- Opinions about AVR32 for Voice Recorder?
- Posted by Ty Roberts on December 28th, 2007
I am beginning the design of a specialized voice recorder device. It
needs to do decent quality voice recording (and play back) and it has
a graphic LCD display that shows certain information about the context
of the recording and shows simple animations. It also needs USB and
memory card I/O capabilities.
I've been researching which microcontrollers might be suited to this
purpose. I'm impressed by the relatively new AVR32 line. These seem
to be inexpensive, low power, and (depending on the model) integrate
a good deal of functionality, like 16bit ADC, USB controller, LCD
controller, SmartMedia, even audio amp, etc. Also, free development
tools are distributed by AVR that run on non-Windows platforms.
This all seems like a win. Am I missing something? Does anyone have
a criticism of the AVR32, or a recommendation for a good alternative?
- Posted by larwe on December 28th, 2007
On Dec 28, 4:02*am, Ty Roberts <ty...@aool.com> wrote:
Yes; it's a sole-source core with little developer interest compared
to the more mainstream offerings. Anything you can get in an AVR32 you
can also get with a mainstream core like ARM or MIPS, and probably
cheaper.
- Posted by Ty Roberts on December 31st, 2007
larwe wrote:
Could you suggest a good resource that gives an overview of the
ARM core market? I'm having trouble finding a comparable ARM based mcu
with USB, 16 bit ADC/DAC, LCD, etc. I don't doubt it's out there
somewhere, but there are so many vendors and products that it's hard
to find your way.
- Posted by John Devereux on December 31st, 2007
Ty Roberts <tyrob@aool.com> writes:
<http://www.gnuarm.com>,
in particular
<http://www.gnuarm.com/ArmDevices_frame.html>.
However I am not aware of anything wih that exact feature mix. AFAIK
16 bit ADC and DAC is non-existant on ARM. Even the few 12 bit capable
devices are smaller devices, at the other end of the scale from
devices with integrated USB and LCD.
Unless the 16 bit ADC/DAC is audio, in which case perhaps there is
something media orientated (like i.mx? not looked.)
Your best bet is probably to have external ADC/DAC.
--
John Devereux
- Posted by Chris H on December 31st, 2007
In message <s9SdncyIToT-ROXanZ2dnUVZ_g6dnZ2d@comcast.com>, Ty Roberts
<tyrob@aool.com> writes
Try analog devices. They usually do good ADC/DAC
Also ST Micros have a reputation for lots of IO on ARM parts.
That said..... almost any part (including MIPS) is going to be less
popular than "ARM" per say. However you want an ARM core with a 16bit
ADC/DAC so you will have to pick a single source implementation. That
implementation may go away if it does not sell well. For example the
Sharp ARM7 cores with the graphics controller. Though in that case NXP
took over the parts.
The AVR32 is not likely to go away any time soon as Atmel have pinned a
lot on it. Their other AVR's are very popular despite being single
source. So I would say there is as much risk going with an Atmel AVR32
and any other chip implementation.
--
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\
\/\/\/\/\ Chris Hills Staffs England /\/\/\/\/
/\/\/ chris@phaedsys.org www.phaedsys.org \/\/\
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
- Posted by steve on January 2nd, 2008
On Dec 31 2007, 3:58*am, Ty Roberts <ty...@aool.com> wrote:
Personally, I wouldn't worry about not using an ARM chip, there are
zero multi sourced ARM chips anyway (if whatever ARM chip you pick
goes obsolete your screwed just like if the AVR32 goes obsolete). Pick
whatever processor has the right combination of peripherals.
- Posted by rickman on January 2nd, 2008
On Dec 31 2007, 7:55 am, John Devereux <jdREM...@THISdevereux.me.uk>
wrote:
This is my web page and I am glad that people find it useful.
However, I have to apologize for not keeping it up to date. I have
intended to update it for some time now, but there are so many new
devices and my time seems to always be so limited. In particular,
there are always new devices from Atmel and Philips as well as some
significant new devices from Luminary Micro. I also am aware of new
Cortex M3 parts (STM32) from ST Micro which I have not even had a
chance to really read up on. So please don't consider the above page
to be complete in any way.
A 16 bit ADC on an MCU is likely not going to be a very good 16 bit
ADC. If you really need anything nearly that accurate, you should
consider a separate device. You will find that for the most part, a
16 bit ADC will run as much as the MCU... 12 bit ADCs, on the other
hand, are a dime a dozen (although not literally).
Steve below makes a valid point. Although there is some commonality
within the ARM family that your code can take advantage of, the
peripherals are largely different between manufacturers. So it will
not be easy to port application code between manufacturers. However,
there are other concerns about using a proprietary device such as the
AVR32. The main one is tools. Although Atmel may have supported tool
vendors to develop compilers and debuggers for this devices, the tool
vendors will sell a lot more of the ARM tool sets and it is very
unlikely that the AVR32 tools will work as well as the ARM tools.
So if you have the resources to spend some extra time with
development, you might get a better match between your application and
the AVR32 devices lowering your unit costs. But if you need to
optimize your development time and cost, you might want to stay with a
more mainstream family.
- Posted by vinnie on January 3rd, 2008
I'd suggest looking at the Renesas SH2A series, in particular the 7263
which is designed for embedded audio and LCD control.
http://america.renesas.com/fmwk.jsp?...Specifications
The Hitach (Renesas) SH core has been around for 12+ years, unlike th
AVR32, and there are dozens of variations of the architechture. All ar
supported by multiple tools vendors like IAR and GCC. Check ou
www.kpitgnutools.com.
I also think an external A/D chip is a requirement if you need more tha
10 bits (if you don't believe me, do a calculation on the noise "floor
for a system running at 3.3 volts and compare it to the resolution of a 1
or 16 bit A/D.)
- Posted by Ulf Samuelsson on January 3rd, 2008
There is a *free* gcc compiler suite, which is supported by Atmel.
This is integrated with the *free* Eclipse based AVR32 Studio.
Linux is free and can be easily built using buildroot
(www.atmel.no/buildroot).
This will also build the U-boot monitor.
You can also get the IAR Embedded Workbench for AVR32.
The AVR32 Gateway is a real low cost board (< $100 single qty)
capable of running Linux (16 MB Flash and 32 MB SDRAM)
which can be used as is, in many applications.
If you happen to have used the JTAGICE Mk II for the AVR, then the
same unit will work with the AVR32, otherwise this is about $300.
--
Best Regards,
Ulf Samuelsson
This is intended to be my personal opinion which may,
or may not be shared by my employer Atmel Nordic AB
- Posted by Chris H on January 4th, 2008
In message
<8f45c81c-088d-444e-b759-245e2bd4da52@u10g2000prn.googlegroups.com>,
rickman <gnuarm@gmail.com> writes
ALL MCU are proprietary except the 8051 where there are multiple cores
(over 40) some of which are FREE, and multiple implementations over 600
from about 60+ chip vendors.. The ARM core is single source and
definitely NOT free.
I think you will find that the IAR compiler will be at a similar
standard for both ARM and AVR32.
The Gcc is a movable feast and depending who's implementation (and
library) you get is not going to be constant on any architecture let
alone across architectures. However I would think that it is going to be
as good on AVR as ARM.
Maybe but the same could be said for the AVR family in general. Or for
that matter the PIC parts which are VERY non-standard and proprietary.
They seem to do very well.
The AVR32 is still worth a look as it does have support from main
stream tools. I am sure you will get a lot of support from Atmel.
--
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\
\/\/\/\/\ Chris Hills Staffs England /\/\/\/\/
/\/\/ chris@phaedsys.org www.phaedsys.org \/\/\
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
- Posted by Wilco Dijkstra on January 4th, 2008
"Chris H" <chris@phaedsys.org> wrote in message news:v54BiHFGuffHFAZD@phaedsys.demon.co.uk...
I wouldn't call ARM cores single source as they are produced by many
different fabs but there are also several architecture licensees who
produce their own implementations. The only thing that is single source
is the ARM architecture, and that's exactly how you want it (or you get the
mess the x86 "architecture" is in today).
Of course it's rare to see MCUs that have the exact same peripherals,
are pin compatible and run at exact the the same frequency, voltage and
power - this simply doesn't make sense commercially. I'm sure few 8051
implementations fit these criteria.
Ie. code quality similar to GCC...
GCC is far behind the state of the art on many architectures.
Wilco
- Posted by David Kelly on January 4th, 2008
Wilco Dijkstra wrote:
So? What does the fact gcc on HC11 sucks have to do with gcc on AVR32 or
ARM? The 8-bit avr-gcc is exceptionally good.
What gcc needs to excel on a platform is a motivated sponsor. Atmel
hired the developer of WinAVR, presumably to ensure that he has time to
keep WinAVR current. I don't know but would think its a safe bet the way
Atmel is backing Eclipse and gcc for the AVR32 that many full time
employees are working there too.
- Posted by Wilco Dijkstra on January 4th, 2008
"David Kelly" <n4hhe@Yahoo.com> wrote in message news:13nsk0v5f8de652@corp.supernews.com...
As I'm sure you know, compilers share a lot of code between targets.
One of GCC's weakest points is the register allocator, and this affects
all targets (though not equally of course).
The latest GCC for ARM is pretty bad by any standard - the Thumb code
it generates is often larger than ARM code from good ARM compilers!
I haven't looked at AVR32 but since the instruction set is very similar to
Thumb-2, there is no reason to believe it is much better than the ARM port.
True, it takes a lot of time and effort to create a good compiler. But you
can avoid the fact that most GCC development is geared towards x86
and Lunix rather than small embedded CPUs. Think of the libraries for
example.
Wilco
- Posted by Chris H on January 4th, 2008
In message <I0qfj.27090$KC3.4908@newsfe6-gui.ntli.net>, Wilco Dijkstra
<Wilco_dot_Dijkstra@ntlworld.com> writes
So who apart from ARM does an ARM core? (One that is not licensed by
ARM)
I think most 8051 vendors have a basic part that is pin for pi with
everyone else's but most are not pin compatible
I doubt it.
And for once it was not me saying that.
--
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\
\/\/\/\/\ Chris Hills Staffs England /\/\/\/\/
/\/\/ chris@phaedsys.org www.phaedsys.org \/\/\
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
- Posted by Grant Edwards on January 4th, 2008
On 2008-01-04, Chris H <chris@phaedsys.org> wrote:
Didn't DEC/Intel do an ARM core? It used to be called
StrongARM and now it's called XScale.
--
Grant Edwards grante Yow! It's the RINSE CYCLE!!
at They've ALL IGNORED the
visi.com RINSE CYCLE!!
- Posted by larwe on January 4th, 2008
On Jan 4, 12:06*pm, Grant Edwards <gra...@visi.com> wrote:
DEC did it, Intel scored it in the buyout, and it was dumped as Intel
exited the embedded market; Marvell Technology Group owns it now.
In any case, Chris is arguing disingenuously, as usual; I wouldn't pay
much attention.
- Posted by Chris H on January 4th, 2008
In message <13nsprrc1ult7d8@corp.supernews.com>, Grant Edwards
<grante@visi.com> writes
I think you will find that it was a licensed ARM core in them,
Somewhere there was a Black ARM which was a non-ARM ARM core clone. I
heard something of it about 4 years ago (It was confirmed in
conversation with someone from ARM) However I have not looked for it
since.
--
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\
\/\/\/\/\ Chris Hills Staffs England /\/\/\/\/
/\/\/ chris@phaedsys.org www.phaedsys.org \/\/\
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
- Posted by Chris H on January 4th, 2008
In message
<e8e9ab9a-786e-491a-970b-7274869bf4d8@e10g2000prf.googlegroups.com>,
larwe <zwsdotcom@gmail.com> writes
Was it based on an ARM(tm) core or something that was nto connected to
ARM at all?
How so?
--
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\
\/\/\/\/\ Chris Hills Staffs England /\/\/\/\/
/\/\/ chris@phaedsys.org www.phaedsys.org \/\/\
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
- Posted by Wilco Dijkstra on January 4th, 2008
"larwe" <zwsdotcom@gmail.com> wrote in message news:e8e9ab9a-786e-491a-970b-7274869bf4d8@e10g2000prf.googlegroups.com...
On Jan 4, 12:06 pm, Grant Edwards <gra...@visi.com> wrote:
StrongARM and XScale are different cores. Intel still sells the networking
XScales. Marvell and Qualcomm also design their own set of high-end
ARM cores. There are others that take an existing ARM core and optimise
it for high performance, for example Samsung makes the high speed ARM11
used in the iPhone.
Wilco
- Posted by Wilco Dijkstra on January 4th, 2008
"Chris H" <chris@phaedsys.org> wrote in message news:0zCnAxPtYnfHFAbB@phaedsys.demon.co.uk...
No, definitely not. These were not licensed ARM cores. They were
new designs based on an ARM architecture. There are many ARM
architecture licensees who make their own cores (see my other post).
Illegal cloning is something different altogether...
Wilco