Without direct examination, I would say, the peak shown in [1] is energy from the Local Oscillator being directly converted. Many receivers that use direct conversion suffer from the oscillator coming through the mixer. If you were to build a Direct Conversion Tuner from scratch with the option of suppressing the "birdie" at the conversion frequency, the mixer would have to designed as a Double-Balanced Demodulator, that suppresses the carrier wave frequency.

It appears you are using GNU RADIO. Not being completely familiar with its operation, I can guess that this software can be configured to suppress the CW energy before the I/Q stream is decoded.

It is advised not to connect the Tuner's RF input directly to an antenna without a coupling capacitor, there will is a DC component and may damage the LNA in the Tuner chip. Coupling Capacitors on the Differential I/Q outputs are to isolate DC paths between the Tuner and the Converter Chip, also advisable not to directly connect them for the same reason.

-----Original Message----- From: osmocom-sdr-bounces@lists.osmocom.org [mailto:osmocom-sdr-bounces@lists.osmocom.org] On Behalf Of Benedikt Heinz Sent: Sunday, April 15, 2012 7:02 AM To: osmocom-sdr@lists.osmocom.org Subject: RTL-SDR ghost signal at center frequency & hangs in GR spectrum view

Hi everyone,

can someone explain to me, what the origin of the ghost signal close to center frequency is? It can be seen at [1] and [2] - both are EzTV668 with no antenna connected. It seems to me, that the higher the frequency, the more one can see the "burst-like" nature of this signal. Is this some mixer artefact from the E4k tuner? The best workaround is probably tuning a bit off the actual target frequency and using frequency xlate?

Also, in [3] (german) the blind spot at 0Hz due to the coupling capacitors between E4k & RTL is mentioned. Is it possible to replace the capacitors with 0-Ohm resistors to work around this issue, or will this cause other unwanted side-effects? Or does this make no sense at all since one should always tune a bit off and use xlate due to the ghost signal?

I guess it should also be possible (w/ some hot glue & wires) to insert some jumpers (like OsmoSDR) instead, so one can use the ADCs directly for LF/MF with some extra analog stuff?

Regarding the OsmoSDR source - I also gave the RTL-source from gr-baz a try and noticed, that with averaging enabled in the spectrum view in GR, the top-block GUI usually won't react to user input any longer when using OsmoSDR, but this doesn't happen with the gr-baz RTL source, although I used the same sample rate in both cases. Does anyone understand why it doesn't hang with the gr-baz source?

Thanks for your answers!

Regards,

hunz

[1] http://hackdaworld.org/~hunz/rtl-sdr/wo_ant_2m_osmo.png [2] http://hackdaworld.org/~hunz/rtl-sdr/wo_ant_70cm_osmo.png [3] http://www.mikrocontroller.net/topic/253371#2610540

-- Benedikt Heinz hunz@jabber.berlin.ccc.de