"Jackson4" <p2@media.mit.edu> writes:

>>
>>STA333 from STMicro.
>
> Thank you Vlad, this looks like a good chip, but we need something more
> versatile, as some of the algorithms we're using are nonstandard. If it
> were only that easy...

Would a small FPGA be any good?  *Lots* of flexibility, and may be cheaper
than you think...

Cheers,
Martin

-- 
martin.j.thompson@trw.com 
TRW Conekt - Consultancy in Engineering, Knowledge and Technology
http://www.conekt.net/electronics.html

Jackson4 wrote:


> I can't go into too much detail, but the output PWM can't be straight
> class D, we there are other requirements and an unusual load.   There are
> also custom filters that the project needs, and other fairly simple, though
> unique, intermediary control/signal tailoring steps.

We develop the special PWM amplifiers. If this is a business project, 
consider my services. The contact is at the web site below.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

>Could you explain what is special and what exactly are you planning to 
>accomplish?
>
>Besides, there is a lot of specialized programmable or semi-programmable

>solutions for the DSP PWM amplifiers from TI, NXP, FreeScale. Using a 
>general purpose DSP for that is an overkill; it is very cost
inefficient.

I can't go into too much detail, but the output PWM can't be straight
class D, we there are other requirements and an unusual load.  There are
also custom filters that the project needs, and other fairly simple, though
unique, intermediary control/signal tailoring steps.

I'll look more at these programmable PWM options, but they don't seem to
allow enough algorithm flexibility.

Jackson4 wrote:

>>STA333 from STMicro.
> 
> 
> Thank you Vlad, this looks like a good chip, but we need something more
> versatile, as some of the algorithms we're using are nonstandard. If it
> were only that easy...

Could you explain what is special and what exactly are you planning to 
accomplish?

Besides, there is a lot of specialized programmable or semi-programmable 
solutions for the DSP PWM amplifiers from TI, NXP, FreeScale. Using a 
general purpose DSP for that is an overkill; it is very cost inefficient.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

>
>STA333 from STMicro.

Thank you Vlad, this looks like a good chip, but we need something more
versatile, as some of the algorithms we're using are nonstandard. If it
were only that easy...

Jackson4 wrote:
> Hello all,
> 
> My group is designing a small, inexpensive audio device (basically a PWM
> amp with variable/custom EQ processing), and we're looking to select a DSP.


STA333 from STMicro.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

 I don't believe it has any
>specific hardware support for DSP (MAC instructions, dual operand
>prefetches, etc). You may find that despite the ~120MIPS max they
>quote that this lack will reduce its performance for basic DSP
>algorithms over processors that do include these features but which
>have lower max instruction rates. Whether or not this is an issue
>depends on your application of course.

The PIC32 does have single-cycle MACs, which makes it a possible
candidate.  Not sure if it's just 16x16 or 32x32, but it brings it close to
the TI C2000 series, which is what I'm now leaning toward.

On Jun 1, 6:39 pm, "Jackson4" <p...@media.mit.edu> wrote:
> >You might check out some of the low-end Freescale 56k processors -
> >there are some very economical parts in that family these days.
> >They're 24-bit fixed point and have sufficient dynamic range for good
> >audio, but may not have enough general purpose peripherals & features
> >for all the other things you want to do.
>
> Ah, I have looked into the DSP56F8xx .  They do look interesting, but
> unless someone talks me out of it, I might go straight into 32-bit and make
> my life easier, since the PIC32 and TMS320 seems to be priced similarly.

PIC32 is basically a MIPS4 architecture which is targeted at general-
purpose MCU applications built with high-level language and possibly
relying on an OS (RTOS or Linux-like). I don't believe it has any
specific hardware support for DSP (MAC instructions, dual operand
prefetches, etc). You may find that despite the ~120MIPS max they
quote that this lack will reduce its performance for basic DSP
algorithms over processors that do include these features but which
have lower max instruction rates. Whether or not this is an issue
depends on your application of course.

> What I'd really love is a floating point processor with onboard flash, for
> $5-$10. :)

Wouldn't we all?

Eric

>Be careful "TMS320" is the prefix for many very different TI processors.

Right.  I guess the family is better classified as the C2000.

Jackson4 wrote:
>> dsPIC33 would probably be a good match for what you're doing. I've
>> used it for simple audio applications, but beware that the 16-bit
>> wordsize really limits the ultimate audio quality of what you can do
>> with it these processors.
> 
> Yes, I wrote some simple routines (filters) using an eval board, and this
> remains a concern.  24 or 32 bit might be a better way.
> 
>> You might check out some of the low-end Freescale 56k processors -
>> there are some very economical parts in that family these days.
>> They're 24-bit fixed point and have sufficient dynamic range for good
>> audio, but may not have enough general purpose peripherals & features
>> for all the other things you want to do.
> 
> Ah, I have looked into the DSP56F8xx .  They do look interesting, but
> unless someone talks me out of it, I might go straight into 32-bit and make
> my life easier, since the PIC32 and TMS320 seems to be priced similarly.
> 
> What I'd really love is a floating point processor with onboard flash, for
> $5-$10. :)

Be careful "TMS320" is the prefix for many very different TI processors.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������