Loading ...
Sorry, an error occurred while loading the content.

2550Re: Spur in the middle of the spectrum...

Expand Messages
  • Leif Asbrink
    Dec 2, 2010
    • 0 Attachment
      Hello Henk,


      > At the moment I am using a M-Audio Delta44 as my soundcard, but
      > also this soundcard has some kind of noise in the middle of the spectrum. I
      > tried also yesterday my onboard soundcard. The result was poor. Much worse
      > than the M-audio Delta44.
      (from your first posting on this thread)

      > In my case I have some spur in the middle of the bandpass of my SDR. Simon
      > has got an option to reduce this. It is a little better, but it is still
      > present and it is anoying me a little.
      >
      > Does someone know why this is present there and what a solution is to reduce
      > this? Is it something about the adjustment of my SDR on the hardware side?
      (from your second posting on this thread)

      It is non-trivial to get rid of the center spur in a direct conversion
      radio that uses a PC soundcard.

      The spur is caused by several different mechanisms in the hardware.
      The way to remove the spur is to first reduce it to a modest
      level and then apply front end RF gain that rises the antenna noise
      level well above the spur level.

      I can not judge from your image what the dominating problem
      might be because the resolution is too low and I can not read
      the frequency scales. As a first step I suggest that you
      replace the antenna with a dummy load. Then look at the
      spectrum and you should see the spur but no other signals.
      In cas this simple test shows an improvement you suffer
      from ground loop problems. Try to connect the antenna
      ground to the dummy load ground. If that affets the spur
      level you can eliminate that problem with an RF transformer
      at the input.

      The Delta 44 suffers from design errors so it has a rather
      strong enter spur even if nothing is connected to it.
      This link shows how you can modify the card to eliminate
      the errors:
      http://www.sm5bsz.com/linuxdsp/hware/delta44a.htm
      Figure 1 shows the card in its original shape with
      nothing connected to the input. The center spur is symmetric
      and consists of two peaks separated by 10 to 20 Hz.
      (Different in the two hannels)

      The spur is 1/F noise that would cause a very high level
      at frequency zero, but there is a digital high pass filter
      in the Delta 44 that starts to attenuate at 20 Hz or
      so (I do not remember the exact response) The high
      pass filter attenuates 0Hz infinitely an that causes
      the double peak.

      The dominating source of 1/F noise is poor decoupling
      of the 5V DC supply lines to the A/D converters. See
      figure 4.

      The modifications lower the center spur by about 20 dB
      as you can see here:
      http://www.sm5bsz.com/linuxdsp/hware/delta44.htm
      Figures 1 and 2 have the center spur at 48 kHz.
      The modifications clean up the spectrum in several other
      ways too.

      I suggest that you disconnect the cable from the Delta 44
      and look at the spectrum. Then connect only the ground
      pin(s) from your radio hardware. In case you see a difference
      you suffer from the ground loop error. Step 1 here:
      http://www.sm5bsz.com/linuxdsp/hware/d44step.htm

      Your mixer is likely to produce 1/F noise and there might
      be pick up of magnetic fields and many other things. They
      can all be mastered and give you nice spectra like these:
      http://www.sm5bsz.com/linuxdsp/run/pmmx266.htm

      There is an infinitely deep notch at the enter but it is so
      narrow that it is unlikely to cause any problems. If the
      carrier of an AM station would fall exactly on it a normal
      AM detector would fail however.

      73

      Leif / SM5BSZ
    • Show all 14 messages in this topic