Which starter kit?

"steveu" <31473@dsprelated> wrote in message 
news:QaudnfUGEO482uzNnZ2dnUVZ_qmdnZ2d@giganews.com...
>>>
> I thought Vlads comments were unusually helpful. I do a lot of DSP with an
> MSP430, but I deal with a lot of *very* slow signals.
> The STM32 lets you go a bit faster.
> You can do some simple audio processing with it, and that's about it.

My favorite processor for low end DSP work is AVR Mega. It is capable of 
more complicated things besides usual control, modem or sensor stuff.  I've 
implemented DVOR ILS, sound recognition, utility locator and many other 
cost/power sensitive things with it.

Vladimir Vassilevsky
DSP and Mixed Signal Consultant
www.abvolt.com

"Waves" <61140@dsprelated> writes:

> Hello,
>
> I'm looking to get into DSP/embedded programming practically.

Hi Dude,

The problem is, even that relatively narrow field has a wide
variety of potential platforms. 

> I was looking at these low cost development kits:
>
> TI MSP430 USB Stick Development Tool
> http://www.ti.com/tool/ez430-f2013#119
>
> STMicro STM32VL DISCOVERY
> http://www.st.com/internet/evalboard/product/250863.jsp
>
> Am I right in thinking that the above two are more for embedded programming
> development rather than DSP?

Don't know about the STMicro, but the MSP430 is a relatively slow
processor intended for very low power applications.

You can do DSP on ANY processor, no matter what type, but the
complexity, sample rate at which you can operate, and pain to develop
your algorithm will vary greatly.

If I were you, I would separate "DSP programming" from "embedded
programming." A fine platform for learning how to program DSP is your
own PC using a basic language like C or C++ [1]. It is much easier to
develop and test DSP algorithms on the PC. Once you've done a couple of
basic DSP programming projects there, you can move to an embedded
implementation fairly easily, assuming you are familiar with embedded
programming. 

Embedded programming entails a largely-orthogonal set of skills (to DSP)
which can include learning how to write interrupt service routines, how
to write device drivers, how to use an embedded OS (e.g., SYSBIOS or
DSPBIOS from TI), how to arrange linked sections in memory, etc.

You could also split "embedded programming" into "high-level embedded
programming" (e.g., embedded linux systems) and "low-level embedded
programming" (e.g., something like the MSP430), so it would help if you
had an idea which one you were more interested in.

--Randy

[1] I would use a linux distro (Fedora 17 recommended), the gcc
compiler, and gnumake as a build system. I would then use a .wav file
I/O library and take input from a .wav file and write your results to a
.wav file. You can generate/examine .wav files, plot results, etc.,
using gnu-octave (or Matlab).
-- 
Randy Yates
Digital Signal Labs
http://www.digitalsignallabs.com

On Sat, 06 Oct 2012 23:38:36 -0400, robert bristow-johnson
<rbj@audioimagination.com> wrote:

>
>
>gee, i thought hams had some DIY standard thingie.  they used to use the 
>Mot 56K just like audio geeks did.  dunno what hams use now.
>
>(long ago, like the '60s, i used to be WB0CCA.  but of course never had 
>ham and DSP on the same plate.)
OFF TOPIC AMATEUR RADIIO COMMERCIAL FOLLOWS,

The world has changed several times since the 60's. There are many
facets to Amateur Radio. Advancing technology has solved many old
problems and many active hams go no further with technology than what
is necessary to pass the tests. I view that as a good thing because
there has been extensive gains in the matter of implementation of
existing technology. DSP support and other computer innovations have
made it possible for many users to share what used to be considered
one voice channel. Digital communications now work around the world
and into space with signal levels so low that they are nearly
undetectable without prior knowledge of where, when, and how
information. Merging radio & Internet resources allow collaboration
between developers that were never practical except for governments
and big business.  There are throw away transceivers available on VHF
(less than $100 and impractical to repair) and similar bargains for HF
equipment.  There are more amateur radio licenses now than there has
ever been. AND there is an interesting niche (or more) for every one. 
Since Morse code is no longer required (International treaty) it is
gaining popularity. 
Most new HF transceivers use some form of DSP now. 

If one has any interest at all, the tests are pretty simple (that does
not mean you won't have to at least read the book!) except for the top
class (Extra Class in the US). That one will humiliate anyone brazen
enough to try to bluff their way through. The test fees are low and
most population centers have frequent test sessions.

de John, W8CCW
John Ferrell W8CCW

On Sun, 7 Oct 2012 10:12:56 -0500, "Vladimir Vassilevsky"
<nospam@nowhere.com> wrote:

>
>"steveu" <31473@dsprelated> wrote in message 
>news:QaudnfUGEO482uzNnZ2dnUVZ_qmdnZ2d@giganews.com...
>>>>
>> I thought Vlads comments were unusually helpful. I do a lot of DSP with an
>> MSP430, but I deal with a lot of *very* slow signals.
>> The STM32 lets you go a bit faster.
>> You can do some simple audio processing with it, and that's about it.
>
>My favorite processor for low end DSP work is AVR Mega. It is capable of 
>more complicated things besides usual control, modem or sensor stuff.  I've 
>implemented DVOR ILS, sound recognition, utility locator and many other 
>cost/power sensitive things with it.
>
>Vladimir Vassilevsky
>DSP and Mixed Signal Consultant
>www.abvolt.com
>
I am more familiar with PIC micro's and I am just getting started with
some TI products. I do have a few Arduino boards & Chips but so far I
have not run across one that has more than a 10 bit A/D and none with
a built in DAC. 

It would help a lot to see some simple examples of DSP math
implemented in silicon with a popular programming language.  It seems
like getting the tool chain set up and working is harder than writing
code. 

OTH, may be I am still lacking enough to ask a reasonable question. 

John Ferrell W8CCW
John Ferrell W8CCW

"John Ferrell" <W8CCW@arrl.net> wrote in message 
news:r8i378htcc1shesc22ipr8r6pb6log14a4@4ax.com...
> On Sun, 7 Oct 2012 10:12:56 -0500, "Vladimir Vassilevsky"
> <nospam@nowhere.com> wrote:

>>My favorite processor for low end DSP work is AVR Mega. It is capable of
>>more complicated things besides usual control, modem or sensor stuff. 
>>I've
>>implemented DVOR ILS, sound recognition, utility locator and many other
>>cost/power sensitive things with it.
>>
> I am more familiar with PIC micro's

For number crunching tasks, PIC18 at 40 MHz is about 3 times slower then AVR 
Mega at 20 MHz.
There are PIC32s and dsPICs, of course; however it is a different class.

> some TI products. I do have a few Arduino boards & Chips but so far I
> have not run across one that has more than a 10 bit A/D and none with
> a built in DAC.

The built-in 10 Bit ADC is often sufficient; PWM would work as DAC. If 
better performance is required, you will need external DACs or ADCs anyway.

> It would help a lot to see some simple examples of DSP math
> implemented in silicon with a popular programming language.

What do you mean by DSP math?

> It seems
> like getting the tool chain set up and working is harder than writing 
> code.

Indeed. Set up of the tool chain and development of application framework 
and low level hardware drivers could take more time and effort then actual 
DSP part.

Vladimir Vassilevsky
DSP and Mixed Signal Consultant
www.abvolt.com

On 10/7/12 1:50 PM, John Ferrell wrote:
> On Sat, 06 Oct 2012 23:38:36 -0400, robert bristow-johnson
> <rbj@audioimagination.com>  wrote:
>
>>
>>
>> gee, i thought hams had some DIY standard thingie.  they used to use the
>> Mot 56K just like audio geeks did.  dunno what hams use now.
>>
>> (long ago, like the '60s, i used to be WB0CCA.  but of course never had
>> ham and DSP on the same plate.)


> OFF TOPIC AMATEUR RADIO COMMERCIAL FOLLOWS,
>
> The world has changed several times since the 60's. There are many
> facets to Amateur Radio. Advancing technology has solved many old
> problems and many active hams go no further with technology than what
> is necessary to pass the tests. I view that as a good thing because
> there has been extensive gains in the matter of implementation of
> existing technology. DSP support and other computer innovations have
> made it possible for many users to share what used to be considered
> one voice channel. Digital communications now work around the world
> and into space with signal levels so low that they are nearly
> undetectable without prior knowledge of where, when, and how
> information. Merging radio & Internet resources allow collaboration
> between developers that were never practical except for governments
> and big business.  There are throw away transceivers available on VHF
> (less than $100 and impractical to repair) and similar bargains for HF
> equipment.  There are more amateur radio licenses now than there has
> ever been. AND there is an interesting niche (or more) for every one.
> Since Morse code is no longer required (International treaty)

i didn't knew that.  no code test at all anymore?

> it is gaining popularity.
> Most new HF transceivers use some form of DSP now.
>
> If one has any interest at all, the tests are pretty simple (that does
> not mean you won't have to at least read the book!) except for the top
> class (Extra Class in the US).

used to be 20 w.p.m.  that's no longer the case?

i *barely* could cut the 13 w.p.m. when i got my "Conditional Class" 
license (same as General, but was administered by a ham rather than the 
FCC, like Novice and Technician Class).  never got higher.  in fact, i 
didn't even renew the Conditional Class license in 1974 when it expired. 
  i sold my HW-100 and everything else except the Heathkit VTVM for 
about $300 in 1975.

> That one will humiliate anyone brazen
> enough to try to bluff their way through. The test fees are low and
> most population centers have frequent test sessions.
>
> de John, W8CCW
> John Ferrell W8CCW


-- 

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."

On Sun, 07 Oct 2012 17:53:47 -0400, robert bristow-johnson
<rbj@audioimagination.com> wrote:

>On 10/7/12 1:50 PM, John Ferrell wrote:
>> On Sat, 06 Oct 2012 23:38:36 -0400, robert bristow-johnson
>> <rbj@audioimagination.com>  wrote:
>>
{snip}
>> Since Morse code is no longer required (International treaty)
>
>i didn't knew that.  no code test at all anymore?
>
>> it is gaining popularity.
>> Most new HF transceivers use some form of DSP now.
>>
>> If one has any interest at all, the tests are pretty simple (that does
>> not mean you won't have to at least read the book!) except for the top
>> class (Extra Class in the US).
>
>used to be 20 w.p.m.  that's no longer the case?

No code for any class!
The best I ever got was 15 WPM. Age has not helped any and I am
currently at 9 wpm IF the code is also readable by machine. I simply
cannot copy a bug at any speed.

>i *barely* could cut the 13 w.p.m. when i got my "Conditional Class" 
>license (same as General, but was administered by a ham rather than the 
>FCC, like Novice and Technician Class).  never got higher.  in fact, i 
>didn't even renew the Conditional Class license in 1974 when it expired. 
>  i sold my HW-100 and everything else except the Heathkit VTVM for 
>about $300 in 1975.
>
>> That one will humiliate anyone brazen
>> enough to try to bluff their way through. The test fees are low and
>> most population centers have frequent test sessions.
>>
>> de John, W8CCW
>> John Ferrell W8CCW
John Ferrell W8CCW

On Sun, 07 Oct 2012 13:31:41 -0400, Randy Yates
<yates@digitalsignallabs.com> wrote:

>"Waves" <61140@dsprelated> writes:
>
>> Hello,
>>
>> I'm looking to get into DSP/embedded programming practically.
>
>Hi Dude,
>
>The problem is, even that relatively narrow field has a wide
>variety of potential platforms. 
>
>> I was looking at these low cost development kits:
>>
>> TI MSP430 USB Stick Development Tool
>> http://www.ti.com/tool/ez430-f2013#119
>>
>> STMicro STM32VL DISCOVERY
>> http://www.st.com/internet/evalboard/product/250863.jsp
>>
>> Am I right in thinking that the above two are more for embedded programming
>> development rather than DSP?
>
>Don't know about the STMicro, but the MSP430 is a relatively slow
>processor intended for very low power applications.
>
>You can do DSP on ANY processor, no matter what type, but the
>complexity, sample rate at which you can operate, and pain to develop
>your algorithm will vary greatly.
>
>If I were you, I would separate "DSP programming" from "embedded
>programming." A fine platform for learning how to program DSP is your
>own PC using a basic language like C or C++ [1]. It is much easier to
>develop and test DSP algorithms on the PC. Once you've done a couple of
>basic DSP programming projects there, you can move to an embedded
>implementation fairly easily, assuming you are familiar with embedded
>programming. 
>
>Embedded programming entails a largely-orthogonal set of skills (to DSP)
>which can include learning how to write interrupt service routines, how
>to write device drivers, how to use an embedded OS (e.g., SYSBIOS or
>DSPBIOS from TI), how to arrange linked sections in memory, etc.
>
>You could also split "embedded programming" into "high-level embedded
>programming" (e.g., embedded linux systems) and "low-level embedded
>programming" (e.g., something like the MSP430), so it would help if you
>had an idea which one you were more interested in.
>
>--Randy
>
>[1] I would use a linux distro (Fedora 17 recommended), the gcc
>compiler, and gnumake as a build system. I would then use a .wav file
>I/O library and take input from a .wav file and write your results to a
>.wav file. You can generate/examine .wav files, plot results, etc.,
>using gnu-octave (or Matlab).
I get hung up in windows with sound card programming. I will look into
wave files.

Each attempt into Linux seems to lead into forks of other things to
learn.

Matlab is priced well out of most Senior Citizens budgets and Sci Lab
is another fork that leads off topic...

Actually, embedded is easy to use for me today.. I miss Turbo Pascal!

Let me choose a fresh direction and come back.
John Ferrell W8CCW

John Ferrell <W8CCW@arrl.net> writes:
> [...]
> I get hung up in windows with sound card programming. I will look into
> wave files.

Hi John,

I have a simple wave file library I've developed and used over the years
that works well for simple linear PCM .wavs. You are welcome to it -
email me if you're interested.

> Each attempt into Linux seems to lead into forks of other things to
> learn.

Yes, pity, eh? :)

There is a solution that is the best (worst?) of both worlds: cygwin
under Windows (xp/7/what-have-you). Check it out: www.cygwin.com

> Matlab is priced well out of most Senior Citizens budgets and Sci Lab
> is another fork that leads off topic...

If you are going to do any amount of DSP, Matlab would be a tool well
worth learning. GNU-Octave is a) free and b) a clone in many ways of
Matlab. And it's available under cygwin!

> Actually, embedded is easy to use for me today.. I miss Turbo Pascal!

I guarantee it will be harder than a PC-based environment. It's a
question of DSP or DSP + embedded headaches ("let's see, am I
overrunning the stack?", "is my ADC ISR working right?", ...).

> Let me choose a fresh direction and come back. John Ferrell W8CCW

Of course.
-- 
Randy Yates
Digital Signal Labs
http://www.digitalsignallabs.com

On Sun, 7 Oct 2012 15:32:01 -0500, "Vladimir Vassilevsky"
<nospam@nowhere.com> wrote:

>
>"John Ferrell" <W8CCW@arrl.net> wrote in message 
>news:r8i378htcc1shesc22ipr8r6pb6log14a4@4ax.com...
>> On Sun, 7 Oct 2012 10:12:56 -0500, "Vladimir Vassilevsky"
>> <nospam@nowhere.com> wrote:
>
>>>My favorite processor for low end DSP work is AVR Mega. It is capable of
>>>more complicated things besides usual control, modem or sensor stuff. 
>>>I've
>>>implemented DVOR ILS, sound recognition, utility locator and many other
>>>cost/power sensitive things with it.
>>>
>> I am more familiar with PIC micro's
>
>For number crunching tasks, PIC18 at 40 MHz is about 3 times slower then AVR 
>Mega at 20 MHz.
>There are PIC32s and dsPICs, of course; however it is a different class.
>
OK, I will focus on the AVR Mega. I think I put a few away this
afternoon.
>> some TI products. I do have a few Arduino boards & Chips but so far I
>> have not run across one that has more than a 10 bit A/D and none with
>> a built in DAC.
>
>The built-in 10 Bit ADC is often sufficient; PWM would work as DAC. If 
>better performance is required, you will need external DACs or ADCs anyway.
I have not used PWM to work as a DAC before. I will check it out. 

How about this:
Feed nominal sine wave (400 Hz or so) into AD
Drive PWM with Value from AD. 
? may need to scale 10 bit AD value to 8 bit pwm
Compare input to output with scope and audio.

Does that sound like a starting point?

>> It would help a lot to see some simple examples of DSP math
>> implemented in silicon with a popular programming language.
>
>What do you mean by DSP math?

My first objective is to decode/encode the common forms of modulation:
FM, AM, SSB.

>
>> It seems
>> like getting the tool chain set up and working is harder than writing 
>> code.
>
>Indeed. Set up of the tool chain and development of application framework 
>and low level hardware drivers could take more time and effort then actual 
>DSP part.
>
The popular Arduino package may be adequate for the initial testing.

>Vladimir Vassilevsky
>DSP and Mixed Signal Consultant
>www.abvolt.com
>
de John, W8CCW
John Ferrell W8CCW