Re: [Discuss-gnuradio] WBX CW transmit problem

[Ales Povalac]

Hi Matt,

thanks for the answer. The problem is not in analyzer setup. Using
large RBW and zero span, it acts as an envelope demodulator and shows
the signal in time domain. Since I used the USRP2 to generate CW
signal only (I=1, Q=0, constant in time), there should be stable
transmitted power over the time. This works well on most frequencies
but not always in the 900MHz band.

It seems like a problem with AGC of ADL5386. RF signal at MODOUT is
nice and its level is stable. Voltage levels at U501/VSET and
U501/VREF are constant. The voltage at U501/VDET_VCTL is oscillating
with the same shape as the RF signal envelope (obviously, it is also
the input of the attenuator).

Regards,
Ales


2011/7/1 Matt Ettus <address@hidden>:
> On 06/30/2011 07:13 AM, Aleš Povalač wrote:
>> Dear all,
>>
>> I have been testing USRP2 with WBX board (#957). Using GNU Radio, I
>> have created a simple CW TX system. The problem: output signal is
>> oscillating when I set the TX frequencies to 900MHz band. Signal
>> output (in time domain) at TX/RX connector from the FSL analyzer is in
>> the attachment.
>>
>> Settings:
>> 868 MHz, 200 kSps, gain 15 dB
>>
>> Same results with:
>> ./tx_waveforms --rate 250000 --freq 868000000 --gain 15
>>
>> The oscillation with ca. 1dB amplitude is present for almost all gain
>> settings except the lowest values. I have observed the problem at
>> several frequencies from 700M to 1G, sometimes with amplitude over
>> 2dB.
>>
>> Any ideas about the cause of this problem? Some PA instability or AGC
>> loop issue?
>
>
> There is DC offset in the signal.  Since you are using a 10 MHz RBW, you
> are seeing both the DC offset and the desired signal beating with each
> other.  If you take your spectrum analyzer out of zero-span mode and
> zoom in on the signal you will see that the amplitude is constant.
>
> Matt
>