3 New UBER Decks - Is this Turntable's SwanSong? 🦢

Re speed stability.
As has been noted, the numbers produced by the platter speed app vary from test to test. This depends upon how the tester placed the record on the spindle, there is a small amount of clearance, so the record can appear to have different amounts of eccentricity each time it is placed on the platter. Also it depends upon when the test is started relative to the point at which the arm is swinging in and out due to the eccentricity. There will be a different reading if the test started when the arm was at mid swing compared to if it was at the inner or outer extreme. You can see the impact of this on the green trace at the start of the plots. Some show an uptick or downtick at the zero second mark due to the arm being at or near full excursion, (AF0, XD1, AC-2, 1000R, K3), where one is neutral with the arm’s swing more or less at its mid point (TT-101).This biases the low pass numbers in favour of the TT-101. The length of time the test runs also impacts on the final outcome. Then there is the low pass filter that is applied to the raw data, this designed to remove the impact of the records eccentricity. It is somewhat a blunt instrument in that it also removes key data about the platters micro speed stability within each revolution.
It is possible to do multiple tests and pick the best or worst numbers to highlight whatever point you are trying to make.  A reasonable conclusion is that beyond a certain point, the numbers produced are of interest but are not particularly robust data.

But what is pretty consistent and robust from test to test is the yellow trace. (amplitude aside as mentioned above due to record/spindle positioning) The platter speed app is interesting in that it first plots the CARTRIDGE OUTPUT as a function of frequency with respect to time.
In a perfect world where the signal generator sent a constant 3150hz to a perfect cutting lathe, we stamped a perfectly concentric record and played it back on a perfectly speed stable TT, we would see the yellow trace as a straight line at 3150hz
But of course, this is impossible and one of the major deviations from perfection is that the record spindle hole is not in the centre. So if all other parameters were still perfect and we considered a non concentric record, we would see the cartridge output a clean sine wave, symmetrically centred about 3150hz with a constant amplitude per cycle.  IOW, a TT producing a yellow trace that is a badly distorted sine wave is NOT speed stable, although its speed could average 33.333.  I hope that this is self evident.
The program then puts this raw trace thru a low pass filter to remove the effect of eccentricity, we now have the green trace. But what the low pass filter does is filter distortion as well. This distortion is a graphical representation of rapid speed changes. It is no longer visible and what we are left with is a smoothed average which is used to compute the low pass filtered numbers.

A far more accurate and useful metric of the speed accuracy of a TT is to simply look at the raw yellow trace, ignoring the numbers. How close is it to the ideal symmetrical, clean, constant amplitude, sine wave?  Some of the TTs mentioned in this thread, plot significant deviation from this ideal with quite rapid and frequent speed changes.
Remember that the yellow trace is the cartridge output plotted as frequency with respect to time, any distortion of the sine wave is a change in frequency, thus a change in speed. These speed changes would be superimposed on any music that was being played. With this in mind, I invite you to look again at all of the speed plots

.
On top of this conundrum, the platter speed app does not load the system dynamically since the 3150hz tone is of constant amplitude.  This is a whole different set of  equally important measurements. We do not listen to constant amplitude, single frequency tones.

BTW two of the TT plots posted have a low frequency oscillation with a period of approximately 9 seconds. A low frequency oscillation with a multi second period like this is not uncommon and is very difficult to eliminate.

Cheers,

Yes.;
The Yellow trace is a a long way away from the ideal sine wave, its pretty messy. This TT is not micro speed stable. 
As I implied before, on average it takes 1.8 seconds to complete a single revolution ( 33.333 rpm) so the filtered Green trace is nice, but not a robust measure of what is going on inside each revolution.   

BTW the TT-101 is one of the TT's that has the subsonic resonance. It takes heavy duty real-time computing power to eliminate this. Maybe it was available back then but probably not in a device that was economically viable. 

Cheers.  

Yeah, one day I will see the light and buy a CD player!

Halcro.

I think that the numbers and green graph are useful provided their limitations are taken into account.

Mean frequency....

This is a good number to have. We are looking for 3150Hz, so nice and close to this is desirable.

Raw frequency…

Max deviation. (relative ) This is the max percentage deviation, negative and positive, from the mean frequency. Ideally the two numbers should be the same

Max deviation (absolute) This is the max frequency deviation, negative and positive from the mean frequency. Ideally the two numbers should be the same.

What the deviation numbers do not show is the variation between the deviation of individual positive and negative swings. They all should be the same, (constant amplitude) and for this information we need to refer to the yellow graph.

Lowpass-filtered Frequency.....

This data taken from the green trace. The limitation here is that this trace is heavily filtered version of the yellow trace. And it is very inconsistent from test to test.

Look at the 1000R yellow trace and compare this with the TT-101 yellow trace. There is an enormous difference yet, post filtering they yield very similar green traces  

I have run sequential tests on my SP10 Mk 3 and got Low pass -0.01%/+0.01% followed by -0.02%/ +0.03%. I have outlined some of the reasons for this inconsistency. As I said earlier, below a certain percentage, these figures aren't robust.

Obviously we do not want to see any standout perturbations in the green trace and in this way it can be very useful.

My point is that this data is derived from the yellow trace which is taken from the cartridge output. After all, it is the cartridge’s output that we listen to. To really see how the TT is behaving in fine detail, we need to analyse the yellow trace from the perspective of what a perfectly speed stable TT would look like. A clean sine wave, symmetrically centred around 3150hz with a constant amplitude for each cycle. As we can see, advances in technology have clearly improved the situation.  

 The subsonic speed resonance in the TT-101 can be seen as periodic cycling every 5 revolutions. If you look at the first traces you posted (yellow) , you will see a positive peak just after 5 seconds. This peak appears again just after14 and 23 seconds. The smaller positive and negative excursions in between these peaks follow the same pattern with their 9 second spaced partners. There is a recurring pattern to the shape of this waveform with a period of 9 seconds.  

 

Cheers,

Halcro. 
I echo Raul in that I'm happy that you are enjoying your TT's. After all, this is what this hobby is all about. 

Dover.
Thank you for your sage input. Yes the arm/cartridge are part of the loop so they do impact the figures, Halcro has just demonstrated that. That said, within its limitations, the platter speed app does give a reasonable snap shot of what the platter is doing and is thus useful in differentiating between TT's, as I have been trying to demonstrate. But we abandoned it during the development phase, partially for the reasons you mentioned and due to its lack of consistency. Primarily though, it simply doesn't have the granularity we were looking for. We are measuring well in excess of 1 million directly read samples per revolution. 

Raul.

Thanks for your comment. Yes the test record is what most of us use if we are to look at W&F, and as I have said earlier, it is useful to broadly differentiate between TTs. This despite the many variables in play. 
I also pointed out that we abandoned  the platter speed app, this was during the development of K1.
An interesting aside is that K1 had a similar subsonic resonance as seen in the TT-101. It was not always there and we could induce it by stopping and starting.  The controller we used enabled us to tune this aberration out. Clearly visible on the yellow trace, this detail was filtered out on the green.
 At that stage, in its early development, K1 had a similar messy yellow trace to some of the TT's posted here.

Mijostyn

"The  wow and flutter of good modern turntables is hidden in the noise of record irregularity and eccentricities. All this is interesting and totally academic."

We are going to have to agree to differ on that one.
Cheers.