FPGA Central

Hello all,

I am on the verge of implementing the DSP Trick "Complex downconverter
for signals at Fs/4 and 3Fs/4", from Allan Herriman, posted in the trick
sections of dspguru.com.
Here is the link to the trick:
http://www.dspguru.com/comp.dsp/tricks/alg/cpxdown.htm

To do this, I will be using a simple FPGA.

My question is about the final notes of the trick, and i quote "an
frequency errors in the input signal will be present during filtering".

I would like to know what Allan meant when he wrote this. Basically, i
there is a variation in frequency of the input signal, will this variatio
still be present in the downconverted signal? I this is correct, i guess
just have to make a FIR filter which will not cut out this variation.

Has anybody used this trick before? Any do's and don'ts about it?

Thanks for your help!

NiCad

in article [email protected], NiCad at
[email protected] wrote on 06/30/2006 10:50:

>
> I am on the verge of implementing the DSP Trick "Complex downconverters
> for signals at Fs/4 and 3Fs/4", from Allan Herriman, posted in the tricks
> sections of dspguru.com.
> Here is the link to the trick:
> http://www.dspguru.com/comp.dsp/tricks/alg/cpxdown.htm
>
> To do this, I will be using a simple FPGA.

i think that is what was the intended target. multiplying by -1, 0 or +1
should be much easier in an FPGA than by some general number.

> My question is about the final notes of the trick, and i quote "any
> frequency errors in the input signal will be present during filtering".
>
> I would like to know what Allan meant when he wrote this. Basically, if
> there is a variation in frequency of the input signal, will this variation
> still be present in the downconverted signal?

yes, there will be a linear (w.r.t. time) phase change in the downconverted
IQ pair if the IF frequency is not Fs/4. there is no mathematical
difference between the IF that is spot on some given frequency but has for
its data a linear increasing angle and another IF that is not changing its
IQ data, but is a little off of the demodulator frequency. now i do not
know where you are getting your IF, but maybe if you can do some kinda
phase-locked synchronous sampling, you can keep it precisely on Fs/4. if
not, then expect I + j*Q to move around the circle at a speed proportional
to the difference of your IF to Fs/4. you could also track that, after the
fact, and mathematically compensate for it (but how would you know the
difference between this and the case where the IF is on the spot but the IQ
data is supposed to be moving around a circle?


> I this is correct, i guess i
> just have to make a FIR filter which will not cut out this variation.

how does one use an FIR filter (or any kind of LTI filter) to adjust the
frequency of a quadrature signal?

> Has anybody used this trick before? Any do's and don'ts about it?

i haven't, but it looks straight-forward enough.

--

r b-j [email protected]

"Imagination is more important than knowledge."

"NiCad" <[email protected]> wrote in message
news:[email protected] ...
>
> Hello all,
>
> I am on the verge of implementing the DSP Trick "Complex downconverters
> for signals at Fs/4 and 3Fs/4", from Allan Herriman, posted in the tricks
> sections of dspguru.com.
> Here is the link to the trick:
> http://www.dspguru.com/comp.dsp/tricks/alg/cpxdown.htm
>
> To do this, I will be using a simple FPGA.
>
> My question is about the final notes of the trick, and i quote "any
> frequency errors in the input signal will be present during filtering".
>
> I would like to know what Allan meant when he wrote this. Basically, if
> there is a variation in frequency of the input signal, will this variation
> still be present in the downconverted signal? I this is correct, i guess i
> just have to make a FIR filter which will not cut out this variation.
>
> Has anybody used this trick before? Any do's and don'ts about it?
>
> Thanks for your help!
>
> NiCad

If you are following the instructions, you should be fine. I've used this
before on several occasions.

Regarding the final note, your understanding is correct. If your IF is off
frequency, your downconverted signal will still have this delta offset. If
you are doing demodulation, it'll contribute to frequency error. I'd expect
these deltas to be pretty small, so your FIR filter shouldn't really need
any special considerations - your software would need to be able to
accomodate this though (would depend a lot on what your algorithms do).

Cheers
Bhaskar