Is Direct Drive Really Better?

Hold up, I think both both TWL and Macrojack have a couple things wrong here (not that I don't agree with most of the gist of what Tom has said -- I think I said some of the same stuff myself above :-)

"...direct drive motors generally (and I use that word advisedly) use their drive shaft as the main bearing, which typically does not have the precision tolerances of a belt-drive system's main bearing..."

I believe this is a common misconception, which I talked about in my first post to this thread. I may not always know about "generally", but specifically, a DD like my SL-1200 >>does not have a drive shaft<<. The main bearing is similar to the main bearing in any conventional BD, passive meaning unpowered. The motive force to rotate the platter is applied purely by touchless electro-magnetic impulse -- no shafts, wheels or of course belts involved. (Please also see my first post.)

"...Upon looking at the design of my Technics DD, I see that there is no main bearing per se but rather a broad based rotor/stator interface. The notion of that being rocked or deviated from its center seems remote given its diameter, mass and magnetic hold. After all it isn't a pencil point on a hard disk balancing a 12 inch diameter spinning disk. With the Technics table the motor is about 4 inches in diameter and in the case of the SP-10 it is screwed to the motor assembly..."

I think maybe you're being fooled by the appearance of the TT with the platter off. If the SP-10 is anything like an SL-1200, the platter fits over the conically-tapered brass sleeve which forms the base of the spindle, which is integral to the main bearing. When you remove the platter, the spindle is therefore left behind -- you can rotate it by hand. That is the main bearing. What you're describing as 4" in diameter is the stator assembly, which is not "screwed to the motor assembly" as you put it (not sure if you meant to write it that way, since it does't make much semantic sense), but bolted to the cast aluminum chassis, the bearing housing of which you can see centrally located within the stator ring at the base of the spindle/bearing. (Again, if it's anything like the SL-1200 -- please let me know if I am wrong in anyway in translating this arrangement to the SL-10.)

I would recommend anyone fuzzy about the details who really wants to get a feel for how this works to take a trip down to your local pro-sound shop that sells DJ gear and ask to see their display SL-1200 with its platter removed. (With the power turned off, place thumbs or fingers in the opposing holes provided for this purpose, alternate gently lifting one side and then the other to unseat and then carefully lift straight up). Everything I'm talking about should become very clear.

Zaikesman,
My SP-10 MK II has a motor assembly with a top plate and the platter is fastened to that top plate by 3 flat head screws. On my SL 1100A, the arrangement is similar but the platter simply rests on the top plate without any fasteners. I also have an SL 150 MK II and that is similar to what you describe in the 1200 where the platter is an integral part of the motor.
The question remains however whether these DD models remain perfectly concentric in response to stylus drag or are spun off kilter by their looser bearings structures. I think that was what TWL was saying.
I questioned him about this because it appears that my DD tables are not vulnerable in the way he described. Maybe they are.
I also have a Luxman PD 441 and it has a magnetic mechanism which reduces the platter weight on the bearing by 80%. Is this better or worse in relation to Tom's premise?

I don't think a SL-1200 has a "looser bearing structure" than a BD. I haven't used an SP-10 in over 20 years though, and never looked under the hood of one.

The Mk.II pictures on this page, though clearly not the same as my SL-1200, still look to me as if there is a conventional central bearing and no drive-shaft. But on this one, there is a difference noted between the SP-10 Mk.II and Mk.III, with the Mk.II described as having an enclosed motor with what sounds like a sub-platter, vs. the Mk.III's construction which is more similar to the SL-1200.

It's still not clear to me, however, whether in the Mk.II the power is actually applied to a drive-shaft, or whether there is more than one central bearing. My assumption is that in any case where torque is transmitted via a shaft, there must be at least two bearings (as in a BD TT, a motor bearing and a platter bearing).

I'm inclined to view the subplatter as being a part of the top platter, and regard the spindle shaft as not being called upon to transmit the twisting force, but I could be wrong, or the difference could be mostly academic. Maybe the more important point is that the motor turns at a low 1:1 speed (33 1/3 or 45 RPM) and is a rigid part of the chassis. The latter means there can be no relative motion between the motor and the platter. The former means torque will be naturally high and vibration naturally low.

Here is an archived thread that has some more interesting comments, including from Twl.

Here's something that's always made me wonder. Folks talk about quartz/PLL-controlled DD as constantly "hunting" for the correct speed, or attempting to compensate for deviations after they've happened.

What I'd like to know is this: In a BD, I'd assume that the "kick and coast" action of the motor, or any dynamic drag flucuation caused to the platter, would cause the elastic belt to be stretched a bit on one side and relaxed on the other, because then the drive-pulley and the platter would be turning at slightly different speeds. The belt of course would attempt to regain a state of equilibrium in tension, but this would set up an oscillation between sides that would take a while to die out, before which another disturbing discrepancy would have come along, etc. So the belt would constantly be in a state of "hunting" for the correct speed, n'est-ce pas?

On an SL-1200 there is a built-in strobe, so you can see speed deviation and recovery behavior. If you use your finger or a brush to momentarily apply some extra friction to the turning platter (enough to noticably slow the platter -- in other words hundreds of times more friction than a stylus playing a record could ever apply), it will come back up to speed in a controlled, deliberate fashion without visible overshoot or oscillation.

If you put a strobe on a BD and do the same thing, does it appear to behave the same way, or is there some degree of "bouncing around" visible in its recovery behavior? This may be a "trick" question -- any differences might well be too small and/or fast to be visible either way.

Turntables drive systems is an interesting topic and one that I have been contemplating and experimenting with a great deal lately.

I have concluded that speed stability is one of the most important factors in turntable sound quality. For that matter it is also one of the key performance factors in digital audio. It is well known and accepted that digital jitter significantly degrades sound quality. What is remarkable about digital jitter is that such extraordinarily small timing errors could be audible at all. The message here is that our ears are far more sensitive to timing errors than with amplitude errors. With analog the principle and effects of jitter/timing errors are essentially the same. In both cases waveforms are being reconstructed and timing errors create similar distortions. Distortions that for some reason are much more audible than one might expect.

So when it comes to turntable speed stability it is a much more complex issue than many would think. Technically there is no such thing as “constant speed”. Any drive system will have micro variations in speed. As with digital jitter, both the frequency and amplitude of these variations are important. Wow and flutter measurements only quantify large, low frequency variations and don't seem to correlate well with sound.

The main source of speed variation is the motor. All motors cog, or have variations in torque as they rotate. The correct term is torque ripple. In general it is beneficial to isolate the torque ripple to reduce the effect it has on platter speed. Belts and idler wheels provide some degree of isolation. How much of course depends on how compliant the material is. At the same time it is beneficial to have the motor tightly coupled to the platter and rely on the motors torque to keep the platter speed constant. Tight coupling of the motor is best way to reduce the effects of stylus drag. So we have two opposing objectives, coupling and isolation. It would seem that for any motor, platter combination there would be an ideal compromise between isolation and coupling. For example AC motors have a lot of torque ripple so the best compromise is usually a lot of isolation using a stretchy belt. DC motors have far less torque ripple so they typically sound better when used with more rigid coupling (½ tape). With direct drive there is no isolation so the torque ripple must be very low to get acceptable sound.

Beyond the isolation and coupling issue belt and idler drives are both are susceptible to oscillation. We end up with two rotating masses connected with a compliant medium. The worst case is when both the motor and the platter have the same inertia. At first glance it would seem that the inertias are quite different. However, the motor typically spins much faster than the platter so that the inertia actually ends up being similar despite very different mass. Increasing platter mass and decreasing motor inertia helps reduce oscillation. I believe that this is one of the reasons that heavy platters tend to sound better.

As I said earlier both the frequency and amplitude of speed variations is important. There is considerable evidence that very small, higher frequency variations are particularly audible. Power regenerators for AC motors and batteries for DC motors have consistently provided better sound. My own experimentations has also shown that efforts to reduce high frequency noise results in better sound. This would indicate that higher frequency speed variations are more detrimental to good sound. This kind of error sounds remarkably like digital jitter. It sounds harsh, edgy and smeared. Not an artifact that would typically be attributed to speed stability.

Various techniques may be used to reduce these effects but they are never completely eliminated. A heavy platter will reduce high frequency variations more than a light platter. However, lower frequency problems are less effected. For example the effect of stylus drag is different but not less with a heavy vs. light platter. With a light platter a heavily modulated passage will reduce the speed but because of low inertia it will quickly recover. On the other hand a heavy plater will be slowed less but it will take longer for the speed to recover. So platter mass only changes the frequency of stylus drag effect and does not eliminate it. However, it would seem that lower frequency variations from a heavy platter would generally be more benign.

So after all the ramblings the question is what is the best approach? As many have surmised it mostly boils down to implementation. It also is a matter of compromises. Some will prefer one set of compromises over another based on their tastes. However, I do believe that direct drive has the greatest potential. With careful design and implementation direct drive can result in less compromises. With sufficiently low torque ripple and noise the results are remarkably good. Good enough that a direct drive offering from Teres is in the works.