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

Re: New guy looking at sound card 'calibration'

Expand Messages
  • sm5le
    Hello Wolf Is it better to use internal soundcard SoundMAX Digital Audio insted of external E-MU 0202 for best freq. accuracies and drift ? tnx /Sven
    Message 1 of 13 , Jun 18, 2013
    • 0 Attachment
      Hello Wolf
      Is it better to use internal soundcard "SoundMAX Digital Audio" insted of external "E-MU 0202" for best freq. accuracies and drift ?
      tnx /Sven

      --- In SpectrumLabUsers@yahoogroups.com, wolf_dl4yhf <dl4yhf@...> wrote:
      >
      > Hello Bill,
      >
      > Without a means to monitor ("calibrate") the soundcard's sampling rate
      > against a reference signal,
      > you can expect accuracies in the range of a few ppm (parts per million).
      > Of course this requires measuring the soundcard's sampling rate *once*,
      > which can be done easily with an off-air standard like an audio tone
      > received in AM (not SSB!).
      >
      > After that, the accuracy depends largely on the soundcard (no big
      > surprise..).
      > The manual contains two examples of a 'good' and a 'poor' soundcard,
      > concering the stability of the clock source:
      >
      > http://www.qsl.net/dl4yhf/speclab/frqcalib.htm#soundcard_clock_drift_measurements
      >
      > As you can see, with the E-MU 0202 (used for the 2nd, "good" example)
      > the sampling rate settles to less then 0.1 ppm after > 1 hour of
      > warm-up. Thus, with a sufficiently large FFT (i.e. sufficient
      > *resolution* of the measurement), a 1 kHz audio signal can be measured
      > with an accuracy 0.1 milliHertz.
      >
      > Thus the accuracy of the entire system (with HF receiver / VFO+BFO +
      > Soundcard) is limited by the stability of the radio, not by the soundcard.
      >
      > All the best,
      > Wolf .
      >
      > Am 16.06.2013 15:39, schrieb Bill Cromwell:
      > >
      > > Hi,
      > >
      > > I'm a Ham radio licensee (I know not everybody here is). I have Spectrum
      > > Lab, Spectran, fldigi, and some other digital mode, sound card packages.
      > > Spectran does some of the same tricks that Spectrum Lab does (not a
      > > comparison) but not nearly as many. Ergo, Spectran has a smaller
      > > learning curve and is dramatically quicker to get into action. Spectran
      > > can do more for me when I get it "mastered".
      > >
      > > I have been working my way into the goodies in Spectrum Lab and I am
      > > ready to try calibration for general "frequency measurements". I have
      > > used my dual loop PLL receiver to apply signals from WWV in AM mode (no
      > > BFO to consider and VFO only needs to be 'close enough') to several of
      > > my sound card programs and they all show me that the standard tones are
      > > within just about 100 millihertz on the displays. I do NOT have my sound
      > > card or computer timebase synced to anything like a rubidium or gps
      > > "standard".
      > >
      > > I can look at the manuals myself to see how Spec Lab wants me to adjust
      > > for the sound card clock error (and the other packages, too). What kind
      > > of overall accuracy can I expect from Spectrum Lab under the conditions
      > > I have outlined? I intend to use the information I can get from Spectrum
      > > Lab and Spectran to correct the alignment of my duall loop PLL and BFOs
      > > in that (and other receivers). For the most part, being within 100
      > > milliHertz is overkill in most ham radio operation (never mind about
      > > WSPR for now). But it's always good to know the limitations.
      > >
      > > Incidentally, when I bring in the BFO the actuall PLL "VFO" frequency
      > > and the BFO come into play and it's definitely "out of whack". Starting
      > > point is sound card calibration.
      > >
      > > 73,
      > >
      > > Bill KU8H
      > >
      > >
      >
    • Werner Karn
      Sven, from my experience, that will depend on where you can best control temperature stability. BR Werner 2013/6/18 sm5le
      Message 2 of 13 , Jun 18, 2013
      • 0 Attachment
        Sven,
         
        from my experience, that will depend on where you can best control temperature stability.
         
        BR
         
        Werner


        2013/6/18 sm5le <sm5le@...>
         

        Hello Wolf
        Is it better to use internal soundcard "SoundMAX Digital Audio" insted of external "E-MU 0202" for best freq. accuracies and drift ?
        tnx /Sven

        --- In SpectrumLabUsers@yahoogroups.com, wolf_dl4yhf <dl4yhf@...> wrote:
        >
        > Hello Bill,
        >
        > Without a means to monitor ("calibrate") the soundcard's sampling rate
        > against a reference signal,
        > you can expect accuracies in the range of a few ppm (parts per million).
        > Of course this requires measuring the soundcard's sampling rate *once*,
        > which can be done easily with an off-air standard like an audio tone
        > received in AM (not SSB!).
        >
        > After that, the accuracy depends largely on the soundcard (no big
        > surprise..).
        > The manual contains two examples of a 'good' and a 'poor' soundcard,
        > concering the stability of the clock source:
        >
        > http://www.qsl.net/dl4yhf/speclab/frqcalib.htm#soundcard_clock_drift_measurements
        >
        > As you can see, with the E-MU 0202 (used for the 2nd, "good" example)
        > the sampling rate settles to less then 0.1 ppm after > 1 hour of
        > warm-up. Thus, with a sufficiently large FFT (i.e. sufficient
        > *resolution* of the measurement), a 1 kHz audio signal can be measured
        > with an accuracy 0.1 milliHertz.
        >
        > Thus the accuracy of the entire system (with HF receiver / VFO+BFO +
        > Soundcard) is limited by the stability of the radio, not by the soundcard.
        >
        > All the best,
        > Wolf .
        >
        > Am 16.06.2013 15:39, schrieb Bill Cromwell:
        > >
        > > Hi,
        > >
        > > I'm a Ham radio licensee (I know not everybody here is). I have Spectrum
        > > Lab, Spectran, fldigi, and some other digital mode, sound card packages.
        > > Spectran does some of the same tricks that Spectrum Lab does (not a
        > > comparison) but not nearly as many. Ergo, Spectran has a smaller
        > > learning curve and is dramatically quicker to get into action. Spectran
        > > can do more for me when I get it "mastered".
        > >
        > > I have been working my way into the goodies in Spectrum Lab and I am
        > > ready to try calibration for general "frequency measurements". I have
        > > used my dual loop PLL receiver to apply signals from WWV in AM mode (no
        > > BFO to consider and VFO only needs to be 'close enough') to several of
        > > my sound card programs and they all show me that the standard tones are
        > > within just about 100 millihertz on the displays. I do NOT have my sound
        > > card or computer timebase synced to anything like a rubidium or gps
        > > "standard".
        > >
        > > I can look at the manuals myself to see how Spec Lab wants me to adjust
        > > for the sound card clock error (and the other packages, too). What kind
        > > of overall accuracy can I expect from Spectrum Lab under the conditions
        > > I have outlined? I intend to use the information I can get from Spectrum
        > > Lab and Spectran to correct the alignment of my duall loop PLL and BFOs
        > > in that (and other receivers). For the most part, being within 100
        > > milliHertz is overkill in most ham radio operation (never mind about
        > > WSPR for now). But it's always good to know the limitations.
        > >
        > > Incidentally, when I bring in the BFO the actuall PLL "VFO" frequency
        > > and the BFO come into play and it's definitely "out of whack". Starting
        > > point is sound card calibration.
        > >
        > > 73,
        > >
        > > Bill KU8H
        > >
        > >
        >


      • Bill Cromwell
        Hi, So I have been playing. I have a number of radios but the stars of the show here are a Kenwood R-599, the receiver in the Kenwood Twins lineup and a Ten
        Message 3 of 13 , Jun 19, 2013
        • 0 Attachment
          Hi,

          So I have been playing. I have a number of radios but the stars of the
          show here are a Kenwood R-599, the receiver in the Kenwood Twins lineup
          and a Ten Tec RX-325 that I am planning some "hot rod" mods. The R-599
          has xtal BFOs, xtal heterodyne oscillators, and a free running L-C VFO
          that is adequately stable. It also has a xtal calibrator with front
          panel switch for either 100 kHz or 25 kHz markers. I have always set
          these to zero beat with WWV the best I could in AM mode as recommended.
          I am aware that human hearing response does NOT extend to zero Hertz and
          the audio in our receivers rolls off at some low, non-zero frequency as
          well.

          I examined the signal format from WWV and looked at it in AM mode as
          well as in USB and LSB mode with the Kenwood. Sometimes there is
          interference in the display...maybe from intermod products in the
          receiver...but I was always able to identify the tones in AM mode and in
          either sideband mode I was able to shift the carrier to 1 kHz on the
          spectrum display and identify the tones. Then I turned on the calibrator
          and found it to be almost 100 Hz off at 10 MHz! That is not a real
          surprise. In the AM mode I cannot hear that and the pip is masked in the
          junk that lives at zero on the display. By moving the control on the
          calibrator I may be able to identify it in that mess and try to move it
          zero. However, it looks like dialing the WWV carrier up to 1 kHz on the
          spectrum display (for example) using SSB mode I should be able to adjust
          the calibrator's pip directly on top of the incoming WWV carrier's pip
          and end up with a much smaller error. I am planning to turn the R-599
          over this evening and make the adjustment. And after that I will couple
          my frequency counter clock into the R-599 and set THAT the same way.
          When I get into the RX-325 I will set it's 10 MHz reference the same
          way. The computer, both radios, and the counter run 24/7 except when
          thunderstorms come into the area so warmup drift is eliminated. I'll let
          the calibrator run for a while, too. Hopefully there won't be any
          surprises from that.

          Did I overlook anything?

          73,

          Bill KU8H
        • wolf_dl4yhf
          Hi Sven, During all tests, the external soundcard was always better - MUCH better - last not least because its temperature-dependent oscillator drift was alway
          Message 4 of 13 , Jun 19, 2013
          • 0 Attachment
            Hi Sven,

            During all tests, the external soundcard was always better - MUCH better - last not least because its temperature-dependent oscillator drift was alway very predictable (decay function). All of the internal devices suffer from rapid temperature changes when the PC's internal fan goes on/off. Unless you have a fan-less PC, or a netbook which hardly ever turns on one of the internal fans (but when doing number crunching they usually have to turn on the fan sooner or later, even a netbook with an Atom CPU was no exception).

            All the best,
              Wolf .


            Am 18.06.2013 16:39, schrieb Werner Karn:
             
            Sven,
             
            from my experience, that will depend on where you can best control temperature stability.
             
            BR
             
            Werner


            2013/6/18 sm5le <sm5le@...>
             

            Hello Wolf
            Is it better to use internal soundcard "SoundMAX Digital Audio" insted of external "E-MU 0202" for best freq. accuracies and drift ?
            tnx /Sven

            --- In SpectrumLabUsers@yahoogroups.com, wolf_dl4yhf <dl4yhf@...> wrote:
            >
            > Hello Bill,
            >
            > Without a means to monitor ("calibrate") the soundcard's sampling rate
            > against a reference signal,
            > you can expect accuracies in the range of a few ppm (parts per million).
            > Of course this requires measuring the soundcard's sampling rate *once*,
            > which can be done easily with an off-air standard like an audio tone
            > received in AM (not SSB!).
            >
            > After that, the accuracy depends largely on the soundcard (no big
            > surprise..).
            > The manual contains two examples of a 'good' and a 'poor' soundcard,
            > concering the stability of the clock source:
            >
            > http://www.qsl.net/dl4yhf/speclab/frqcalib.htm#soundcard_clock_drift_measurements
            >
            > As you can see, with the E-MU 0202 (used for the 2nd, "good" example)
            > the sampling rate settles to less then 0.1 ppm after > 1 hour of
            > warm-up. Thus, with a sufficiently large FFT (i.e. sufficient
            > *resolution* of the measurement), a 1 kHz audio signal can be measured
            > with an accuracy 0.1 milliHertz.
            >
            > Thus the accuracy of the entire system (with HF receiver / VFO+BFO +
            > Soundcard) is limited by the stability of the radio, not by the soundcard.
            >
            > All the best,
            > Wolf .
            >
            > Am 16.06.2013 15:39, schrieb Bill Cromwell:
            > >
            > > Hi,
            > >
            > > I'm a Ham radio licensee (I know not everybody here is). I have Spectrum
            > > Lab, Spectran, fldigi, and some other digital mode, sound card packages.
            > > Spectran does some of the same tricks that Spectrum Lab does (not a
            > > comparison) but not nearly as many. Ergo, Spectran has a smaller
            > > learning curve and is dramatically quicker to get into action. Spectran
            > > can do more for me when I get it "mastered".
            > >
            > > I have been working my way into the goodies in Spectrum Lab and I am
            > > ready to try calibration for general "frequency measurements". I have
            > > used my dual loop PLL receiver to apply signals from WWV in AM mode (no
            > > BFO to consider and VFO only needs to be 'close enough') to several of
            > > my sound card programs and they all show me that the standard tones are
            > > within just about 100 millihertz on the displays. I do NOT have my sound
            > > card or computer timebase synced to anything like a rubidium or gps
            > > "standard".
            > >
            > > I can look at the manuals myself to see how Spec Lab wants me to adjust
            > > for the sound card clock error (and the other packages, too). What kind
            > > of overall accuracy can I expect from Spectrum Lab under the conditions
            > > I have outlined? I intend to use the information I can get from Spectrum
            > > Lab and Spectran to correct the alignment of my duall loop PLL and BFOs
            > > in that (and other receivers). For the most part, being within 100
            > > milliHertz is overkill in most ham radio operation (never mind about
            > > WSPR for now). But it's always good to know the limitations.
            > >
            > > Incidentally, when I bring in the BFO the actuall PLL "VFO" frequency
            > > and the BFO come into play and it's definitely "out of whack". Starting
            > > point is sound card calibration.
            > >
            > > 73,
            > >
            > > Bill KU8H
            > >
            > >
            >



          • wolf_dl4yhf
            ... Not a good idea, because in SSB mode, the audio from the receiver depends on the radio s VFO/BFO. That s why I mentioned AM (because even when the AM
            Message 5 of 13 , Jun 19, 2013
            • 0 Attachment
              Am 19.06.2013 14:34, schrieb Bill Cromwell:
              > Hi,
              >
              > So I have been playing. I have a number of radios but the stars of the
              > show here are a Kenwood R-599, the receiver in the Kenwood Twins lineup
              > and a Ten Tec RX-325 that I am planning some "hot rod" mods. The R-599
              > has xtal BFOs, xtal heterodyne oscillators, and a free running L-C VFO
              > that is adequately stable. It also has a xtal calibrator with front
              > panel switch for either 100 kHz or 25 kHz markers. I have always set
              > these to zero beat with WWV the best I could in AM mode as recommended.
              > I am aware that human hearing response does NOT extend to zero Hertz and
              > the audio in our receivers rolls off at some low, non-zero frequency as
              > well.
              >
              > I examined the signal format from WWV and looked at it in AM mode as
              > well as in USB and LSB mode with the Kenwood. Sometimes there is
              > interference in the display...maybe from intermod products in the
              > receiver...but I was always able to identify the tones in AM mode and in
              > either sideband mode I was able to shift the carrier to 1 kHz on the
              > spectrum display and identify the tones. Then I turned on the calibrator
              > and found it to be almost 100 Hz off at 10 MHz! That is not a real
              > surprise. In the AM mode I cannot hear that and the pip is masked in the
              > junk that lives at zero on the display. By moving the control on the
              > calibrator I may be able to identify it in that mess and try to move it
              > zero. However, it looks like dialing the WWV carrier up to 1 kHz on the
              > spectrum display (for example) using SSB mode I should be able to adjust
              > the calibrator's pip directly on top of the incoming WWV carrier's pip
              > and end up with a much smaller error.

              Not a good idea, because in SSB mode, the audio from the receiver
              depends on the radio's VFO/BFO.
              That's why I mentioned AM (because even when the AM receiver is detuned,
              the audio frequency of a 1000.000 Hz signal will still be at 1000.000 Hz
              at the AF output).
              Not sure about WWV / WWVB but afaik they have very stable, calibrated
              audio tones similar to the 1000 Hz 'test tone' transmitted by TV
              stations before going QRT at night. Oh, those were the days.. long gone.

              Cheers,
              Wolf .
            • Bill Cromwell
              ... Hi Wolf, WWV and WWVB are frequency standards. Yes, the carriers and the tones are all very accurate. When I use SSB (or CW) it moves the WWV carrier off
              Message 6 of 13 , Jun 19, 2013
              • 0 Attachment
                On 06/19/2013 01:28 PM, wolf_dl4yhf wrote:
                >
                > Not a good idea, because in SSB mode, the audio from the receiver
                > depends on the radio's VFO/BFO.
                > That's why I mentioned AM (because even when the AM receiver is detuned,
                > the audio frequency of a 1000.000 Hz signal will still be at 1000.000 Hz
                > at the AF output).
                > Not sure about WWV / WWVB but afaik they have very stable, calibrated
                > audio tones similar to the 1000 Hz 'test tone' transmitted by TV
                > stations before going QRT at night. Oh, those were the days.. long gone.
                >
                > Cheers,
                > Wolf .
                >
                >

                Hi Wolf,

                WWV and WWVB are frequency standards. Yes, the carriers and the tones
                are all very accurate.

                When I use SSB (or CW) it moves the WWV carrier off the zero end of the
                spectrum display. It moves the calibrator (or other signal being
                measured the same amount. So with WWV at an arbitrary 800 HZ on the
                spectrum display the signal being examined - in this case my calbrator -
                should be on 800 Hz, too. It's entirely so true that if my VFO changes
                or my BFO changes or my heterodyne oscillator changes (yes, three
                oscillators - one signal) then WWV moves on the display too. But the
                device under test moves also the exact same amount. On the R-599 the
                drift was very small and slow but observable. It was not enough to be
                distracting. In fact, I opened the receiver and aligned my crystal
                calibrator to display exactly on top of WWV's carrier. The error is now
                less than on Hz. As I approached zero I could see the sidebands on the
                WWV carrier that were created by the Crytsal calibrator diminish to the
                point of small undulating bumps on the slopes of WWV's carrier and then
                finally - completely disappear. As I watched over several hours the
                bumps would return, undulate slowly on the slopes, and then again
                disappear. There is nothing I can to about that with the present gear
                but that is much better accuracy than is needed for day to day ham
                operation.

                I am pleased with the results. I fully intend to put the tools to work
                daily here because they are more than adequate. And now that I can
                actually see fine errors I will gradually make improvements to my gear
                as I am able. I know that people who are using WSPR and techniques to
                produce signals from vapor have to be concerned about those "miniscule"
                errors. But those are not a part of my day to day operating - yet. If I
                keep playing with the tools I am sure the improvements will come and
                when I can make WSPR actually work I will certainly turn it on.

                73,

                Bill KU8H
              • sm5le
                RRR Tnx Wolf /Sven
                Message 7 of 13 , Jun 20, 2013
                • 0 Attachment
                  RRR
                  Tnx Wolf

                  /Sven

                  --- In SpectrumLabUsers@yahoogroups.com, wolf_dl4yhf <dl4yhf@...> wrote:
                  >
                  > Hi Sven,
                  >
                  > During all tests, the external soundcard was always better - MUCH better
                  > - last not least because its temperature-dependent oscillator drift was
                  > alway very predictable (decay function). All of the internal devices
                  > suffer from rapid temperature changes when the PC's internal fan goes
                  > on/off. Unless you have a fan-less PC, or a netbook which hardly ever
                  > turns on one of the internal fans (but when doing number crunching they
                  > usually have to turn on the fan sooner or later, even a netbook with an
                  > Atom CPU was no exception).
                  >
                  > All the best,
                  > Wolf .
                  >
                  >
                  > Am 18.06.2013 16:39, schrieb Werner Karn:
                  > > Sven,
                  > > from my experience, that will depend on where you can best control
                  > > temperature stability.
                  > > BR
                  > > Werner
                  > >
                  > >
                  > > 2013/6/18 sm5le <sm5le@... <mailto:sm5le@...>>
                  > >
                  > > Hello Wolf
                  > > Is it better to use internal soundcard "SoundMAX Digital Audio"
                  > > insted of external "E-MU 0202" for best freq. accuracies and drift ?
                  > > tnx /Sven
                  > >
                  > > --- In SpectrumLabUsers@yahoogroups.com
                  > > <mailto:SpectrumLabUsers%40yahoogroups.com>, wolf_dl4yhf
                  > > <dl4yhf@> wrote:
                  > > >
                  > > > Hello Bill,
                  > > >
                  > > > Without a means to monitor ("calibrate") the soundcard's
                  > > sampling rate
                  > > > against a reference signal,
                  > > > you can expect accuracies in the range of a few ppm (parts per
                  > > million).
                  > > > Of course this requires measuring the soundcard's sampling rate
                  > > *once*,
                  > > > which can be done easily with an off-air standard like an audio
                  > > tone
                  > > > received in AM (not SSB!).
                  > > >
                  > > > After that, the accuracy depends largely on the soundcard (no big
                  > > > surprise..).
                  > > > The manual contains two examples of a 'good' and a 'poor'
                  > > soundcard,
                  > > > concering the stability of the clock source:
                  > > >
                  > > >
                  > > http://www.qsl.net/dl4yhf/speclab/frqcalib.htm#soundcard_clock_drift_measurements
                  > > >
                  > > > As you can see, with the E-MU 0202 (used for the 2nd, "good"
                  > > example)
                  > > > the sampling rate settles to less then 0.1 ppm after > 1 hour of
                  > > > warm-up. Thus, with a sufficiently large FFT (i.e. sufficient
                  > > > *resolution* of the measurement), a 1 kHz audio signal can be
                  > > measured
                  > > > with an accuracy 0.1 milliHertz.
                  > > >
                  > > > Thus the accuracy of the entire system (with HF receiver /
                  > > VFO+BFO +
                  > > > Soundcard) is limited by the stability of the radio, not by the
                  > > soundcard.
                  > > >
                  > > > All the best,
                  > > > Wolf .
                  > > >
                  > > > Am 16.06.2013 15:39, schrieb Bill Cromwell:
                  > > > >
                  > > > > Hi,
                  > > > >
                  > > > > I'm a Ham radio licensee (I know not everybody here is). I
                  > > have Spectrum
                  > > > > Lab, Spectran, fldigi, and some other digital mode, sound card
                  > > packages.
                  > > > > Spectran does some of the same tricks that Spectrum Lab does
                  > > (not a
                  > > > > comparison) but not nearly as many. Ergo, Spectran has a smaller
                  > > > > learning curve and is dramatically quicker to get into action.
                  > > Spectran
                  > > > > can do more for me when I get it "mastered".
                  > > > >
                  > > > > I have been working my way into the goodies in Spectrum Lab
                  > > and I am
                  > > > > ready to try calibration for general "frequency measurements".
                  > > I have
                  > > > > used my dual loop PLL receiver to apply signals from WWV in AM
                  > > mode (no
                  > > > > BFO to consider and VFO only needs to be 'close enough') to
                  > > several of
                  > > > > my sound card programs and they all show me that the standard
                  > > tones are
                  > > > > within just about 100 millihertz on the displays. I do NOT
                  > > have my sound
                  > > > > card or computer timebase synced to anything like a rubidium
                  > > or gps
                  > > > > "standard".
                  > > > >
                  > > > > I can look at the manuals myself to see how Spec Lab wants me
                  > > to adjust
                  > > > > for the sound card clock error (and the other packages, too).
                  > > What kind
                  > > > > of overall accuracy can I expect from Spectrum Lab under the
                  > > conditions
                  > > > > I have outlined? I intend to use the information I can get
                  > > from Spectrum
                  > > > > Lab and Spectran to correct the alignment of my duall loop PLL
                  > > and BFOs
                  > > > > in that (and other receivers). For the most part, being within 100
                  > > > > milliHertz is overkill in most ham radio operation (never mind
                  > > about
                  > > > > WSPR for now). But it's always good to know the limitations.
                  > > > >
                  > > > > Incidentally, when I bring in the BFO the actuall PLL "VFO"
                  > > frequency
                  > > > > and the BFO come into play and it's definitely "out of whack".
                  > > Starting
                  > > > > point is sound card calibration.
                  > > > >
                  > > > > 73,
                  > > > >
                  > > > > Bill KU8H
                  > > > >
                  > > > >
                  > > >
                  > >
                  > >
                  > >
                  >
                Your message has been successfully submitted and would be delivered to recipients shortly.