Tiva C Series versus Concerto?

Thanks, Vito!

We will not need Ethernet.

Jordan Rendall said:

will require a 32-bit processor, and at least 256kB of memory.

Unstated is the calculus to make this 256KB specification.  Our group also data-logs - address multiple, varied sensors (even MEMS) and has found the M4 a good choice.  (from this & multiple other vendors - depending upon Ap specifics)

Perhaps of benefit to you/others - we found that meeting both memory requirements and speed of acquisition was best achieved via 2 MCUs.  And - when acquisition speed would allow - one of these MCUs, "went to sleep." 

Beware use of RF - both for robustness of signal xmsn. - and for extra/demanding current draw.  We found wired (RS485 multi-drop) link provided much improved robustness - and w/out power demands of "key-down."   (beware - some RF devices demand way high current - even to receive!   And may be in active receive - for prolonged periods...) 

cb1_mobile said:

Unstated is the calculus to make this 256KB specification.  

We will need to store 500 acceleration data samples per second for each of four accelerometers, as well as 2 temperatures (requiring 21 bits - 1 for sign, 8 for integer, 4 for 1 decimal place, 8 for addressing the machine). Therefore, we need a 32-bit processor.

We need to transmit this data from the individual machines to a central computer, constantly updating the information.

32*500*4+32*1*2 = 64064 bps

Originally, we had thought that we would need to send data from node to node, hence each node would need to be able to hold 3*64064 = 192192 bits.

Hence, 256KB would meet the requirements.

256KB calculus now described/stated - and causes much concern!   256KB of SRAM is far beyond capacity of most (maybe all) ARM MCUs - in this spectrum.  (believe these newest devices "top-out" at/around 32KB)  And - moving 32 bits along an 8 bit bus?  Thus - in spite of initial responder's, "both could do the job" - this reporter simply asks, "How?"

Data flow & interface - MCU and 4 accelerometers - @ 500 captures/Sec - constantly - perhaps highlights our earlier, multi MCU suggestion.  (and - in your case - multiple MCUs may be the most reasonable means for you to reach SRAM requirement - and save the cost, complexity, added transfer time & power draw dictated by external memory)

Our experience saw 11 channels - and was simplified & speeded by first limiting one sensor to one ARM MCU.  We experimented/tweaked this component base - and that analysis argued clearly for, "multi MCUs." 

And - as often happens - client soon required "more" - thus a "cram into single MCU" would have imposed a limitation - expensive & time-consuming (even prohibitive) to overcome.

Use of an RF link (always attractive) may also prove a speed/power challenge.  Multi-channel RS-485 (with parallel guns blazing) was the clear winner - our case... (RF could never match wire's robustness - nor speed - nor design EASE!)

Wish you well...

Note that Concerto is a 2-core device - with DSP plus Cortex-M3.

Tiva are just single microcontrollers.

Has anyone programmed both the Stellaris/Tiva and Concerto microcontrollers to provide an opinion on relative ease of programming?

No I haven't used Concerto.

But the two cores are entirely different architectures - so they will require entirely different colpilers. I can't imagine how that can possibly be "easier" than programming a single architecture...

We are designing a hand held instrument that will only have about 1000 units built. We have two people devleoping the product. We also liked the idea of having a processor dedicated to har real time activites and another handling user interface and comms. Rather than going for a Concerto we went for two Stellaris/Tiva chips. One programming envirnments. Another benfit is that with two peopel rpgramming, we each have our own processor - no fighting for limited RAM, processor cycles etc.

So, if you have low volumes, rather than a Concerto chip, consider two Tiva chips.

Vito Casa said:

if low volume... rather than Concerto...consider two Tiva chips.

Indeed - volume plays an important role.

However - in real estate it is, "location - location - location."

And - in our MCU world - might it not be, "application - application - application?"  (primary MCU determinant...)