1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Walkman comparison: tape speed fluctuation

Discussion in 'Tech talk' started by mrn, Feb 23, 2017.

  1. mrn

    mrn Member

    Messages:
    69
    Trophy Points:
    18
    Location:
    germany
    Tape-speed stability is cool!
    Crowd-courcing is cool, too:)

    Here is one more project for those, who want to contribute a bit:
    Comparing walkmans from tape-speed-fluctuation point-of-view.
    I will post some of my measurements, just for comparison.
    Feel free to post the results of your measurements too!
    The more devices we will cover, the better.

    Tools you will need:
    1) Master-tape with constant tones pre-recorded.
    You can get such tapes online for few Euros, or create your own by
    recording 10.000 12.500 and 16.000Hz using the following free online tool:
    http://onlinetonegenerator.com/
    and a Deck having stable tape speed.

    2) Measuring SW:
    For the measurements below, I used SpectrumView, a smart-phone app from OxfordWaveResearch.
    You can use also PC-based apps, which can display spectrograms (baudline, Matlab, Octave, etc).
     
    PJ0688 likes this.
  2. mrn

    mrn Member

    Messages:
    69
    Trophy Points:
    18
    Location:
    germany
    1. Sony TC-DM5 (very stable tape speed.)
    Tape used: Cz-Kontrolní Kazeta (w pre-recorded tones from 30Hz to 16kHz).
    Setup: TC-D5M -> headphone-> phone Mic -> SpectrumView -> screenShot.

    IMG_0370.JPG IMG_0373.PNG IMG_0374.PNG
     
    PJ0688 likes this.
  3. mrn

    mrn Member

    Messages:
    69
    Trophy Points:
    18
    Location:
    germany
    2. Sony WM-F20 (less stable in comparison to TC-D5M)
    30+ yrs old machine with signs of usage :)
    new belt, new battery, rest original.

    IMG_0586.JPG IMG_0582.PNG IMG_0585.PNG
     
    Last edited: Feb 23, 2017
  4. mrn

    mrn Member

    Messages:
    69
    Trophy Points:
    18
    Location:
    germany
    3. Sony WM.DD30 (direct drive technology, no belts used!)
    This machine has already fixed center gear ( all DD walkmans suffer from cracking center gears, and this one got 1-2 extra teeth to fill the crack).
    Still the rpm seems to be fluctuating, almost periodically.
    IMG_0428.JPG IMG_0444.PNG IMG_0445.PNG
     
    Last edited: Feb 23, 2017
  5. mrn

    mrn Member

    Messages:
    69
    Trophy Points:
    18
    Location:
    germany
    4. Sony WM-EX50 (disaster :( )
    the machine is in good condition and has new belts.
    the center gear mini toothWeel has been fixed (glued only), and
    the the auto-stop mechanism removed in order not to influence the rpm-s.
    Still the result is crazy. just look at the fluctuating 16kHz signal.

    IMG_0660.JPG IMG_0638.PNG IMG_0641.PNG
     
    Last edited: Feb 26, 2017
  6. enryfox

    enryfox Active Member

    Messages:
    288
    Trophy Points:
    43
    Location:
    close to Milan (Italy)
    Hi,

    Not sure what your scope exactly is, but if you want to measure the transport stability in terms of W&F, the procedure is a bit different: the tone frequency is ~ 3 kHz (recorded on a properly made reference tape) and you need a tool to properly evaluate W&F by accurately measuring the zero crossing.
    You can find such a tool here : http://www.ant-audio.co.uk/index.php?cat=post&qry=library

    Measuring tones at 10 kHz or more on a cassette tape e is quite complicated: with such high frequency, even the best tape decks do not have a really stable amplitude and the cassette shell is also important for a good amplitude stability; a walkman is even more likely to exhibit amplitude fluctuations. Moreover sampling such high frequency with a decent accuracy requires a very high sampling rate (192 kHz) so that zero crossing are correctly measured.

    bye
     
    mrn likes this.
  7. mrn

    mrn Member

    Messages:
    69
    Trophy Points:
    18
    Location:
    germany
    Hi,
    and thanks for the link!
    The aim is to compare walkmans in order to see, how much they are reliable after 20-30 years.
    I am not insisting on posting spectrograms, at all.
    If we get the W&F values for different walkmans directly, that is even better :)

    I also agree: visualizing spectra or spectrograms of a 16kHz pure tone, sampled at 48kHz is not the most precize thing. However, the analysis window is long (1024 samples) covering 340+ periods of the signal (48ooo/16ooo=3, 1024/3>340).
    Why I proposed spectrograms:
    - just by looking at the spectrograms of the outputs of TC-D5 (1) and EX50 (4) one can clearly see, how different the two machines are. No need for comparing numbers.

    Beside, specrograms can show, where the problem might come from.
    example 4 (see above): the EX-50 produces a non-periodic fluctuation of the tape-speed (hence, also non-periodic fluctuation of the the measured tones), probably caused by the glued center gear, which shows now some un-predictable resistance once a while.
    ... and in
    example 5 (see below): the Dual C828 deck shows an amazing, periodic fluctuation, probably caused be some irregular wheel-surface, or hardened pitch-roller, etc.

    Just a remark on the 3kHz vs 10kHz vs 16kHz: visualizing fluctuating 3kHz signal is not as eye-catching (striking) as visualizing a fluctuating 16kHz signal, since the 16kHz one shows 6x more deviation, when displayed on the same 0-20kHz scale.

    Thanks again for the feedback!
    M.
     
    Last edited: Feb 23, 2017
  8. mrn

    mrn Member

    Messages:
    69
    Trophy Points:
    18
    Location:
    germany
    5. Dual C828 deck (only for comparison) having periodic rpm fluctuation:

    IMG_0386.JPG IMG_0348.PNG IMG_0349.PNG
     
  9. enryfox

    enryfox Active Member

    Messages:
    288
    Trophy Points:
    43
    Location:
    close to Milan (Italy)
    Hi,

    I agree that spectrogram are more intuitive than a number, my only concern is that by using high frequency tones you are mixing other effects: most walkman's have a frequency response specified up to 15 kHz and I think that at that frequency the response is already dropping by 3 dB. At 16 kHz you are also including the roughness of the tape, the stability of the shell and the playback head response.

    On top of that, W&F is weighted with a standard curve because the human hear is mostly sensitive to certain frequency fluctuations (I think in the range of few hertz).

    Anyway judging by the graph of the DD (3) I would investigate the origin of the spurious signals as they appear to be distortion rather than tape stability (the frequency is quite stable) ; the latter example (5) shows indeed a quite terrible playback :) !

    bye
    Enrico
     
  10. walkman archive

    walkman archive Administrator Staff Member

    Messages:
    1,883
    Trophy Points:
    113
    Location:
    Barcelona
    It's an interesting and different point of view, but I think it would be better if accompanied of standard weighted W&F measurements with the software Enrico pointed.
    Thanks for the info!
     
  11. walkman archive

    walkman archive Administrator Staff Member

    Messages:
    1,883
    Trophy Points:
    113
    Location:
    Barcelona
    That would be not very precise, though...
     
  12. mrn

    mrn Member

    Messages:
    69
    Trophy Points:
    18
    Location:
    germany
    hi.
    ok, I will add the measurement results using the WFGUI app soon.
    I see the argument regarding "the origin of the spurious signals as they appear to be distortion rather than tape stability".
    I will switch to 3kHz then.
    Thanks for the feedback!
    cheers!
     
  13. mrn

    mrn Member

    Messages:
    69
    Trophy Points:
    18
    Location:
    germany
    tcd5...0.05-0.06
    wm-d6c...0.06-0.065
    wm-dd30...0.07
    dual-c828...0.17(!)

    w_f_tcd5.JPG

    basf cr-e90 tape, recorded in a tcd5 (i know...cascading black boxs...) using the tone generator from the link above.
    soundcard used: iU2 from tascam.
    looking forward for your inputs :)
    cheers!

    ps: @ wm-dd30: scrape flutter
     
    Last edited: Feb 26, 2017
  14. enryfox

    enryfox Active Member

    Messages:
    288
    Trophy Points:
    43
    Location:
    close to Milan (Italy)
    Those numbers are too good indeed !

    the specification for the DD30 is 0.13 DIN, 0.08% WMRS(NAB) .
    the specification for the FX43 (a belt driver plastic walkman comparable with your (2) ) is 0.6% RMS (JIB)

    Some years ago I did a test measure with a DD33 (same transport) using a 3.150 kHz tone recorded with a good tape deck and I got 0.15~0.16% that makes sense because I'm actually measuring the W&F of both tape deck and walkman.

    In the above screenshot you set the frequency selection to 3.150 kHz but then used 3 kHz; does that change the result ?

    bye
     
    alasdair likes this.
  15. mrn

    mrn Member

    Messages:
    69
    Trophy Points:
    18
    Location:
    germany
    hello!
    I knew you will not like the results :)
    honestly, recording with the TCD5, and playing back with the same TCD5 should lead to worse results than expected.
    And definitely not better. So either my sound-card does de-modulate the slightly AM/FM-modulated 3000Hz, or the SW is not really calibrated.

    Anyway, to compensate the good results above, pls let me add few more machines...

    HS-PX510
    > 0.15 (0.30 max)
    < 0.14 (0.25 max)

    WM-EX615 white
    > 0.25 (0.33 max)
    < 0.20 (0.30 max)

    WM-EX670 grey
    > 0.25 (0.40 max)
    < 0.25 (0.40 max)

    WM-EX678 silver (almost new machine)
    > 0.14 (0.20 max)
    < 0.14 (0.30 max)

    WM-EX90 with new belts but motor from an EX50 (shorter shaft :( )
    > 12% (19 max)
    < 15% (19 max)

    PS: I have repeated some of the measurements like 5x,
    - sometimes the values are different between measurements,but only little (0.25 instead of 0.23, etc.),
    - while in case of the D6C, the frequency display shows 3000 +/-2Hz,
    in case of the WM-EX6xx series it is 3000 +200/-0Hz! The difference is so huge, it is unbelievable.
    Below the D6C measurement, now the switch set to 3000Hz.

    wmd6c_wf_005.JPG
     
    Last edited: Feb 27, 2017
  16. walkman archive

    walkman archive Administrator Staff Member

    Messages:
    1,883
    Trophy Points:
    113
    Location:
    Barcelona
    These numbers look more real. However, recording your own calibration tape is something you can do just for 'home use'. If you really want useable results you need a professionally made reference calibration tape.
    If not possible, at least try to use a Direct-drive, quartz locked close-loop dual capstan deck to ensure the best tape stability. I've made my own with my SONY K909ES and though they seem good, they aren't good enough to measure W&F, though they work well for speed.
     
  17. mrn

    mrn Member

    Messages:
    69
    Trophy Points:
    18
    Location:
    germany
    Hi Hugo.
    could you post the results you got with the K909ES?
    ... just for comparison.
    Thanks.

    PS: in ideal case it should be somewhere around "0,022 % (WRMS), ± 0,04 % W peak (IEC)", source here.
    Theory: I believe you got values roughly 1.5x, maybe 2x as high, since your measurement roughly indicates the "Rec-and-Playback" W&M together, ie.

    W&F_total = W&F@Recording +W&F@playback

    So in our case it is
    at max 2x W&F@Recording or at max. 2x W&F@playback.
    Why do I say "at max"?
    Because tiny fluctuations of the capstain speed probably do not happen at the same moment in Recording and in Playback, so not really 2x, but more realistically

    W&F_total = 1.5x W&F@playback, or
    W&F_total = 1.5x W&F@recording.


    Of course, the above theory holds only when using the same setup, same tape, stable generator and stable rec. tools. Correct me if I am wrong.

    If that theory then works, then it is easy to subtract the W&F@Recording values of the TCD5 from all above numbers I presented.
    And if that theory works, then the W&F@Recording value of the TCD5 is roughly 1/2, maybe 2/3 of the W&F_Total published above, ie. 2/3 of the 0.06. That leads to 0.04% (I know you are skeptic ;) )
     
    Last edited: Feb 28, 2017
  18. enryfox

    enryfox Active Member

    Messages:
    288
    Trophy Points:
    43
    Location:
    close to Milan (Italy)
    Hugo, your Revox B215 should be good enough to record a very close approximation of a test tape :) Alex (ANT audio) uses a Revox for its test tapes. It has a modified head for full track recording, but the tape transport is pure swiss made.
     
  19. mrn

    mrn Member

    Messages:
    69
    Trophy Points:
    18
    Location:
    germany
    ... I´ve just sent him an eMail regarding possible spectrogram colormap lookup-table for his spectrogram app.
    (on my machines, only horizontal scrolling worked, and that one still displayed the specrogram in pink-white color scale only).
    cheers!
     
  20. mrn

    mrn Member

    Messages:
    69
    Trophy Points:
    18
    Location:
    germany
    ok..
    the huge fluctuation measured by the w&f tool was caused by the error in storing audio samples.
    sometimes some samples were missing, sometimes empty buffers were stored and analyzed.
    that made the w&f tool think that 3000hz fluctuated a lot.
    and it did not.
    below the same signal in audacity.
    forget win.

    win buffers 1.JPG win buffers 2.JPG win buffers 3.JPG win buffers 4.JPG
     

Share This Page