Hi Guys,

I'm new to embedded design and I just want to discuss the concept of conventional process a designer would take from start(picking a new micro) to finish (micro on PCB with implemented algorithms and functions).

Lets start of with my situation: I am currently starting an "improvement" project of a previously designed embedded system and one of the improvements I want to implement is the replacement of the current mico ( http://www.digikey.ca/product-detail/en/CY8C27443-24SXI/428-1598-ND/679709 ) , a Cypress chip, to an NXP ARM based chip ( http://www.nxp.com/products/microcontrollers/cortex_m0_m0/lpc1100/LPC11C24FBD48.html#overview ). I chose this chip because its ARM based, and it has SPI , CAN functionalities and its cheaper.

I have never experienced this before and would like to hear input.

How much would buying the dev kit of the new chip help me? I think that when the time comes to just buying the chip and installing/implementing on PCB board, I would start at zero again and make my own libraries and protocols to use SPI and CAN. Also, how would programming the chip and running my algorithms (ie. main functionalities) be different?

Thanks,
MrBuggy

 

I have stated this before in previous posts.

You don't begin embedded systems design by choosing the microcontroller unit (MCU).
You select the MCU to suit the specifications of the project.

You begin by giving your project a name or title.
Next, you layout all the requirements of the project (not of the MCU).
Next, write the User's Manual. This is one step that is often overlooked and left for the conclusion of the project. With the User's Manual, everyone involved with the project will be able to see completely all the requirements of the project.

You can see that even at this stage the choice of MCU has not yet come into consideration.

 

I have stated this before in previous posts.

You don't begin embedded systems design by choosing the microcontroller unit (MCU).
You select the MCU to suit the specifications of the project.

You begin by giving your project a name or title.
Next, you layout all the requirements of the project (not of the MCU).
Next, write the User's Manual. This is one step that is often overlooked and left for the conclusion of the project. With the User's Manual, everyone involved with the project will be able to see completely all the requirements of the project.

You can see that even at this stage the choice of MCU has not yet come into consideration.

Yes I agree with your opinion. In my case though, there already has been an identification (a general overview) of what is needed and what needs to be done. The selection of the MCU was based on the peripherals I needed which are the SPI, CAN, Timers, PWM and also of my interest in the ARM architecture.

The problem I have is my unfamiliarity of what happens after picking a micro. Right now I have my pseudo code ready, its the configuring, instantiation, programming hardwares required that is till way over my head.

Any advise is appreciated!

 

After you have selected the MCU the process of coding, analyzing, debugging and maintaining proper documentation remains the same regardless of the selected MCU.

 

I consider it bad form to pick a totally new micro architecture, and expect to learn how to use it at the same time you are making a commercial product.

You should "cut your teeth" on an unfamiliar micro with a number of hobby style projects, to get good with that micro first, before expecting to have good success on a large important project.

This might be an oversimplification, but it's a bit like someone saying; "I want to learn to drive in a F1 car, because next week I'm going to race against the pros in Formula 1. Please tell me how to pick which F1 car to learn how to drive in?"

 

I am currently starting an "improvement" project of a previously designed embedded system

So you want to throw away all the existing hardware, purge stock of any existing boards, lay the board out again, throw away any and all code written, then start from scratch and redo the entire project over again? All that to save about a buck per copy?

I'd check the math again to see if there is any savings there. Your boss will note each of these points, and he's the guy who knows what the "loaded labor rate" is.

How much would buying the dev kit of the new chip help me?

If you can use it to prototype your new system then it can be very helpful. You need to build the hardware so it reliably works so you can get the bugs out of the software.

I think that when the time comes to just buying the chip and installing/implementing on PCB board, I would start at zero again and make my own libraries and protocols to use SPI and CAN.

Does the cheap new chip not come with any support?

Also, how would programming the chip and running my algorithms (ie. main functionalities) be different?

Different from what?