Imagine the stylus is at 5.73" radius. (Weird number but makes the rest easy).
That's 18 linear inches per rev.
At 33.3 that's 10"/sec.
So the period of 10kHz is a mil ... And therefore 0.0005" (five tenths of a thou) of offset between the contact point in the groove walls will cancel in the L+R, and create a false vertical displacement. 
Somebody want to calculate how many degrees out of plumb/perpendicular to tangent that is? 
A laser on the front of the cartridge head tells a sensor about that degree of dynamic error and generates a voltage  a piezo-twisty cartridge mount  corrects for it.
Go!!!


Please pardon the misspellings and occassional insane word substitution I'm on an iPhone

> On May 7, 2015, at 11:44 AM, "Richard L. Hess" <[log in to unmask]> wrote:
> 
> Jamie,
> 
> You are not being a pain in the ass. I alluded to the complex adjustments of the cartridge that Doug Pomeroy has explained so well at other times/places.
> 
> My goal is for everyone to understand that a polarity flip/inversion/reversal is not the same thing as a phase shift.
> 
> The M/S transform involves polarity inversion and summing. Yes, there may be phase errors at the higher frequencies introduced (mostly by) designed-in capacitance that was intended to maintain unconditional circuit stability.
> 
> I have done far more research and experimentation on summing dual-mono tape playback to true mono. The theoretical goal there is to properly adjust the azimuth to minimize single-channel high-frequency combing roll-offs. Once that is done, then adjusting the inter-channel delay using an outboard processor finishes the job.
> 
> In reality, it is not easy (or even possible on program material) to adjust azimuth to a single channel by ear. The Nakamichi Dragon cassette players will do that with some program material as the inner (right) track heads are split and the system compares one half of the right channel to the other half of the right channel. This defines the playback head, in reality, as a six-track head.
> 
> In most instances, with well-designed and well-manufactured heads like those from Nakamichi in the Dragons and even the MR-1s, proper single-channel azimuth will occur close to the point of time-alignment between the two channels (assuming again a mono (single track) recording). So it seems OK to use that to make the mechanical alignment.
> 
> Still, after doing the mechanical alignment, there are end-to-end variations and also once-around cyclical variations, which can best be accommodated by a product such as StereoTool or the "Azimuth" correction in iZotope RX Advanced.
> 
> I do not know how much the alignment of the groove walls change during playback. I think it is in great part due to the differences between the linear-motion cutter head and the pivot-motion tone-arm. I wonder if StereoTool or iZototpe could offer that final correction.
> 
> Yes, I could see in theory how the very highest frequencies could be improved by tracking and correcting this inter-channel timing error. On the other hand, when I attempted a calibration of a disk playback system over a decade ago, the output above about 12 kHz was not stable enough for a truly accurate adjustment, and low-frequencies introduced by centre-hole eccentricities (I think) also made reading the meters more difficult.
> 
> However, I was pleased that I think my system was within 2 dB at 12 kHz and I had gotten it there by comparing CD and LP releases of some of my favourite albums and throwing out the ones that obviously had been remastered.
> 
> I don't do professional disk transfers because I lack the production cleaning tools, the collection of stylii, the ability to deal with off-centre pressings, or the ability to play 16-inch ETs. I am, however, building up an SP-10 with a short SME arm in a made-up wood base on sorbothane pads to see if I can hear a difference between that and my SL-1200 MK II. Most of this stuff came my way, but the two SP-10s' bearings seem to be in great shape.
> 
> The caveat with StereoTool is that I believe (at least at one time) its internal processing was limited to 48 kHz sampling frequency and it downsampled internally when inserted in a 96 kHz stream with no warning. I have not tested the latest versions.
> 
> While in all practical purposes, I think Goran might be correct that the inter-channel delay in disk reproduction may be insignificant, going after that last bit of correction (as long as nothing else gets harmed) might fulfill at least a desire for perfection that may or may not be practical to implement on a large-scale basis.
> 
> Cheers,
> 
> Richard
> 
>> On 2015-05-06 9:54 PM, JAMES HOWARTH wrote:
>> Not to be more of a pain in the ass than I already am, but I was referring to phase, and not just polarity. Perhaps interchannel timing is more precise, but that manifest as rotated phase between the two groove walls.
>> 
>> You guys are talking about polarity as if that were the only variable, and in my humble estimation, despite Goran’s opinion that it’s insignificant I respectfully submit that there’s times recently that taking a more granular approach than a simple polarity flip.
>> 
>> Also this is for me at least pursuant to the statement that hill and dale can’t be derived from MS encoding of the Left and Right channels: it clearly can. The S is the vertical, is the hill and dale, QED.
> -- 
> Richard L. Hess                   email: [log in to unmask]
> Aurora, Ontario, Canada                             647 479 2800
> http://www.richardhess.com/tape/contact.htm
> Quality tape transfers -- even from hard-to-play tapes.