I'll take belt drive over the over versions and here is why. First, belt drive tt's in general have high mass platters. That provides not just speed stability but also dampening; dampening not just external vibrations but also motor cogging. The mass of the platter provides inertia to help keep the speed constant. For example, my tt has a 14lbs (6.36kg) platter. A 1 gram change in stylus drag (assuming a drastic change in the record groove) generates 0.00155Nm of torque at the outer rim of the record. Assuming all else is constant except for this torque change, the platter is going to lose 0.29% of its speed in 1 second. Of course the motor is going to increase torque to compensate, but for this analysis you can see the impact of stylus drag on the platter. The deceleration is proportional to platter mass, so if your platter weighs 28lbs, then the speed loss would be half or 0.145%. Conversely, the stylus drag will have a proportionally larger impact on lighter flywheels. Motor speed control and belt compliance play a larger roll in speed stability than inertia, but I wanted to point out the value of higher mass platters. Now imagine a low mass platter directly coupled to a motor. Not saying it is impossible, but it is definitely an engineering challenge to smooth out the torque ripples from the motor and isolate the record from external vibrations. |
Motor speed control circuits have to be designed for the various types of drives. For belt drive tt's the circuits must likely be tuned specifically for the belt compliance/platter mass being used. DD tt's must have altogether different speed control circuits as compared to idler or belt drive. Back to belt drive; I think changing belt types on a tt affects the sound because the spring rate of the belt has changed thereby affecting the dynamic response of the system. The speed control circuits are designed around a specific set of parameters including the belt type. Changing belt types alters the spring rate side of the equation and may make the table sound better or possibly worse. It might be hit or miss. The same goes for adding mass to the platter.
You have to try the Dr. Feickert Analogue iPhone/iPad app. I think it is a killer app. You need a test record with a 3150Hz test tone. You can order one through the maker of this app or in my case, I have a test record with a 3150Hz test tone. This app does three things; it lets you dial in platter speed while the cartridge is tracking in the groove, it analyzes and computes your table's Wow&Flutter (it automatically filters out the record's runout which can accentuate Wow&Flutter) and it charts your platter's speed over time. My particular tt has Wow&Flutter measuring +0.02%/-0.03%. That is within specifications. The speed plot over time shows a small, smooth sinusoidal wave. That might be motor cogging smoothed out by the platter mass, but I think it might be the motor speed controller cycling about the setpoint. I measured the speed of an old Oracle tt and it did a little better than my tt. It had a Wow&Flutter measurement of +0.01%/-0.01%. It also had a smooth, low frequency sine wave for speed vs. time. The Oracle was harder to adjust and dial-in an exact speed of 33 1/3. I would really like to hear from others how their idler drive and direct drive tt's compare with the speed vs time plotted by this app. It might give us some insight into why the different drive systems sound different. |
I also meant to mention that by setting platter speed with this iPhone app, it significantly improved the PRAT of my tt. I had been using the little strobe disc for years to set platter speed. The speed was off by a significant amount using the little manufacturer supplied strobe disc. |
I'd say that high mass would be something like 100 times greater than the mass of the record. That would be about 15 kg.
This discussion got me in the mood to check my platter speed again. After playing several records the tt was warmed up and I know from measurements before that warmup is critical. I had a small bit of drift (about 0.5Hz out of 3150 Hz) after setting the speed several months ago. I dialed speed back in and had some good results: -0.01%/+0.02% Wow&Flutter and the speed plot over time shows a sine wave with a 5 second period. I believe the +/- speed variation over 5 seconds is the control system holding speed about the setpoint (At any rate I feel these values are well below audible detection and are an order of magnitude better than the records themselves.) This app is a very good objective measurement tool, I think, for dialing in speed and checking to see if your tt is operating within the manufacturer's specifications. No more, no less. |
I have never had a chance to try the Timeline device. Some have reported very impressive results with the Timeline. They have seen zero drift of the red line playing the entire side of a record. That is approx. 833 revolutions. I figure if my tt looses one revolution out of 10,000, I would be ok with that. That is 0.01% accuracy. Does that seem reasonable? Well, if so then that would translate to a 30 degree drift of the red line from the Timeline after playing one side of a record. I would still call that good. |
An accuracy of 1/1000 rpm is 0.003%. This is only as good as most high end tt's. It needs to be an order of magnitude better than that. It will be good for dialing in speed, but not for examining Wow&Flutter. If it is off 0.003%, then the platter will gain/loose 1/3 of rotation per 10,000; assuming all else is constant. I'd call that good. Why doesn't anyone talk about the iPhone app? It at least plots speed over time for you. |
I have an FFT analyzer app on my iPhone. I used a B&K calibrated 1000Hz test tone and the FFT plot was dead on. It is about the frequency counter tied to the internal clock. I don't know the accuracy, but I figure it must be within the range of the timeline- quartz crystal clocks inside their processors. |
The iPhone app filters out the record runout. Easy to do since record runout has a 1.8 second period. And I go right back to the record runout issue. Record speed accuracy is at least an order of magnitude worse than any high end table made today. Sure, you might loose between zero to 1/3 of a rotation during one side of record play (approx 833 rotations total). So what? Don't tell me anyone can hear that. WoW&Flutter is more critical an issue and that gets back to the motor quality and it's speed control circuits. Maybe I am wrong, but I don't believe drive types matter that much regarding speed accuracy. I think drive types affect the sound character of the turntable as much as all of the other design features of a given turntable such as plinth, mass, and dissapation of vibrational energy. If Wow&Flutter is <0.05% then it is all about the record and how the table/tonearm extracts the information. I just can't get past the fact that the record center hole accuracy produces a Wow&Flutter in the 0.5% range. Without a self centering tt/platter design, that number cannot be bested. |
Your tt platter is a flywheel. Like any good flywheel design, the bulk of the mass is on the outer edge (Think 2001 A Space Odyssey). The moment of inertial equations illustrate the effectiveness of having the bulk of the mass at the outer edge. A uniform mass flywheel is I=1/2mr^2, but when the mass is concentrated at the outer edge it is I=mr^2. The moment of inertia is doubled. Therefore, a "hollow" flywheel half the mass of a solid flywheel has the same moment of inertia. And the moment of inertia is directly related to the amount of torque it takes to accelerate the platter- double "I" and torque must double to get the same amount of acceleration. Conversely, double "I" and the impact of stylus drag change on platter speed drops by half. |
Hi Dover, one thing to quickly recalculate is that the record spins 0.5556 revolutions per second. (Inverse of 1.8) So the speed error would be 0.008% if it takes 7200 seconds to lose/gain one rotation.
What I believe is that we will hear a difference in PRAT if speed is off before we hear sour notes (tonal pitch); at least most of us. If speed is off by 0.1%, then concert A would be off by 0.44Hz. I do not think that I could tell the difference even if I had a tuning fork to my ear. But I now know that I can sense the pace of the music by small changes in speed. An orchestra, for example, tunes to the oboe. If the oboe is off a Hertz or two, it doesn't matter because the whole orchestra will be matched to it; but the tempo of the music doesn't change- can't do that with a recording. |
Here is another thought to feed on our neurosis. (I hate to suffer alone :) Like I have said before, the tolerance on the center hole of records is such that the record Wow&Flutter is going to be around 0.5%. My favorite iPhone app will show you that too. It will show you the raw W&F as well as the filtered W&F of your tt. Last Winter, I filed out the center hole of my test record a bit so I could center it on the tt platter. It worked and I was able to reduce/change the record W&F. It is not easy to do. The outside diameter of records is not that round either. I had to try to center it relative to the grooves. But now, think about this. Your turntable has W&F in the 0.03% range and a typical record, say 0.5%. The two values are additive, so depending on the position of the record relative to the platter the total W&F could be 0.53% or 0.47% or somewhere in-between. (Min and max values worse case). So it is very likely that you could play a particular record one day and it seems to sound really great- pace is on and the music flows and the next time it sound kind of dead- all based on the randomness of the record position to the platter. Other than my test record, I have never experimented with this concept. I try not to think about it. |
Actually, high precision bearings are not that expensive. Balls with sub micron tolerances can be purchased for pennies- even in ceramic and ruby. Shafts can be had very cheaply that are ground to high precision tolerances. That is the product of an industrial base that today produces robust machinery that lasts 10s of million to 100s of million cycles. The markup that an individual may pay for a bearing at retail is 10-100 times the price that a high volume manufacturer would pay. A good example are the ball bearings in my mower deck. The autoparts store wanted $55 each for these things and I needed 6 of them. I bought the exact same bearings at an online auction site- a pack of 12 for $22. A large volume manufacturer can probably get them for less than a dollar a piece. Think I got junk for $22? That was several years ago and these bearings have lasted just as long as the factory originals.
My tt has a hardened steel ball turning on a flat sapphire disc. It is not only very quiet, but would likely last for a billion cycles. At 33 1/3 rpm that is 57 years continuous running. My great great grandkids will be playing records on it. |
Works for me too. I invite anyone to come listen to the Appassionata on my tt and tell me what's wrong with it. I don't mind constructive criticism- except that it ends up costing big $$$ in upgrades. :) |
I just did an experiment using the first track on Fresh Aire III. Some great drums and synthesizer on that track. I have both the CD and the record so I listened to the first track back to back. I focused on the bass. I found that the bass on the CD is just a bit crisper and deeper. Not a big difference but it is there. So is my tt deficient? Is the difference due to speed stability, cartridge, tonearm or tubes? My preamp tubes are a couple of years old now. |
Another observation that I should note: The record has a much bigger, broader soundstage so the drums sound further away. The CD soundstage being smaller makes the drums sound closer more immediate. I know it is not the case, but it is like the band recorded the song in two different locations. Which one is right? Maybe I shouldn't have both versions. As per the old adage; man with two watches is never sure what time it is. |
Hi Halcro,
Thanks for uploading those pictures. It is exciting to see how different tables perform. First, note that the Raw Frequency plots are nearly the same on both tables as it should be. That is due to the eccentricty of the grooves to the center hole of the record. Is the Raven a belt drive table? I also see different results on my table (belt drive); but the numbers improve the longer I let my table run. Warm up time seems to be more critical for belt drive vs. your direct drive.
What is interesting too is that I see about a 5 second period in the filtered plots which is the same for my table. Since the period is 5 seconds, I would not call that motor cogging. I think it is the speed control circuits. The designers have probably purposefully put in long time constants to make sure the speed correction is slow and gentle so as not to be heard. |
I see variation about the setpoint which is normal. That is the speed control system working. The DD table has a funny pattern to it. Notice the two bumps and then a flat line in the speed plot? The DD table speed control seems to do a correction over about 10 seconds and then rest. Then the cycle repeats. The Raven on the other hand has a pretty constant cycle about the speed setpoint which is very similar to my table. Note that these corrections are within 0.01% to 0.02% of speed. Very fine control.
Also note that even if your speed is off by 0.1Hz, it will take 945 minutes of running time for your platter to gain/lose one rotation. Nothing against the timeline, just that your turntable would have to run for almost a day with the timeline to see how accurate it is. These plots show you results in a matter of seconds. |
Dear Halcro, what an excellent job you did testing your turntables. Your dedication to the hobby is impressive. It looks like the DD table actually increased speed just a bit with the 2nd tonearm while the belt drive dropped by about the same amount. The magnitude of change being just at or under 0.1%. If I were to guess, it looks like you dropped the 2nd tonearm at around the 10 second mark? And on the Raven at about the 11 second mark? Notice that the speed on the DD recovers after 2 rotations, but the belt drive speed remains about 0.1% lower. As for being able to hear that differnce in terms of pace/rhythm, I don't know. Keep in mind too that this is an extreme test beyond any music- dropping a 2nd tonearm onto a platter. |
Hi Halcro, the DD table seems to like a little more load perhaps. Notice that the filtered sine wave has a period of 1.8 seconds or one complete rotation of the platter. That could be the runout of the platter, the motor controller oscillating gently around that speed set point or the way that the app filters the raw signal. The plot of my tt shows the same 1.8 second period. Since the raw signal will show record runout every 1.8 seconds, it is an easy filter to make; but some small amount may be left in the final output.
The belt drive Raven filtered sine wave is showing the slow, gentle speed correction being made by the motor/controller- at least that is my guess. Someone who has done design work and measurements of turntables would have a much more valued opinion than mine. On the Raven, I wonder if given more time, ie. beyond the 10 seconds after you dropped the 2nd tonearm, it had gently corrected the speed back to the original set point- even if it was just 0.1%.
As for the raw waveforms, I'm thinking that one possibility might be the response of the phono cartridge. I see the similar waveform like your TT-101 on my plot- the little breaks at the tips of the sine wave. Or maybe the more likely answer is limitations of the iPhone or the app. The sampling frequency may be too low to round out the peaks. |
Hi Lespier,
Very impressive results. Congratulations on your tt design. I am also amazed at the unfiltered results. Your test record is centered nearly perfect based on the extremely low Wow&Flutter results. Did you do anything special to your test record to center it? I think we all want to know how to do that. |
All measurement equipment has an accuracy and a repeatability factor. As engineers we have to measure our measurement equipment and it can get frustrating and confusing at times. Like the old saying: "A man with two watches is never sure what time it is." I still think the weak link is the test record. Eccentricity tolerances of the center hole is the first and foremost issue; but another problem is flatness. Any variation in the surface of the record is going to cause a speed change at the stylus- ie. the linear speed of the stylus in the groove. I see the same things. The longer I play the test tone the higher the values. The better that I center the record on the platter, the lower the values that I see. Overall, I have results that vary from 0.01% up to 0.04% with my turntable. I have my own 25 year old test record. Maybe the 7" disc is better with respect to flatness as compared to my test record. It would be nice to find a 180g test record.
As for the iPhone app, it occurred to me that I have a test CD with a 1kHz test tone. So I just did a test with the CD. The app reports a mean frequency of 1000.7Hz. The line is dead flat. The deviation is 0%. Short or long test had the same results; but here is an interesting observation. First, I held the iPhone in my hand and the result was 0%; but the raw result showed waviness in the line. Very small peaks and valleys. So I placed a small table in front of my speaker, placed the iPhone on that table and reran the test. This time the line was dead flat, but still had a small wiggle right at the beginning where I had to hit the start button on the screen. So as I suspected before and just saw, holding the iPhone in your hand while recording will cause additional variation. I suspected that before and always rested my iPhone on a table for the tt tests. This just proved my suspicions. |
Also, it looks like to me that the app is reporting peak values, it is not averaging. I noticed that while looking at the handheld CD test results vs. resting the phone on a table. So like I was saying, one blip on the test record, like a high spot and the resulting values will increase. |
I know what you are saying. Our simple measuring equipment has limitations. If you stare at this data too long and try to interpret something that is just not there, then you will start to see fairies dancing on the screen. I think the app is perfect for dialing in platter speed and it gives us some assurance that our turntables are holding speed an order of magnitude better than the capability of the records themselves. And because of that limitation built into the records, the app is limited on accuracy probably to the 0.03%-0.05% range. Keep in mind too, like I said before; the Timeline needs to run for several hours on your tt in order to get a reasonable data set for evaluation. One side of a record play, 600 rotations, is not enough data points. You need thousands of rotations to conclude anything in the 0.01% range or else you will be seeing those fairies dancing around. |
Hi Ketchup,
The fewer data points that you have, then the higher the standard deviation which let's say is increased uncertainty. Just the fact that you are going to make two 0.100" pencil marks 0.02" apart is going to add to the measurement error. A few hours of running time will make the marks far enough apart to at least minimize the pencil mark measurement error.
Ok, you have a measurement after 666 revolutions. Go ahead and let it run for a few hours and see if your results are the same. The results might look worse, or they might look better, but I doubt they will be exactly the same. (I've been known to eat my words before). :) |
I used to think I was normal until I heard a difference in power cords. Why oh why did that have to happen to me. |
Hi Ketchup,
I'm not sure how else to say it. If you have a timeline or someone who has a timeline could do that test- measure after 20 minutes and then do a 2 hour measurement. See if the per rotation error calculates out the same or not. That would dispel my notion about too few data points.
I have a MIT powercord on my CD player. It made a difference; but I would not call it a cheap powercord. btw- seems like I saw a thread somewhere once that claimed a better powercord on their VPI controller, maybe, made a difference? |
I have a pragmatic question. I did some maintenance on my tt last weekend. I checked screw torques and surprisingly, all of the armboard screws turned about a quarter turn. I got it all back together and dialed in speed perfect. All is looking good. I put on Fresh Aire III again. Nice bass, a little tighter than before, I think. So I put on the CD version and give a listen to track 1. On the CD I can hear the skin on the drums; on the record- almost, but just not quite there. My question is: is that the smearing of detail you all are talking about? Is that really due to speed control, or could it be tonearm, preamp, etc? |
