Back in the early 1980's a friend of mine visited a high end audio salon with a test record and a guitar tuner--a handheld device which has a needle meter showing how flat or sharp a particular musical note is. He tested a bunch of turntables and told me that they all ran at the wrong speed. Especially the Linn Sondek, he added. The only belt drive with accurate speed, he said, was the Music Hall MMF-5. Meanwhile, he claimed, all the quartz locked direct drive turntables were correct, and especially his Sony PS-X800.
I didn't believe him then, and I think now he was at best misinterpreting or overgeneralizing now, but it has remained a nagging fear. My friends probably bogus claim has given me a largely unwarranted fear of turntables playing at the wrong speed ever since. Another variant of audiophilia nervosa. As teenager I leaned in the opposite direction. My very first turntable was a Dual 1209 which had a 6% adjustable speed control that changed the position of an idler wheel. It should probably have been adjusted for accurate speed using a strobe light. It came with a tiny strobe disc. I never did that adjustment, I considered it unimportant, and the required neon light seemed unobtainable in those days before online. Using my Tensor lamp was useless. I lusted for a Dual 1219 which had the strobe indicator built in and much more readible. For years I routinely listened with a 1% or thereabouts speed error which I never thought much about, except being pissed that I hadn't gotten the 1219. I figured even a 1% error wouldn't make that much audible difference in a playback situation, but I hated not being able to brag about it. I still haven't actually explored the audibility of these kinds of differences. Such an investigation, done correctly with blind listening tests, might take forever because the change is so small (unless you have perfect pitch, or maybe even if you do). A semitone is a 5.9% increase or decrease, so a 1% error is about 1/6 semitone.
My friend argued that the new turntables with quartz locked motors were the only kind that could really maintain the correct speed. I argued that a belt drive turntable with synchronous motor stays locked either to line frequency or a precision quartz oscillator, and it basically cannot have overall speed error (just wow and flutter) unless the pully is manufactured to the wrong size, which would be a serious manufacturing defect. Any decent belt drive turntable with a synchronous motor ought to have the correct speed. (I found out much later that some early Rega models, from the early 80's and before, did have such a defect--aka "feature"--and ran slightly fast, a problem which was corrected after many complaints.) Meanwhile, a direct drive turntable with a DC motor has to use a servo just to maintain any constant speed, since otherwise the speed is not fixed but a function of current and loading. It has to have a low voltage reference oscillator and the only way to make that reasonably accurate is with Quartz. Meanwhile the frequency of the AC current used in a synchronous motor is already sufficiently accurate. My friend said well then maybe the Linn wasn't really made that well. He was not the first to say that based on what I know now to be misunderstanding.
I went with my friend (who also says he has perfect pitch) to that high end store sometime later and he showed me the guitar tuner needle going up and down on a particular turntable. I couldn't explain the up and down at the time, but I pointed out that the going up and down seemed to center around the correct speed. Yes, he said, but see how it goes a little higher than it goes lower. It was hard to tell actually, sometimes it peaked higher than it dipped, and other times it dipped lower than it peaked, and the ultimate speed accuracy wouldn't depend only on how much higher or lower it goes on one particular round, but how much time it spends higher or lower, which I couldn't estimate, I said. He argued that it couldn't go any higher than it went lower, and that I was just rationalizing. (I have subsequently figured out why it might go higher than it goes lower on any particular round, and vary a lot in between, read on.)
A few years ago I got the Feickert Adjust test record and the PlatterSpeed app for an android phone. (Nothing like that app seems to be available for iPhone.) Now I could do what my friend used to do, but with the kind of accuracy required to do it correctly.
I applied this test to my Mitsubishi LT-3 turntable (which was my only working turntable at the time). I saw right away (as well as heard) the up and down in the 3150 Hz playback tone. The up and down aligns perfectly with record rotation and represents the effect of an off-center hole in the record. It turns out that all LP records have off-center holes. It's apparently impossible to fix during record manufacturing. There have been a number of audiophile products to correct this well known problem, including ways to punch new center holes by hand, and turntables (a couple of Nakamichi models) which automatically re-center records for playback using lasers and computers. Bob Green of The Absolute Sound has written about how important this problem is, and complained that hardly anyone pays attention to it. And yet, it might be one of the most audible defects of LP playback generally. The off-center wow in virtually every LP far exceeds the wow caused by most turntable platters themselves. Various WOW specification weightings eliminate the effect of this very low frequency wow, and it is generally considered more audibly innocuous than higher frequency wow caused by the drive system and bearing. I've sometimes wondered if it adds a euphonic coloration that turntable lovers actually prefer. You can see the off-center wow prominently in turntable tests published by Stereophile. Stereophile reviewers look past the off-center wow to see what the actual platter wow looks like, because that is what actually varies between turntables. Off-center wow is a constant determined by the records you are playing.
The PlatterSpeed app does what I couldn't do by eyeball. It averages the speed over many seconds and gives you a digital readout of the exact average speed. If I remember correctly, the Mitsubishi did have very close to correct speed.
As I was getting my Linn Sondek LP12 repaired a few years ago, the Linn guru Mark (a former dealer) did the speed adjustment AFTER the turntable was set up in its intended location. He said that was the time to do it. I happened to have a $30 optical test record, and we used that to perform the speed testing. The speed on an LP12 (except for the newest models) is adjusted by adjusting the 3 motor screws. The motor pully is either tilted more towards the turntable, or more away from the turntable. It can pretty much be set as accurately as you have patience for. He got it accurate enough that you couldn't see any of the markings on the test disc advance or decline over 30 seconds. That was as good as we had patience for.
At first I thought maybe the distance of the pully from the platter was being changed. But a friend convinced me that would not make any difference. The only thing that could vary the speed would be varying the size of the pully the belt goes around, and the size of the subplatter the belt goes around. Nothing else should matter. The motor speed is locked to a precision crystal oscillator on the Valhalla. On other belt turntables, the speed is taken is taken from the AC line, which has incredibly accurate frequency which can't vary more than a second over many months because power equipment and clocks depend on that, but considerable noise.
So then it bothered me because the method of speed adjustment on the LP12 seemed like a hack.
But now I've determined that the motor screw adjustment on LP12 is not at all a hack. It deals with a fundamental problem with a suspended-subchassis type turntable like the LP12. The problem is that the subplatter might not line up perfectly with the motor, since the subplatter is supported by a suspended subchassis which can move up or down and even slightly to the sides. As the belt ages, or the springs age, this alignment may shift ever so slightly. When the belt wraps around the motor pully at a greater or lesser angle than 90 degrees, the effective radius of the pully changes slightly. The motor screw adjustment fixes this by restoring the pully axis to 90 degrees from the angle of the belt. It actually fixes the problem as close as possible to the source of the problem. Or you could re-tune the suspension, but that is notoriously difficult on a Linn Sondek LP12 and best left to a Linn expert. Tuning the speed with the motor screws is much easier, and it can easily be dialed in as accurately as you choose.
Mark told me I should re-check the speed every year because it could be affected by belt aging. I didn't understand that at the time, but later figured that the belt itself has some influence on the suspension. As the belt loosens, the suspension can tilt slighly away from the pully, affecting the speed. The pully can then be re-adjusted for the new tilt.
Simpler belt-drive turntables which lack the suspended subchassis don't have this problem. They can stay in alignment forever. But then what you give up for that is potential motor and bearing noise and acoustic feedback. The Linn LP12 is one of the quietest turntables made primarily because of the suspended subchassis, which was an idea invented much earlier and originally popularized by the AR Turntable. Using a suspended subchassis, motor noise can be reduced to nil, even with inexpensive motors. And acoustical feedback can be very much reduced also.
But anyway, it is possible that the LP12 my friend once measured hadn't been adjusted in awhile. The store he did these tests at was known for being a bit sloppy, stuff often didn't work right. Or it's still possible that it was just his inaccurate "eyeball" method of estimating the average speed.
So now, a couple years later, I thought it would be good to re-test my LP 12. This would be an indication of how much speed drift there might be in the first two years after changing the Linn belt, when it probably changes the most.
Understanding as I do the way the Linn suspension can affect the speed, I realized I should test records of different weights to check that out also. The Feickert test record itself is 7 inches and very light, but I figured I could stack at least one record underneath it. (That's about all I can do.) All the tests included the Feickert adjust LP itself, combined with another record and/or clamp.
The record has a 3150 Hz tone.
Test Condition Avg Speed 10 sec Readings
Heavy LP and clamp 3150.7 3150 3150 3150
Medium LP and clamp 3151.8 3150 3151 3150 3151
Feickert plus clamp 3152.2 3152 3151 3152 3151
Feickert w/o clamp 3152.7 3152 3152 3151 3152
This shows the small increase in speed as the weight of the LP increases, just as I suspected. I'm not sure what the time interval for the "Average" speeds are and whether they are actually better or not than the "10 sec" readings, but for the purposes of this analysis I am using them. If I were writing hyperbole, I'd go with the 10 sec readings and say my turntable has perfect speed.
With a heavy record, the speed was fast by a mere 0.02%. This is very accurate, 1/300 of a semitone. A review of the very latest Technics SP-10R shows a long term speed error of 0.12%, about six times larger (unless I'm misunderstanding this). I'm not even sure if the Feickert Adjust record or the PlatterSpeed app are 0.02% accurate. I'm worried that if I tried to readjust the motor screws myself I might never get it that accurate again. It's hard to get any kind of adjustment that close.
Update: Specs for the SP 10 Mk3 include less than 0.001% speed drift. But perhaps you still have to set the speed, since it's variable? If there's only a speed knob, it might be difficult to set to 0.02% in the first place, as on many table with variable speed. Speed drift for the Sony PS-X800 is 0.002% and it has a "quartz lock" button you can press, so there's no need to dial in the correct speed first. If the table is currently working, as mine isn't. So perhaps 0.02% is not so above and beyond. Or maybe "speed drift" isn't deviation from actual speed, but change, and I don't see the Linn ever changing either--you can play records all night and it's just not going to change, I believe*, perhaps it has "speed drift" even less than 0.001%. Curiously specs for average speed, or whatever it is I'm measuring here, are hard to find generally. (*The measurements here were taken at significantly different warmup times, on two different nights, and still seem to fit a weight curve that seems plausible. I grasped that meant stable speed, but it wasn't very good proof or measurement.)
The Heavy record was Broken from Nine Inch Nails with a fake "blank" side. I didn't want to risk any of my actual 180g records, which are probably slightly heavier and might therefore play even more accurately.
With a typical medium weight record, the speed was fast by 0.06%. That's 1/100 of a semitone. The speed with the Feickert Adjust record by itself aren't important because no 12" records will be that light, and it's still better than the Technics.
If the speed has shifted from 2 years ago, it hasn't shifted much. Since I didn't measure it with this method, I can't really tell if it has changed at all. Maybe it's exactly the same as two years ago, or maybe it's drifted toward even more accurate speed.
I do now have an explanation for why, even if accurately assessed, the center wow might go slightly more in one direction than another. The center wow interacts with the cartridge-tonearm horizontal resonance, which could be zigging on one side of the record and zagging on the other, as well as constantly fluctuating. Further influences include the damping in the cartridge itself, the anti-skating mechanism, and the levelness of the turntable.
The curious design and construction of the Linn Sondek LP12 has drawn ridicule from my friends for decades. And I used to be of that kind of opinion myself, though my opinions evolved until I actually bought one in 1998. Since about then, I've believed that what often appears as cheapness or madness in the original LP12 is really cleverness. Essentially the design relies on acoustical interface principles to reduce noise transmission from the motor, bearing, and acoustical feedback. For example, the slightly flexible subchassis is designed to reject the bearing noise caused by the single point bearing design. It was "tuned" for that purpose. Only when Linn developed an even quieter bearing design was it possible to replace it with a stiffer subchassis. Even the tiny rubber feet on the bottom of the plinth are tuned to a different frequency than anything else to keep external vibration from entering the turntable.
The hyperbole official (listen to the tune) and unofficial (pace and rhythm) obscures the actual fundamental design principle behind the LP12, which comes from basic audio engineering. Reduce noise.
The Linn is lower noise than the seemingly more solid original AR turntable, and even the almost identical seeming--but heavier built--Ariston RD11. I see the LP12 as a tweaked RD11. The LP12 has a single point bearing, and a subchassis designed to reject the noise that bearing makes. Those were refinements added to the RD11, which was otherwise a very good design, to reduce noise even more. And since then, of course, there have been endless more upgrades.
Though, I add my own principle to that when possible (as to the Ittok XVII tonearm, which I've covered in grams of hockey tape). Reduce resonance.
No comments:
Post a Comment