the hardware was predefined and the software job was to make everything work.  This approach violates all reasonable system design processes, but you can still do the job.  It just makes it more risky to satisfy all of the requirements.

Your MCU selection is based upon your familiarity with the processor.  I can see your point in selecting an item you know well.  You will have to keep us posted on your progress so we can see how your decision plays out during your project.

Paul, I apologize for jumping in here, but I sensed the need to assess the issues and open up options for Ed.  While his approach is risky, it is not an unprecidented method used to develop prototypes for future production projects.

To all of you following these posts, I am prepared to start a series of blogs on the Systems Engineering Process if there is interest in doing so, but since the effort is non trivial, I need some assurance that my effort is desired.

Thanks

DAB