EmbeddedRelated.com
Forums
The 2024 Embedded Online Conference

32-bit MCU-family inquiry

Started by P_B January 30, 2012
Hi everyone! 

This is my first post here :)

I'm interested in making a platform which has an 32-bit MCU as the central
component. This platform is supposed to be configurable, i.e. do things
that an 8-bit MCU can handle OR do more advanced signal processing.
What I have come up with is ARM Cortex M. From what I've read on the
internet is that NXP has pin- & software-compatible devices in Cortex M0,
M3 and M4.
In addition, STMicroelectronics will have the same this year, they
currently only have Cortex M3 and M4.

I would like to hear some reviews from you guys on the NXP and ST devices.

What would you recommend if you don't want to recommend these, Renesas,
Microchip etc. I am open to all suggestions.
I don't have special criterias at this point, not more than low cost and
low power consumption. But those are almost always criterias, I guess.

Really appreciate any input!!


/P






	   
					
---------------------------------------		
Posted through http://www.EmbeddedRelated.com
On Mon, 30 Jan 2012 07:24:21 -0600, "P_B"
<patrik.bjorklund@n_o_s_p_a_m.rejlers.se> wrote:

>Hi everyone! > >This is my first post here :) > >I'm interested in making a platform which has an 32-bit MCU as the central >component. This platform is supposed to be configurable, i.e. do things >that an 8-bit MCU can handle OR do more advanced signal processing. >What I have come up with is ARM Cortex M. From what I've read on the >internet is that NXP has pin- & software-compatible devices in Cortex M0, >M3 and M4. >In addition, STMicroelectronics will have the same this year, they >currently only have Cortex M3 and M4. > >I would like to hear some reviews from you guys on the NXP and ST devices.
They both work. I've used both and will use both in the future. As to which family is "better" for some definition of better and for a given application, that's why we get the big bucks to make a sensible choice. Read the datasheets and errata and look to see if what you want to do can be mapped onto a given device. -- Rich Webb Norfolk, VA
> >They both work. I've used both and will use both in the future. > >-- >Rich Webb Norfolk, VA >
Which compiler(s) did you use, if I may ask? /P --------------------------------------- Posted through http://www.EmbeddedRelated.com
On Mon, 30 Jan 2012 08:31:26 -0600, "P_B"
<patrik.bjorklund@n_o_s_p_a_m.rejlers.se> wrote:

> >> >>They both work. I've used both and will use both in the future. >> >>-- >>Rich Webb Norfolk, VA >> > >Which compiler(s) did you use, if I may ask?
I'm using Rowley's CrossWorks now. Handles all the ARM varieties I've come across (and many that I'll probably never see!) and supports a large number of JTAG pods, including the old Macraigor Wigglers and the inexpensive Olimex ARM-USB as well as the "majors" like the Segger J-Link, plus their own model. It's based on gcc with their own non-GPL libraries. -- Rich Webb Norfolk, VA
On 30/01/2012 13:24, P_B wrote:
> Hi everyone! > > This is my first post here :) > > I'm interested in making a platform which has an 32-bit MCU as the central > component. This platform is supposed to be configurable, i.e. do things > that an 8-bit MCU can handle OR do more advanced signal processing. > What I have come up with is ARM Cortex M. From what I've read on the > internet is that NXP has pin-& software-compatible devices in Cortex M0, > M3 and M4. > In addition, STMicroelectronics will have the same this year, they > currently only have Cortex M3 and M4. > > I would like to hear some reviews from you guys on the NXP and ST devices. > > What would you recommend if you don't want to recommend these, Renesas, > Microchip etc. I am open to all suggestions. > I don't have special criterias at this point, not more than low cost and > low power consumption. But those are almost always criterias, I guess. > > Really appreciate any input!! > > > /P > > > > > > > > > --------------------------------------- > Posted through http://www.EmbeddedRelated.com
Currently ST have M3 and M4, NXP have M0 and M3, TI have M3 and R4 (you probably don't want this) and Freescale have M4. There are others. The price difference between M0 and M3 in the same footprint is small and using a low pin count will keep you from using the better M3 and M4s. If I were you (and bear in mind I don't know the whole story) I would go for M3/M4 from ST because you can get them now and if you ever hit volumes where it is worth using M0 you can re-spin the board. (10k off you can get an ST M3 for <&#4294967295;0.60, so even if the M0 is half the price (and it won't be that cheap) you can only save &#4294967295;0.3 per board. When it comes the NXP M4 looks very nice with an M0 in one corner sharing the memory and peripherals and lots of new and innovative stuff. Obviously too much new stuff because they seem to be taking a very long time to get it out to market. Michael Kellett
On Jan 31, 2:24=A0am, "P_B" <patrik.bjorklund@n_o_s_p_a_m.rejlers.se>
wrote:
> Hi everyone! > > This is my first post here :) > > I'm interested in making a platform which has an 32-bit MCU as the centra=
l
> component. This platform is supposed to be configurable, i.e. do things > that an 8-bit MCU can handle OR do more advanced signal processing.
When you say platform, do you mean Chip, or Eval-Module level ? Many vendors make 32 bit micros, but the cost of entry-level boards varies quite a lot. eg On the ARM front ST and Nuvoton have low cost entry level boards; Microchip have one that comes with Multiple samples, covering more than one of their cores. Atmel have a AT32UC3L0-XPLD, for their AVR32.... - and new entrant Infineon have an M4 in release, with what looks (so far) to be good Software and Eval support. You might want to look at the peripherals on these variants, as they can vary more than the cores... -jg
>> >> --------------------------------------- >> Posted through http://www.EmbeddedRelated.com > >Currently ST have M3 and M4, NXP have M0 and M3, TI have M3 and R4 (you >probably don't want this) and Freescale have M4. There are others. > >The price difference between M0 and M3 in the same footprint is small >and using a low pin count will keep you from using the better M3 and M4s. >If I were you (and bear in mind I don't know the whole story) I would go >for M3/M4 from ST because you can get them now and if you ever hit >volumes where it is worth using M0 you can re-spin the board. (10k off >you can get an ST M3 for <&#65533;0.60, so even if the M0 is half the price >(and it won't be that cheap) you can only save &#65533;0.3 per board. > >When it comes the NXP M4 looks very nice with an M0 in one corner >sharing the memory and peripherals and lots of new and innovative stuff. >Obviously too much new stuff because they seem to be taking a very long >time to get it out to market. > >Michael Kellett >
Hi Michael! This gave me a lot to think about!! I like your arguments. An ST M3 definitely sounds like the best choice now. In addition, I just found out that 12-bit ADC is needed, so if I wanted to go for NXP, I would have to go for an external ADC or use an "overkill" M3. I have been using IAR compiler for alot of other MCU-families. So using it for ARM as well, seems like a natural step. Anyone have something bad to say about the IAR ARM compiler?! If yes, can you specify? /P --------------------------------------- Posted through http://www.EmbeddedRelated.com
>When you say platform, do you mean Chip, or Eval-Module level ? > >Many vendors make 32 bit micros, but the cost of entry-level boards >varies quite a lot. > >eg On the ARM front ST and Nuvoton have low cost entry level boards; >Microchip have one that comes with Multiple samples, covering more >than one of their cores. >Atmel have a AT32UC3L0-XPLD, for their AVR32.... >- and new entrant Infineon have an M4 in release, with what looks (so >far) to be good Software and Eval support. > >You might want to look at the peripherals on these variants, as they >can vary more than the cores... > >-jg >
Hi With platform I mean a platform which can be used for different projects that have a lot of things in common, but also can have individual things. This is for professional stuff, just to be clear. I just come to think about one criteria, an MCU-family, which currently does NOT have a huge errata-list. So, something that has been around for a couple of years is better, I think. --------------------------------------- Posted through http://www.EmbeddedRelated.com
Op Tue, 31 Jan 2012 09:26:57 +0100 schreef P_B  
<patrik.bjorklund@n_o_s_p_a_m.rejlers.se>:
> I have been using IAR compiler for alot of other MCU-families. So using > it for ARM as well, seems like a natural step. > Anyone have something bad to say about the IAR ARM compiler?! If yes, can > you specify?
Not all architectures that IAR support receive the amount of attention that you might expect, but I think it is quite likely that ARM is the architecture that receives most of their attention. -- Gemaakt met Opera's revolutionaire e-mailprogramma: http://www.opera.com/mail/ (Remove the obvious prefix to reply privately.)
"P_B" <patrik.bjorklund@n_o_s_p_a_m.rejlers.se> writes:

>>> >>> --------------------------------------- >>> Posted through http://www.EmbeddedRelated.com >> >>Currently ST have M3 and M4, NXP have M0 and M3, TI have M3 and R4 (you >>probably don't want this) and Freescale have M4. There are others. >> >>The price difference between M0 and M3 in the same footprint is small >>and using a low pin count will keep you from using the better M3 and M4s. >>If I were you (and bear in mind I don't know the whole story) I would go >>for M3/M4 from ST because you can get them now and if you ever hit >>volumes where it is worth using M0 you can re-spin the board. (10k off >>you can get an ST M3 for <&iuml;&iquest;&frac12;0.60, so even if the M0 is half the price >>(and it won't be that cheap) you can only save &iuml;&iquest;&frac12;0.3 per board. >> >>When it comes the NXP M4 looks very nice with an M0 in one corner >>sharing the memory and peripherals and lots of new and innovative stuff. >>Obviously too much new stuff because they seem to be taking a very long >>time to get it out to market. >> >>Michael Kellett >> > > Hi Michael! > > This gave me a lot to think about!! > I like your arguments. > An ST M3 definitely sounds like the best choice now. In addition, I just > found out that 12-bit ADC is needed, so if I wanted to go for NXP, I would > have to go for an external ADC or use an "overkill" M3.
Be aware that internal ADCs tend to be poor performers, in terms of noise and linearity, compared to external ones. However you have a lot of processing power available to you so may be able to compensate for this if important. For example, using signal averaging and/or linearization. And the STM32 ADCs are the most powerful I have seen in terms of flexibility and features.
> I have been using IAR compiler for alot of other MCU-families. So using it > for ARM as well, seems like a natural step. > Anyone have something bad to say about the IAR ARM compiler?! If yes, can > you specify?
It's not gcc :) -- John Devereux

The 2024 Embedded Online Conference