Friday 7 November 2014

MEASUREMENTS: Roksan TMS (1) Wow & Flutter - PlatterSpeed test.

Greetings from this place:
Port city of Xiamen, China - just across from Taiwan.

Wrote this up right before I left Canada to go to Asia...


Hello folks, while you're working on the vinyl LP test and as I prepare to go overseas for a few weeks, I thought it might be interesting to post the objective PlatterSpeed results from the Roksan turntable as well as a couple more pictures of the setup...

It is a thing of beauty:

Although there have been updates to the TMS design (I think TMS 3 is the latest, here's a TSM 2 review from 2003), the appearance and general design principles seem to have stayed constant over the years. The build of this device is top notch and smoothness of operation appears excellent. Switching from 33.3rpm to 45rpm is easy with the push of a button up front.

Using the same Adjust+ Test LP to measure as I did previously with the SL-1200, here are the 33.3 rpm charts and calculations as per PlatterSpeed measured off the same iPad:

As you can see, it measures quite well. It's running a little fast with a mean frequency of 3156.5 Hz instead of 3150 over 60 seconds or so. Wow measurements with the DIN IEC 386 plugin is a little higher than the SL-1200, a similar pattern with the raw frequency maximum deviation. But when lowpassed, the maximum deviation was the same as the Technics at +/-0.02%.

Here it is measured at 45 rpm:

Again, it's a bit fast at 45rpm as well. Remember the target "mean frequency" at 45rpm should be 4252.5Hz. There's only a slight increase in the DIN IEC 386 wow measurement and slight increase in the lowpass-filtered result compared to 33rpm demonstrating good stability with the increased table speed.

It'll be fun to see the LP blind test results! Keep the entries coming - you have to end of November :-).

Enjoy the tunes.


  1. I think it will be safe to say that the hole in the testrecord isn't exactly in the middle.

    Perhaps it is possible to enlarge the hole and try to center it more accurately by freely moving it and 'securing' it with the weight once a perfect center has been reached (might take a very long time to get it right ? )
    perhaps even 'fix' the hole by glueing a new hole on top of the enlarged one.
    That should get rid of the large 'wobble' (that is an exact 33.3 RPM) and yield more accurate results perhaps.

    1. As a digital-only person,I make these observations:
      1. The wow could be genuine off-centred-ness of the spindle or some other problem synchronised with platter rotation, and not just an off-centre hole in the record.
      2. If we can't even trust a test record to have a hole in the middle then the whole exercise seems a little 'academic'!
      3. An off-centre hole in the test record could be compensated for if the test track carried information about rotational position as well as frequency - each disc could be calibrated in one go using a blameless record deck (if such a thing exists), or by making recordings of the disc with it placed at several different rotational positions on the platter of a real world record deck.

      But there you go: as an engineer I am being drawn into the seductive world of the steampunk-computer hybrid. All engineers love motors and bearings and shiny bits of metal that do physical things. And electronics-oriented engineers are aware how they can augment the metal with electronics that provide crystal-locked accuracy. It's a lovely little problem to work on: how accurate can we make it? A 'retro' digital readout would look good - nixie tubes! Introducing a laser somewhere would be fun, too. And so on.

      Non-engineers may be unaware that engineers do this stuff for fun, and may make the mistake of thinking that all 'boutique' 'high end' hi fi is created as a solution to a genuine problem, rather than engineers just playing!

    2. My remark was based on earlier tests made with the exact same disc on a different turntable resulting in similar 'wow'.
      For this reason it is unlikely (given they differ mechanically) that the similar look of the signal is caused by eccentricity problems originating from the decks as they are about equally 'big' in both tests.

      More likely, somewhere down the whole pressing route (making stampers and steps in between) and in creation of the final pressing, perfectly 'acceptable' tolerances create slightly off-centre pressings.
      So in the end the seller of the test discs is at the mercy of (quite old ?) mechanical presses with very small but measurable (perhaps not audible ?) tolerances.
      Acc. to the website below the centering of the disc in question doesn't seem as accurate as most engineers would have liked it to be.

      Indeed as you say... 'how accurate can we make it' is the final question.
      I believe not nearly as accurate as we can measure it but accurate enough for our ears to enjoy what's on vinyl.

      Can't see in the heads of designers and their motivations though, but suspect a LOT of boutique high-end vinyl players are created to make a profit in the end.
      Sell a few with a high profit margin style.
      Most of those decks are truly awesome to see and if I had that kind of cash to burn would buy them just to look at them (like paintings)

      No matter how large, heavy and constant those platters may turn in the real world, one will always be at the mercy of the company that made the vinyl pressing when it comes to tonal accuracy within one rotation. These companies are all about manufacturing speed and quantities.

      In the many years I have played vinyl records (stopped doing that decades ago) I have seen that tone-arm sway from left to right noticeably with some records indicating the hole wasn't centered as well as I would have liked.

      Digital audio increased tonal accuracy, dynamic range and reproducability but other technical limits and financial decisions have replaced the old issues for other technical challenges.

      It keeps engineers playing and aficianados buying !

  2. My well thought out question was pulled from under me by Google during the log in to post process, with a blank text field and no cached fill by using the back button. Thanks, Google! I'll restate in far more simple language.

    Do these tables rectify to DC and re-clock to achieve the resulting platter speed/stability? If not, are you measuring the AC frequency out of the wall during the wow measurements and correcting for AC instability during measurements? AC has a good deal of variation from it's target frequency, usually corrected later to keep clocks accurate by inverse deviation to correct the average target frequency. Thanks for the interesting reading!

    1. Acc. to the manual it is fed by DC (so rectified first) and the motor itself seems to be a 24 pole synchronous motor.

      It would be logical to feed such a motor from a quartz based frequency instead of using the mains.

      Yes... that blank field happens every time. I write what I have to say, copy it, slect the profile and 'preview'. I paste the previously typed text in the blank field and from there on it works. Very annoying.