FPGA Central

Is it possible to perform an FIR filter on 8 bit digital data using
only the simplest tools? Like an 8bit MCU with very limited RAM? How
can I generate very low level code from an FIR filter so I can
translate it into assembler for the MCU I am using? I am thinking of
using the Atmel ATTINY13.

An FIR filter is just a sequency of steps, albiet lots of steps, and
even more steps if you cant do floating point in a native way, right?
So what are these steps?

[email protected] wrote:

> Is it possible to perform an FIR filter on 8 bit digital data using
> only the simplest tools? Like an 8bit MCU with very limited RAM? How
> can I generate very low level code from an FIR filter so I can
> translate it into assembler for the MCU I am using? I am thinking of
> using the Atmel ATTINY13.
>
> An FIR filter is just a sequency of steps, albiet lots of steps, and
> even more steps if you cant do floating point in a native way, right?
> So what are these steps?

An FIR filter can take many steps, but they fit into a small loop.
Fixed-point filters work, and they don't need floating point. (They do
need multiple precision.) Your arithmetic routines of any kind can (and
should) be written as callable functions, so their complexity needn't
add to the conceptual complexity of the filter.

A problem you might have aside from RAM size is execution speed,
especially if you choose to simulate floating-point arithmetic in
software. The inner workings of the filter are well explained in the
appropriate chapters at http://dspguide.com/pdfbook.htm. The author
gives examples in BASIC that can be run on a TRS-80.

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