Dealing with piano distortion on inner grooves with VR

[asqus]

Hey all,

I'm cutting some solo piano and I've noticed that there's a fair amount of distortion when I cut to 7" (worst) or the inner grooves of a 12". Sounds perfectly clear on the outer grooves, but worse and worse as I head inwards.

I know why this is, and why piano is especially hard hit by it (sustained pure tones, rich harmonics, etc.) but I was wondering whether anyone here had some decent mastering advice to try minimise this issue? It's mostly the high notes that are affected, but I'd hate to roll them off too much. Would de-essing help?

[TheCrates]

im new here too... but i guess you will recieve more help if you tell the people which setup you use. for me, i would have no idea where to start helping you. cheers, sven

[asqus]

Thanks sven.

I'm using a standard T560 vinyl recorder setup, although I think the issue of cutting piano-type sounds towards the center of a record isn't limited to this setup - it's a general potential problem related to decreasing speed as you hit the inner edges of the record.

[SpringMan]

It is a well recognized problem considering the absolutely parallel groove walls from the linear tracking cutterhead vs. the increasingly inaccurate tracking from the turntable. Neumann created the TS-66H module to deal with this. In simple terms it adds second harmonic (distortion) to the cutter signal in increasing amounts towards the centre of the disc, above a certain frequency. It works really well.

If you were really super keen you might be able to create a similar effect with an overdriven valve stage creating f2 ( or really accurate simulation thereof) and dial in the right amount progressively in the master. It would take a lot of research to get it right.

Alternatively, cut at 7"33 and use the greater real estate to crowd the cut to the outside. But would anybody buy it?

It would be interesting to consider whether the average degradation due to 33 would be better than the progressive degradation at 45 due to tracking error. I do know that its a mistake to cut slow grooves ( at 33 with small disc diameter) with high levels of top end because the groove lateral vs. circumferential velocity creates a groove that cannot be tracked, and hence creates inevitable (sibilant type) distortion at smaller diameters.
I have the schematics of the TS-66 if needs be.

Heres another tack - do what the classic producers did and choose less demanding, quieter material for the inner tracks, rather than try to beat nature.

[asqus]

Thanks SpringMan, this is really interesting stuff!

I don't think my technical skills are up to what you're suggesting, so I think for now I'll just take your advice and not try to beat nature

[GeorgeZ]

45 rpm is better in any case for a common audio track, see the attached picture with my analysis of 4:30 long audio - 1) ending at 64% for 33 rpm and 2) ending at 100% for 45 rpm - it still has better parameters on any part of the audio for both the cutting (angles) and playback (angles + radii)

[asqus]

Thanks GeorgeZ, this is good to know. What software did you use for this analysis?

[GeorgeZ]

Our own software
See our web pages...

[markrob]

Hi George,

Just curious. How much lower in level would you have had to drop the 33 RPM example to achieve the same or similar results as the 45 RPM version?

Mark

[GeorgeZ]

It is not 100% comparable due to different diameters of the same audio at 45/33 and also much faster worsening of radii in grooves (curvatures depending on accelerations) compared to angles (depending on velocities). But when the land is increased at 33 rpm to reach the same space consumption, it should give the same/similar sound with decrease approx. by 3 dB.

Edit: of course, if there are no highest frequencies (radii problems) or not very much hi-mid range content (possible problems with angles), then the difference between 45rpm and 33rpm single may be lower .

[Greg Reierson]

45 rpm is better in any case for a common audio track, see the attached picture with my analysis of 4:30 long audio - 1) ending at 64% for 33 rpm and 2) ending at 100% for 45 rpm - it still has better parameters on any part of the audio for both the cutting (angles) and playback (angles + radii)

Can you explain that graph in a bit more detail?

Thanks,

[GeorgeZ]

Yes, of course.

There are two layers of information - the first (background) layer consists of dark green profiles, the second (foreground) layer is on top of the dark green ones.

The first two dark green tracks in both simulations are left and right channels (levels in dB seen as square root profiles so not linear, but it doesn't change anything). You can see that both the 45rpm and 33rpm versions have the same loudness.

In the third tracks, there are calculated horizontal amplitudes (dark green) in the background layer and horizontal angles in the foreground layer with color coding: cyan = OK, lightgreen-yellow-orange-red means worsening, orange is on the edge of cuttable groove, red means distorted cut (damaged grooves by the back side of the cutting stylus).
The fourth tracks show the same, but for vertical - amplitudes and angles. It is typical for most audio masters that there are less problems in vertical than in horizontal angles.
Angles also affect the so called pinch effect distortion on playback side, therefore higher angles mean more risk of PED (worst for conical, least for Shibata, SAS).

The fifth tracks have an acceleration parametr in the background, but it is not dependent on diameter or rpm so it is less important compared to the foreground layer showing horizontal radii (curvatures) in the upper part of the fifth track and vertical radii in the bottom part. These are important only for playback styli and often affect loss of highest frequencies (orange color means still good for elliptical styli, but bad for conical ones, red means bad for both, but maybe can be trackable by Shibata, SAS etc).

I chose randomly one 7'' side from our cutting queue to visually show worsening of these geometrical parameters towards the centre and the difference between 33 rpm and 45 rpm. When the simulation is zoomed in, many red parts show only very short peaks, which cannot be audible by human ear. Maybe a better audio master could be found to show it better... but in principle it would show similar results.

[TheCrates]

Very informative, thx Jiri!

[JJZL]

Dear Mr. George

I am quite new here but I dare to make some questions.

Trying to understand your graphics I don't get what you mean for radii and angles.

I understand that you have horizontal and vertical coordinates of the cutter tip, but how that gets converted to angles?

How do you decide which waves are impossible to cut, or before that extreme, to reproduce? Shouldn't that depend on the cutting head and on the cartridge?

Thank you in advance for your patience

[GeorgeZ]

OK, more explanation about angles and radii. Imagine 1kHz sinus signal cut at 33 rpm using +2dB level (peak NTP, in phase) on 100mm diameter. See the attached picture of a groove created by DMM cutting using a standard diamond stylus. The picture shows the ideal groove - how it should look like when the cutting stylus doesn't affect the angles, but in reality max. horizontal angles allowed would be exceeded due to the back side of the stylus resulting in damaged grooves (not showed in the simulation as it depends also on hardness of copper and other factors). The similar is true also for lacquer cutting and any other type of cutting and embossing, but results may be a little bit different due to different styli shapes and their mounting angle.
This problem is often called "velocity limits" in specialized literature.

Edit: + one more answer: Almost any supplied audio files can be cut, but some of them with only decreased cutting levels - reduced amplitudes mean reduced angles (and radii as well). Or so kind of compression or limiting could be used for the problematic part of audio and/or a particular frequency range.

[tragwag]

Very much appreciate the graphics and explanations in this thread, especially GeorgeZ.
I've known these things as a sort of 'accepted truth', but the hard data is super helpful

[JJZL]

Thank you a lot, It is a very clear explanation.

Best regards

[jds265]

Neumann did try pre-distortion and it does work to some degree. The figure shown by GerogeZ shows how the problem occurs, and it varies according to the playback stylus too. An elliptical playback stylus ( Shibata etc) reduce this effect but not everybody uses elliptical stylii. The Neumann pre-distortion system does work but must be set for the desired playback stylus, 0.5mil conical, eliptical etc. Otherwise keep as far out from the label as possible and lastly reduce the level of the recording for the last track. Pity they did not set the standard for records as inside start, as the crescendo mostly occurs at the end where the highest level cutting is possible!