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

Re: [wsjtgroup] SignaLinkUSB freq response and S/N

Expand Messages
  • Joe Taylor
    Hi Peter, ... I would be interested in seeing your data for the SignaLink device. The soundcard frequency response that s important for digital modes (at
    Message 1 of 3 , Sep 3, 2008
    • 0 Attachment
      Hi Peter,

      Peter Frenning [OZ1PIF] wrote:

      > Has anyone else tried to characterize the SignaLinkUSB frequency
      > response and S/N properties?
      >
      > I Have, and, to put it mildly, I'm underwhelmed. Compared to my old
      > homebrew interface combined with an aged SoundBlaster16 PCI it sucks as
      > far as pure data is concerned (I have spectralab dumps to document my
      > statements; if anyone are interested mail me directly)

      I would be interested in seeing your data for the SignaLink
      device.

      The soundcard "frequency response" that's important for
      digital modes (at least as I have coded them in WSJT) is
      *not* the digitized amplitude as a function of frequency.
      It's also unimportant for the soundcard to have a very low
      noise figure (high dynamic range) -- as long as the
      essential selectivity against strong unwanted signals is
      established by the SSB filter in your radio. This is nearly
      always the case.

      What *is* important is that the soundcard's sample-rate,
      both input and output, is reliably what the calling program
      wants it to be. For WSJT this is 11025 Hz. Many of today's
      cheap sound systems always sample at 48000 Hz; the data are
      then "resampled" by the driver to (supposedly) give the
      requested rate. Some drivers do an amazingly bad job of
      this, and that's where the biggest problems lie. It's why
      WSJT needs to have those annoying "sample-rate correction
      factors".

      > The real interesting question here is: how important is this in real
      > life? Both interfaces work fine doing everyday PSK31, RTTY, SSTV etc,
      > but how critical does it become when working signals near the detection
      > limit as in WSJT JT65, FSK441 and very weak HF PSK31?

      JT65 transmissions have 126 symbols. Therefore, to avoid
      significant symbol-synch errors the sample rate must
      therefore be known to better than 0.1%. Similar
      requirements apply to the other WSJT modes. This is why the
      WSJT User's Guide states on page 5:

      "7. Take note of two numbers in the first panel of the
      status bar at lower left of the main screen. After WSJT has
      been running for a minute or so they should stabilize at
      values close to 1.0000. If both numbers are between 0.9995
      and 1.0005, the input and output sampling rates of your
      soundcard are close to the nominal 11025 Hz. If either
      number is outside this range, enter it as Rate in (first
      number) or Rate out (second number) on the Options screen.
      WSJT will then make necessary adjustments for the incorrect
      hardware sampling rate or rates."

      -- 73, Joe, K1JT
    • Peter Frenning [OZ1PIF]
      ... I ll not repeat my my message here, just indicate which message it s about. I have been accused of damaging Tigertronics reputation and possibly their
      Message 2 of 3 , Sep 3, 2008
      • 0 Attachment
        Peter Frenning [OZ1PIF] skrev:
        Has anyone else tried to characterize the SignaLinkUSB frequency response and S/N properties?

        .....
        I'll not repeat my my message here, just indicate which message it's about. I have been accused of damaging Tigertronics reputation and possibly their business with my statements. If indeed so, it was not my intention, and I sincerely apologize.

        My questions are still valid (even if my data are bogus, in which case I need to learn why), and I hope to get to the bottom of this. The fact still remains that my particular combination of Radio, Interface and computer, put together as recommended by the manufacturer, does exhibit a data transfer which does, on the evidence of my data, have some disturbing characteristics. Whether this is even significant, due to my errors or faulty equipment remains to be determined.

        I will report back to the groups, when I'm in the clear with this.
        -- 
        Vy 73 de OZ1PIF/5Q2M, Peter
        
        ** CW: Who? Me? You must be joking!! **
        email: peter(no-spam-filler)@...
        http://www.frenning.dk/oz1pif.htm
        Ph. +45 4619 3239
        Snailmail:
        Peter Frenning
        Ternevej 23
        DK-4130 Viby Sj.
        Denmark
        ***********************************
        
        
        
      Your message has been successfully submitted and would be delivered to recipients shortly.