[Hackrf-dev] Signal Spikes and Jumping

[Michael Ossmann]

On Tue, Aug 06, 2013 at 11:15:02AM -0500, ryan wrote:

>

> I'm seeing strong signal spikes at very round numbered frequencies

> (i.e. 800mhz, 850mhz, 900mhz, screenshot in gqrx:

> http://imgur.com/wRP5ScX,PBZKnN2#0 and in gnu radio-companion

> http://imgur.com/wRP5ScX,PBZKnN2#1)


Spikes at integer multiples of 50 MHz are likely harmonics of the 50 MHz
reference clock that goes to the RFFC5072.  I've observed them, but I
haven't had a chance to work on reducing them.

If anyone gets a chance, it would be nice to confirm this by tweaking
the frequency of that reference clock.   We may be able to reduce or
avoid the harmonics with hardware or firmware changes.


> It's worth noting that these strong spikes seem to appear and

> disappear depending on the hardware frequency and sample rate.  An

> example here http://imgur.com/OnvdPX2,4c9lE14 shows a huge spike that

> disappears after starting and stopping the radio and chaining the

> sample rate, as illustrated here http://imgur.com/OnvdPX2,4c9lE14#1.


You changed the center frequency when you changed the sample rate, and
that is why that spike went away.

When tuned to 1732 MHz, the LO is 2600 - 1732 = 868 MHz.  You were
seeing the second harmonic of the LO, 868 * 2 = 1736 MHz.

When tuned near 1735 MHz, the second harmonic of the LO moved to about
1730, out of view.

Adjusting the IF is the way to avoid these harmonics, but IF selection
is experimental and not yet implemented in any user software as far as I
know.