Direct Drive

Yes, tangential tonearms sound better. But they are a higher end thing, and you probably won't hear the benefits until you're spending a fair bit.

How handy are you? It's not hard to make a DIY belt drive table that will be far better than anything you can buy for the same price. I had a pretty hefty budget for a TT and used the money to DIY - and the result is quite phenomenal. And the journey was a treat not to be missed.

DIY forever!

@pedroeb

My turntable is built around a New Way thrust bearing, an air bearing in all three dimensions. This right away puts you in rarified (6 figure) company.

Next is inspiration from the late Tom Fletcher, who reasoned that a tiny motor couldn’t do much to influence a large rotating mass. I use a Premotec 1.8W precision synchronous motor, which does not catch up using feedback, but instead produces a constant rotational speed. Of course, with this setup, you have to bring the platter up to speed by hand, which is a bit quirky, but not at all inconvenient.

Motor resides on its own massive board made of Panzerholz, a wonderfully dead and dense and strong plywood from Germany, which is isolated from the main chassis and the platter.

Motor controller can be as simple as a capacitor or two, or you can build one with (electric) quadrature, which is better. Maybe better left as a project to get around to someday.

Platter is cast iron base covered with a 1" graphite top, both located precisely on an air bearing spindle. (Inspiration Tom Fletcher again) While each rings a little when separated, the assembly is dead as a tomb. If I were doing it again, I would bite the bullet and get New Way to machine the spindle.

Suspension is something which I did not need, so I did not build, because my listening room has a concrete floor built on bedrock, and miles from a highway.

Air supply depends on your situation. If you use an oil-free compressor you don’t need to do much in the way of filtration, but if you need to use oil (quieter, cheaper) you need to go with heroical air filtration to protect that $1K thrust bearing.

There you have the basics. Let us know how you get along! Good luck!

PS: There is a thread on DIY Audio about this. Very long, very detailed.

How does it sound? Well, I can HEAR the noise from the plastic sleeve bearings on that 1/500 HP motor! Only faintly audible as a slight grainy brightness, but audible nevertheless.

Paired with my DIY air bearing tonearm and higher end Koetsu (diamond), it’s a pretty stunning combination. Makes my Nottingham Analogue Mentor upgraded to Dais standard (Fletcher again!) sound highly coloured and even a bit nasty by comparison. And the stock Mentor was a DD killer, IMO.

Oh - and use a short belt.

@lewm 

I don't doubt that the average speed of a modern turntable is right on the money. But, according to accepted statistical theory, there is more to a phenomenon than just the mean.

Consider the platter speed to be a random variable whose distribution of values is some distribution D. Since a normal or Gaussian distribution is completely determined by two parameters, mean and standard deviation, if D ~ N(m,s) the situation is more complex than the mean. In general, the utility of mean as a unique determinant of a quantity is obvious when you consider wealth, like B Gates and one of us, etc. etc.

And a normal distribution is very well behaved at two parameters. Most distributions are described by more. These are the higher level moments, standard deviation (actually its square, the Variance; second), skewness (third), kurtosis (fourth), and the rest of the infinity of central moments have no common name. It is a theorem that any distribution can be uniquely described by its moments.

What this is all getting to, is that the higher order moments of the speed distribution are all noise, noise which is not much considered and not much measured. Our ears measure it though - it comes through as an almost sibilant brightness, a nasty sound. The best DD don't have much of this, but the big new Technics with its associated system seemed to produce way too much of that for me, when I heard a factory audition.

And what can we measure? Mean speed over a window of some duration.

To measure these higher order moments requires very many, very short window measures of mean speed, and applying the correct estimation algorithms. Shorter is better, and 14 bit resolution should be the absolute minimum. The fact that 'speed stability' is reported while standard deviation, skewness, and kurtosis is not, is revealing. It seems to be a matter of the engineering not keeping pace with either theory or perception, IMO.

What do you think?

@pedroeb  Make that the Premotec 9904 111 31813 (as it was then designated), then available from Element 14, IIRC.

@lewm Sorry to be obscure, but some were questioning @dover about the theoretical basis, so I thought there was an interest.

Not saying that it was worse than previous Technics DD. Just saying that there is no such thing as an instantaneous speed measurement in the real world. It’s all averages, or more precisely, means. And means don’t tell the whole story.

That’s the take-away. What is measured isn’t the totality of what’s there.

IMO belt drives sounds far better, but that’s just personal opinion. The theory on how that could be so, given the measurements, is upstream.

@drbond

You say, "AnalogMagik software does measure instantaneous speed. I think Fremer may use it, or possibly have some other software by which he determines the variability and stability of speed control of the turntables that he evaluates."

I was speaking of fundamental physical constraints, not marketing claims. Speed is defined as displacement / time. We’ve known since Planck that the world is quantized, not infinitely smooth. Hence instantaneous speed is a theoretical concept which cannot be physically realized.

It remains only to specify how close a given measurement is to the ideal. Visible light brings constraints in terms of wavelength. Markings to denote the boundaries of the average are required - how finely are they drawn? What is the sampling rate? Each of these introduces quantization AND each of these introduces its own source of error.

That’s all. My point is simply, "How coarse are those averages, and even if they were analyzed completely, what can those averages tell us?" I suspect that our ears/brains, the result of a billion years of evolution (with survival at stake), tell us more.

@pedroeb

Three phase is easily obtained by driving a 3 phase motor with a 1 phase, and tapping the 3 phase power inputs. Viola, 3 phase power. But 3 phase is more complicated than it looks, and so is better left alone.

But it’s pretty much academic if you have a two phase motor with a quadrature motor controller, that is, producing sine and cosine. Then run the motor with sine and whatever mixture of sine and cosine gives the smoothest results. Pretty much the same performance as 3 phase with less grief.

And must agree with Lew on this. Turntable is most important, tonearm next, cartridge last. IMO. Anyway, you can’t get close to top performance from, say, a 10K cartridge with anything less than a first class TT and tonearm, say 50K or preferably more. My own tonearm adjusts azimuth to less than 1 minute of arc, and just this evening I found myself changing over a +/- 2 minute range with clearly audible results. Not many arms can do this, and unless they can, cartridge performance suffers.

@lewm  Yes, what I wrote was ambiguous. I meant, "I agree with Lew (about TT and tonearm being important) AND TT is most important ...".

@pedroeb  Absolutely. Which is also why some mechanism of keeping the record flat is so important. Vacuum hold-down is effective but cumbersome, but a reflex clamp is just as good for most records.

@ketchup  My tonearm is an air bearing DIY with rock solid shafting and precision adjustments in all directions. The actual fine measurement of azimuth is from a fine scale counter marked in 100 micron increments, but capable of repeating to within 20 microns. Since the beam is 254 mm, resolution is calculated quite accurately by the small angle relation tan ~ sin ~ angle in radians, and 20 microns corresponds to about 30 seconds of arc. Adjustments on this scale are not sufficient to impair horizontality much at all, and so I have azimuth on the fly. Adjustment is secured by a brake.