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

Re: Newbie problems

Expand Messages
  • ki6lql
    Hello Wolf: Adding an option to find the output sample rate seems like a very good idea. The Creative Labs Audigy card I am using is one of the most common
    Message 1 of 5 , Aug 25, 2012
    • 0 Attachment
      Hello Wolf:

      Adding an option to find the output sample rate seems like a very good idea. The Creative Labs Audigy card I am using is one of the most common cards in the installed base of cards, so I think supporting it this way is a great idea -- and would save you from having to support people with my problem directly. How long before such a feature might be ready to test? Just a question -- no pressure.

      I will try your stopwatch method. How close to I have to get to the actual sample rate before the buffer overflows will go away? Couldn't I just measure the time between buffer overflow errors and steer in the direction that lengthens the time, and repeat until no more overflows?

      "Steering" is an appropriate word, but a little unusual in this context.

      Thanks for your help. You've made an amazing tool available!

      Rick KI6LQL


      --- In SpectrumLabUsers@yahoogroups.com, wolf_dl4yhf <dl4yhf@...> wrote:
      >
      > Hello Rick,
      >
      > Since the problem with "slightly different sampling rates for in- and
      > output", even for the same soundcard, seems to be more common than I
      > expected, I will add an option in SL to try to find the correct output
      > rate (almost) automatically. The problem is it cannot be measured
      > directly (in contrast to the input, where the sampling rate can be
      > measured easily by feeding a reference signal with known frequency).
      > The principle to measure the output rampling rate can be done with a
      > pocket calculator and a stopwatch (but future versions will do this in
      > software): Measure the number of seconds until the output buffer usage
      > increments by a certain amount (say 50 percent, or a certain number of
      > samples). Knowing the output buffer size, and the input sampling rate,
      > and the number of seconds in which the 'excess' samples accumulate, one
      > can calculate a rough estimate (!) of the output sampling rate. Then,
      > the resampler can be 'steered' (is this a proper english word?) to keep
      > the output buffers happy. Downside: Due to the ever-changing resampling
      > parameters, output tones will not be as stable as they could be. So
      > knowing the precise output sampling rate, and entering it in Spectrum
      > Lab's calibration table, will always be the preferred method.
      >
      > All the best,
      > Wolf .
      >
      > Am 25.08.2012 19:22, schrieb ki6lql:
      > >
      > > Thanks Wolf!
      > >
      > > Yes, it is an output buffer overrun error. I don't have a frequency
      > > counter to measure the output sample rate, so I tried a variety of
      > > sample rates. With the input sample rate at 8000, I tried several
      > > rates, even down to 4000, but after each change within a few seconds
      > > the buffer fills to 100% and then a few seconds later the overflow
      > > occurs. The help system does not suggest a method other than the
      > > counter. I have a scope, but it would be far less accurate than a
      > > counter. Or can I keep trying different rates blindly through some
      > > procedure you might suggest.
      > >
      > > As for problem 2, I will look through the complex signal routing in
      > > the KX driver for my Audigy card.
      > >
      > > I appreciate your advice and hope for a little more. Thanks again.
      > >
      > > Regards,
      > > Rick KI6LQL
      > >
      > > --- In SpectrumLabUsers@yahoogroups.com
      > > <mailto:SpectrumLabUsers%40yahoogroups.com>, wolf_dl4yhf <dl4yhf@>
      > > wrote:
      > > >
      > > > Hello Rick,
      > > >
      > > > The 1st and 3rd problem you described may be caused by this:
      > > > The sampling rate used for input (ADC) and output (DAC) run at slightly
      > > > different sampling rates. This causes either the input buffer to
      > > > overflow, or the output buffer to underflow periodically. By default,
      > > > Spectrum Lab assumes that both clocks are the same, and there is no
      > > need
      > > > to resample the output to a "slightly different" sampling rate because
      > > > this costs more CPU time (which may be an issue when running on
      > > batteries).
      > > > How to cure the problem of "slightly different sampling rates" is
      > > > explained in the help system:
      > > > http://www.qsl.net/dl4yhf/speclab/settings.htm#output_resampling
      > > >
      > > > The 2nd problem (unwanted "echo") can only be cured in the soundcard
      > > > settings. SL's own 'component' window is only aware of the signal paths
      > > > implemented in its own software, it doesn't know (nor would it care)
      > > > about 'what's going on int the soundcard'. A few notes on that, and how
      > > > it could be cured in some cases are here:
      > > >
      > > > http://www.qsl.net/dl4yhf/speclab/startins.htm#unwanted_audio_bypass
      > > >
      > > > Best regards,
      > > > Wolf DL4YHF .
      > > >
      > > >
      > > > Am 25.08.2012 06:52, schrieb ki6lql:
      > > > >
      > > > > I've just installed the latest released Spectrum Lab on my core 2 duo
      > > > > 3.0GHZ Windows XP-SP3 machine, and using an Audigy Platinum EFX sound
      > > > > card fed by my old Kenwood TS440SAT which has no built in DSP
      > > > > functions. I have three newbie problems:
      > > > >
      > > > > 1. Both the spectrum and the waterfall are halting, with about 3
      > > halts
      > > > > per second. This is still the case when the FFT is reduced to 1024
      > > > > bins and when the waterfall is not displayed. I have a pretty fast
      > > > > processor of which only a few percent is being used. Is there a
      > > way to
      > > > > get a smoother display?
      > > > >
      > > > > 2. When using the CW filter to listen to a CW station, I hear an echo
      > > > > of the code delayed a couple seconds at about 20 db less, followed by
      > > > > another quieter one, and so one. I've looked for a feedback path from
      > > > > the output to the input but I do see it in the component view, or in
      > > > > the way my sound card is configured by the kX driver.
      > > > >
      > > > > 3. Every minute or so I get a second or so when the sound cuts out,
      > > > > and the ADC converter block in the component view goes red for the
      > > > > same period and then back to green.
      > > > >
      > > > > Obviously this is an incredibly powerful tool for which I would
      > > > > normally expect to pay hundreds of dollars, but I don't seem to be
      > > > > able to configure it properly yet. I also have a fancy Infrasonic
      > > > > Quartet pro sound card with 192 kHZ/24 bit sampling rate that I use
      > > > > with my homebrew SDR receiver, and the built-in Realtec sound "card"
      > > > > on the motherboard that I could switch to if necessary.
      > > > >
      > > > > Please help a newbie-- Thanks.
      > > > >
      > > > > Regards,
      > > > > Rick Meyer
      > > > > KI6LQL
      > > > >
      > > > >
      > > >
      > >
      > >
      >
    Your message has been successfully submitted and would be delivered to recipients shortly.