De : Matt Ettus <address@hidden>
À : Mathias Coinchon
<address@hidden>
Envoyé le : Mercredi 4 Août 2010 15h55
Objet : Re: Re : [Discuss-gnuradio] spurious sidelobes with WBX
On the USRP1, the TX only has CIC filters, not halfband. CIC filters
have some rolloff (which you see), and they do not perfectly remove the
aliases. You can significantly reduce this effect by interpolating the
signal more in the host and correspondingly less in the USRP. This will
move the passbands further out and reduce the amplitude of those sidelobes.
The USRP2 does not have this problem, since it has halfband filters. As
long as you use an interpolation rate that is a multiple of 2 you will
get their benefit.
Matt
On 08/04/2010 05:51 AM, Mathias Coinchon wrote:
> Hi,
>
> Thank you for your answer.
However I am not sure to understand how you
> changed your ant-alias filter because this is something that is
> implemented in the USRP FPGA isn't it ?
>
> As I said when monitoring the baseband IQ signal produced by the
> modulator beofre going to USRP, I don't see these sidelobes, they only
> appear after USRP.
> Is it some rounding problem ? and where does this aliasing come from as
> the signal is centered on 0Hz with no further energy after +/- 750Hz. A
> sampling frequency of 3.2MHz (complex) should be enough. I can try use a
> lower interpolation and so use 4MHz instead of 3.2MHz.
>
>
> Best regards
>
> Mathias
>
>
> ------------------------------------------------------------------------
> *De :* Charles Brain <
address@hidden>
>
*À :* Mathias Coinchon <
address@hidden>
> *Cc :* gnuradio <
address@hidden>
> *Envoyé le :* Mer 4 août 2010, 8h 32min 40s
> *Objet :* Re: [Discuss-gnuradio] spurious sidelobes with WBX
>
> On 03/08/2010 21:50, Mathias Coinchon wrote:
>> Hello,
>>
>> I am using the USRP1 + WBX to create a DAB OFDM signal (1.5MHz bandwidth) in VHF
>> band.
>> The sampling rate used is 3.2MHz (interpolation 40)
>>
>> However, there's something I don't understand. On the picture attached, you can
>> see 2 spurious sidelobs that appear at about 2.5MHz from the center (red arrow).
>>
>> They are not present when monitoring the spectrum of the baseband IQ
signal
>> before going to USRP.
>>
>> They also don't appear when sending a tone signal.
>>
>> Is there any explanation for this ? and any hint to avoid or reduce it ?
>>
>> Thank you for your help.
>>
>> Mathias Coinchon
>>
opendigitalradio.org>>
>>
>> _______________________________________________
>> Discuss-gnuradio mailing list
>>
address@hidden>>
http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>
> It looks a bit like this doesn't it.
>
http://tinyurl.com/2vhge7j> These ears on my DVB-T signal were caused by aliasing.
> when I
changed my anti-alias filter they went away
>
> - Charles
> *
> *
>
>
>
> _______________________________________________
> Discuss-gnuradio mailing list
>
address@hidden
>
http://lists.gnu.org/mailman/listinfo/discuss-gnuradio