Choosing a micro-controller for mass production

[Matienzo]

Hi all,

I'm working on a personal project that I prototyped using an Arduino uno. Now I'm paying a guy to make a more streamlined version (mine had several issues I didn't know how to solve) and I would like to make the prototype using the microcontroller that will be used on the final design.
My question is what would be a good choice for mass production to replace the Atmega 328?

My product has 2 knobs to set temperature and time, 4 thermistors, and 4 PID loops (one for each heating element).

Please let me know if I'm omitting information that would allow you to give me a proper answer.

Thanks a lot!

Matienzo

PS: Someone suggested SMT32. Any thoughts?

 

[ronsimpson]

You should look at the cost of production. I am thinking about how to program the micro. In volume you do not want to plug each part into a programmer. Some parts I just send a file to the distributor and for a small amount they program 10,000 parts in no time.

What percent of the memory are you using? AT328
What percent of the CPU power are you using?
I think you are only using 1/2 the pins.

 

[Matienzo]

Hi ronsimpson

What percent of the memory are you using? AT328
What percent of the CPU power are you using?
I think you are only using 1/2 the pins.

When I compile my crude sketch the memory used is 18%. The person writing the code now might use, say, 30%.
%CPU power I'm not sure how to tell. Also, my temp control is not PID, instead is a simple cut off temp control.
I'm certainly not using all the pins.

 

[ronsimpson]

%CPU power

Most temperature projects do not really care of you loop 1000/second or 1/second or 0.5/second.

I'm not sure how to tell

I often add "pin-hi/pin-low" in the main loop so I can watch how often things happen.
Some times I "pin-high" at the start of something and "pin-low" at the end.
Adding a "heart beat" on a unused pin can help with real time events. (LED and resistor) You do not have to populate the LED for production but helps for testing.
My guess is you could use a much slower part.
If you will make 100 parts then it is not worth the headache to save money. If you will make 100,000 then $$$$$.

One option is to go into production the ways things are. If the product looks like high volume then redesign for better money.

 

[KeepItSimpleStupid]

You want to define I/O, not knobs.

Knobs could be potentiometers (Analog) or quadrature encoders (Digital).

 

[Pommie]

Why not stick with the ATMEGA 328. They are only $1.50. Does your programmer know how to setup all the peripherals without the Arduino kernel?

Mike.