Avoiding 8 Hz in tonearm cartridge resonance

A friend strongly believed there is record crap at 8 Hz which must be avoided.  It seems he was right, the lateral resonance must be centered on 9 Hz or higher to avoid it, otherwise sounding dull from excess resonance stimulated at 8 Hz.  In most 2 axis tonearms the lateral resonance is going to be lower, hence first subjected to this issue first.

And that may be why in long tradition, I believe, excess tonearm mass is associated with "dull" sound.

And also now I have a good explanation as to how a resonance with two peaks and null in the middle can shift in seemingly non-linear ways.

Say I measure maximum resonance at 8 Hz.  It could will be that some of the lower peak area is actually in 7 Hz.

Now suppose I subtract mass to raise the resonance frequency.  But the 8 Hz sounds exactly (or nearly) the same !!!

What may have happened is instead of the lower mode being in a different Hz range, now the upper mode is in a different Hz range, in this example, 9 Hz.  Well, then the level at 8 Hz will sound exactly the same.

When will the 8 Hz sound substantially different?  Certainly as soon as the entire upper mode in in the 9 Hz range, that is with the null at exactly 9 Hz or higher.

That is what exactly seems to have been the case when I reduced the extra headshell mass from 6.1 Hz to 4.9 Hz.  The 8 Hz stayed almost exactly the same, because the null merely shifted from the lower 8 Hz to the upper 8 Hz by about the same amount.

Listening with Newly Optimized Tonearm

When I reduced the "added" headshell mass from 4.9g to 1.7g in order to finally budge the horizontal resonance above 8 Hz mostly (as described in previous article) I wasn't imagining the consequences.

Last night I did my serious listening test playing Side 2 of Blood On The Tracks by Bob Dylan in the Mobile Fidelity Sound Labs audiophile 200g pressing, which I've used in every tonearm test since the Linn was fixed last month.

The bass, while musically fine now, isn't any "stronger" in fact it might be a very little bit less.  The higher mass kept the vertical resonance centered at 12 Hz, the new lighter mass only keeps it down to 13 Hz.  I suspect the effect of these resonances is to absorb and reduce frequencies up to about twice the frequency.  So I'd predict a resonance of 13 Hz to cause reductions as high as 26 Hz.  That's not ideal, and a vertical resonance at 12 Hz or less would be better.  By extrapolation I can see the resonance moving up to 16 Hz with with a stock Ittok with no tape on it at all and my Dynavector 17D3, which would cause bass reduction up to 32 Hz, and at that point it's musically deadly, at least in the bass.

But something else has happened that I just wasn't thinking about, though I've long worried about.

By far and away the biggest change from "optimizing" the mass at 1.7g "added" (this probably means almost 3 g more than a stock Ittok arm, because just the damping tape all over including the headshell itself probably raised the effective mass by about 1g) is that the transparency and resolution has improved enormously.

I'd always thought my recent turntable playback had a bit of obscurity compared with digital inputs.  I noticed this especially listening to Purple Rain earlier this year and it inspired me to do a whole new wave of adjustments to the midrange EQ, then the Bass EQ, and ultimately the Bass Limiting as well.

But this didn't entirely fix the slight opacity in vinyl playback.  I was beginning to just accept it, after all the rest of the sound was sumptuous and wonderful with recent adjustments.

Now, finally, with the optimized effective mass, I have reduced that opacity.  And I believe that change results from moving the horizontal resonance above 8 Hz.  When the horizontal Resonance is centered at 8 Hz, and reaches down a bit to 7 Hz and below as well, it's then picking up and resonating in response to all the physical imperfections of the vinyl.  And all that resonating is not so audible in itself as in how it's constantly modulating the signal, adding an opacity to the sound.

Now I could take this improvement all the way by going back to a stock Ittok arm with no tape added to it anywhere, then the horizontal resonance would probably be centered at 11-12 Hz with not a bit at 8 Hz and below.  That would be even better, except for one thing.  It would push the Vertical Resonance up to 16 Hz, leading to loss of audible bass up to 32 Hz and higher, and making it sound "thin" (and "metallic" as well...with the stock undamped Ittok).

That was the sound I originally had with the stock Ittok and pretty much hated after awhile.  Way too thin.  I presumed then that the remarkable transparency (which was remarkably good in revealing tonearm metallic resonances in particular) was caused by the lack of bass.  Rolling off the bass often gives and artificial clarity...but it's a clarity that can quickly become tiresome.

Then the year or so I listened with the huge 6.1g extra rolled mass gave nice extended and even luxuriant bass, but lack of clarity.  I did not know how to fix that, but now I do.

Optimizing Tonearm Mass

After the turntable was repaired in early August, I was at first not using the "rolled" extra added headshell mass I had been using last year (though the arm and headshell had damping hockey tape) and that was clearly wrong, then I added back in the rolled 6.1g mass I had been using last year--but then decided it was too much, so I replaced it with a "stacked" 4.9g added mass. Comparing all three I guesstimated that I needed to reduce that 4.9 mass to about 2.9g to increase the horizontal resonant frequency (as most easily measured by the Feickert Adjust records) up from 8 Hz to 9 Hz or so.

But yesterday I found it wasn't that simple.  With the mass reduced by 2.2g from 4.9g added (therefore about  2.7g "added") the lateral resonance was still clearly loudest at 8 Hz, as if it had barely changed.  (It's complicated because the lateral resonance is actually bimodal, with high and low resonant frequencies and a null in between them...the true center of the resonance.  So with 4.9g added mass, both of the modes are centered around 8 Hz.  With 2.7g added mass, the upper mode is just barely edging up to 9 Hz, but 8 Hz is still by far the loudest.)

Finally with about 3.2g removed from 4.9g and therefore 1.7g "added" headshell mass* I found the resonance was no longer clearly centered at 8 Hz, though it was still noticeable at 8 Hz.  9 Hz was barely different, and now the center of the resonance seemed to be around 10 Hz (often considered optimal) with a bit more at 11 Hz.  Because of the bimodality it seemed that the resonance barely moved from 8 Hz until it nearly jumped to 10 Hz.  And at 1.7g added mass the bimodality of the resonance is most clear on the Hifi News Test Record.  There's a visible null at about 9.5 Hz and nodes at 11 and 8-9 Hz.  This is about as low as I want the lateral resonance to go.

(*1.7g "added mass" is an educated guess.  I first took off 2.2g from the 4.9g, and then took off another 2g and then added about half of it back in, losing track of the exact changes.  The range of possible values is about 1.2g - 2.5g.)

Meanwhile the vertical resonance is higher.  With no added mass, the vertical resonance is around 14 Hz  or slightly higher (because of low frequency rolloff, the resonance peak is where the graph shows it beginning to trail off).  With 1.7 g added mass it's lowered to about 13 Hz, a desirable improvement.  With 4g added mass it's lowered to 12 Hz and with 6g to about 11.  6g or more added mass might be preferable with regards to the vertical resonance, but it makes the horizontal resonance go to 8 Hz and below and is therefore undesireable.

There's more horizontal mass because horizontally the base of the tonearm has to move on it's ball bearings.  This is a lot more mass than the tonearm itself, but it's centrally located and doesn't increase the effective mass by a lot.  But clearly by several grams, just looking at the present measurements.

This means the optimal tuning is going to be a compromise between making the horizontal resonance frequency high enough and the vertical resonance frequency low enough.  Exactly as I have now done.  And this also means accounting for the lower mode of the bimodal horizontal resonance.  I feel that the lower mode should be somewhat above 8 Hz, and it's OK for the vertical resonance to be as high as 13 Hz.

Korf (previously linked) showed that the tonearm/cartridge resonance is not a simple spring/mass resonance because the stylus is actually suspended in a circular donut.  That makes the resonances bimodal among other deviations from the simple model, and prone to seeming to change in non-linear jumps because of that bimodality.  The only way to get it right is by actual measurement with test records.

It's also desirable to keep the added mass as low as possible to achieve the desired resonance frequencies, because more added mass loads the stylus on warps and warp wow.

So I think I've optimized it now, and it took less added mass than I was thinking it would need.

Antiskate compensation seemed to require 2.5g setting on the Feickert test record.  With these settings the Feickert tracking test is clean until the very highest level where it seems to distort somewhat.  When I first tried this record a few weeks ago, I was getting some distortion on the preceding two tracks.

In the following graphs made using the Hifi News Test Record 2, the horizontal resonance is on the left, and the vertical resonance is on the right.  Each has to be played separately then they are combined in a single track with editing (which is not exactly perfect so things may not line up perfectly among the different mass conditions).  The top pair is 4.9g added mass, the middle is 1.7g added mass (the current adjustment) and the bottom is 0g added mass.

In the top set of 6 graphs the marker is placed at 9 Hz in the lateral resonance test.  Because of rolloff that is occurring below 20Hz at >6dB/octave, the point where the resonance seems to end on the downside is actually the center of the resonance.  With that in mind, you can see in the top pair that the horizontal resonance is clearly at 8 Hz, about 1 Hz lower than the 9 Hz marker.  This is more clear on the Feickert test record where it seems like nearly all of the resonance is at 8 Hz with hardly any at 9 Hz or 10 Hz.

In the middle pair, with about 1.7 g added mass, the horizontal resonance has moved upwards and now it looks like the resonance ends pretty much at 9 Hz though it is still fading away down to 8 Hz.  There is a null around 9.5 Hz, this is the true center of the resonance, and a upper mode that reaches up to 11 Hz.  I am calling this "optimal."

In the bottom pair, with 0g added headshell mass, the horizontal resonance as moved upwards so it's thickest portion is now centered around 11 Hz and reaches up to 13 Hz.  This is high enough that it begins interfering with bass response and is undesirable.  The horizontal resonance seems larger and lumpier as well.  It still doesn't go away completely at 8 Hz despite now being centered at 11 Hz.

Marker at 9 Hz in Horizontal sweep for 4.9, 1.7, and 0g added mass

The next set of 6 graphs has the marker at 12 Hz in the vertical Resonance.  With 4.9g added mass on top, it's clear the peak resonance is about 11 Hz.  With 1.7g added mass in the middle the peak resonance is now around 12-13 Hz, higher but still acceptable.  With 0g added mass, the peak is 14-15 Hz, clearly too high.

Marker at 12 Hz in Vertical Resonance for 4.9g, 1.7g, and 0g added mass

More Anti-Skate Testing

The Feickert test disk has a serious of tracking tests.  The last one breaks up horribly at my previous antiskating compensation setting of 3.1g.  Dialing the antiskating down to 2.2g passes the test.  At 1.5g antiskating, it begins to get ugly again.  It's fairly decent from to 1.8-2.7g.

On the silent groove, 2.65 is as low as it can go without pulling toward the center.  It can go as high as 2.9g without strongly pulling outwards, which it does do at my previous setting of 3.1g.

The overlap between these two tests being 2.65-2.7g, that's where I set the antiskating compensation.

Later I took out the "Orion Test Record."  It has a very intense trackability test.  My cartridge with the above settings was breaking up somewhat on the second to last series and skipped the middle frequency on the last series

Checking the antiskating blank space it seemed like the Linn antiskating adjustment followed the tracking force almost perfectly.  This blank space is much larger than the one on the Feickert disk and you can try different spots more easily.  Though I wonder if the Feickert has some special "roughness" that makes it more correct.

With this in mind I tried lower and higher tracking forces than my usual 2.2g, with the antiskating set the same.  Lower tracking forces resulted in worse tracking (as expected, but I wanted to be sure).  Forces higher than 2.3g sounded no better (I tried 2.5g and 2.7g and it was the same or possibly worse).  It might have improved a tad still further with 2.4g antiskating.  At 2.5g it was pulling toward the side on the blank grove and no better on the trackability test.

So as of 8pm, this is the setting: 2.3g tracking force (on the Linn dial, measures 2.26g) and 2.4g antiskating compensation.  This is only a hair above the 2.2g recommended maximum tracking force and therefore I think it's OK.

The cartridge still skipped the middle frequency of the last series.  I think this test record is damaged from previous testing and I ordered a "mint+, possibly unplayed" one on Discogs.

 

New "Stacked" Headshell Mass

The new mass looks and sounds much better than before.  I aimed for 4.5g added mass but actually ended up with 4.7g after stacking a fair amount of hockey tape on top of the lightweight "headshell" of the Linn Ittok arm.  I also removed some hockey tape from the arm which was getting too close to warped records.  I added a single layer of hockey tape around much of the cartridge too (which was not easy to do) in the hopes of attenuating a high frequency resonance apparent on the Hifi News frequency sweep test.

Sonically I'm feeling I now have turntable reproduction in the big leagues, like the megabuck turntables at shows.  Probably most of what they do is simply getting the basics right, such as effective arm mass for the cartridge.

In order to test turntable speed, I brought out the Feickert test record, which works with the PlatterSpeed app on my Android phone.  Sadly it does not appear there is an equivalent app for iPhone anymore.  The iphone apps have you place the phone on the record, which introduces a lot more weight on the platter and therefore may change the speed slightly on a Linn (which it seems to do) because the suspension is pushed down, changing the angle with the motor pulley and hence speed.

The Feickert record also has a low frequency spot frequency test which seems easier to read visually than the Hifi News low frequency sweep.  The announcer announces each frequency, and it appears my low frequency resonance starts and is very intense at 8 Hz, then just a bit at 9 Hz, and gone at 10 Hz.

The previous measurement based on examining the measured response of the Hifi News test because visual inspection was not revealing, with the larger 6.1g mass, was that the lateral resonance went down to 7 Hz, which is why I decided to use less mass.  But perhaps even less is called for, perhaps 3.5g added mass.

Some say 7-14 Hz is OK.  Others recommend 9-11 Hz for best performance.

Here is some excellent research on arm/cartridge resonance.  The usual spring/mass model is an oversimplification.  Also the specs provided are not as useful as they would seem.  Ultimately you must do measurements and mass adjustments to get the arm/cartridge resonance correct.

 

New LP12 adjustment speed accuracy

 Platterspeed app using Feickert test disk shows my LP12 now at +0.11% speed error, not much, but not as good as the +0.02% it had been a few years ago.  But still very tiny.  I had figured the new much improved suspension alignment would make a slight difference, and it did.  But it would probably be pretty hard to see the error with a strobe disc.

Linn LP12 specification is 0.03% accuracy.  It can be made that good, or even better, if you adjust the motor screws well enough to get the pulley tilt just right.  The synchronous motor is controlled by a very accurate oscillator in the Valhalla board.

1000 Hz reads as 1001.1 Hz, with raw -.31% to +.19%.  Which means if you looked at any tiny bit, it might go in the other direction, you have to get a long term average to see the 0.11% fast.

Wow and Flutter reads 0.009% using two different standards.  That's better than the specs of many top turntables.

Update: I see now that measurements may be even more accurate with the 3150 Hz test.  This shows an average speed of 3153.9, which is a 0.13% error.  The two wow measurments are 0.14% and 0.12%, still within the specs of the better turntables (the best are often spec'd at 0.15% if there is any spec at all).   The raw speed deviations are -0.47% and +0.38%.

Measurements on LP12 with Ittok arm and additional headshell mass

Additional Headshell Mass*

Upon getting my Linn LP12 fixed, and finding that I liked it slightly better with my 6.1g additional headshell mass, which is a squished roll of hockey tape,  I decided it was time to do some measurements with and without the extra mass, and with different levels of antiskate.

The results seem to indicate that the mass DOES improve the low frequency tonearm/cartridge resonance by lowering it considerably in frequency and slightly in level (except it increases the amplitude of the resonance in the right channel).  But it seems to have no measurable impact on anything else, including the 20-20kHz response using the 20 second sweep on Hifi News test record (HFNTR).  So how does reducing the resonance from 13Hz to 9Hz make the bass above 20Hz sound better?  Apparently not by a straightforward effect on the frequency response as such, but by reducing resonances that are intermodulating with the rest of the sound.  Note that a 20 second 20-20kHz sweep will also tend to downplay low frequency resonances which take a while to develop, a pink noise test (not yet done) may be more revealing.

Meanwhile, cranking the antiskate compensation up to 3.1 g (the maximum) seems to have no effect on the vertical resonance.  

The 3.1g antiskate setting also the cleanest performance on the maximum trackability test, which sounds slightly distorted at the 2.2g setting which would nominally match my 2.2g tracking force.  I have been using the 3.1g setting since discovering that it worked best two years ago.  The Ittok antiskate mechanism seems to add desireable horizontal damping as well as antiskating force, and both are useful to me up to the 3.1g setting.

I was motivated to test vertical resonance vs antiskate compensation specifically because of a strange channel imbalance in the vertical resonance.  There is much more vertical resonance in the right channel than the left channel.  This could mean the stylus is climbing the outermost side of the groove, which is what excess antiskate force might tend to do.  But no antiskate setting made any measured difference on the vertical resonance.

Here is a picture showing first the horizontal and then the vertical resonance test in the No Added Mass condition (on top) and the 6.1g Added Mass condition (on bottom).  Both resonances are clearly lower in frequency (desireable) with the added mass.  However, strangely, the vertical resonance in the right channel gets lower and perhaps slightly larger.  The marker is at 9Hz in the horizontal resonance test.

Horizontal and Vertical Resonances vs Added Mass (bottom)

You can see why I was motivated by this result to see if excess antiskating force was causing the increased right channel (bottom in Audacity) resonance amplitude with added mass.

But antiskate does not appear to be causing this issue.  The image below shows the result of testing vertical resonance vs antiskate.  There appears to be no change in the vertical resonance while varying antiskate compensation from 3.1g to 0g.  And it's always worse in the right channel anyway.

Vertical Resonance for 3.1g, 2.2g, 0g, and 1g antiskating compensation

Meanwhile, the Frequency Response sweeps show no bass difference from added mass (on bottom).  In the highs the added mass slightly changes the high frequency measurement (I thought it sounded better WITH the mass, but the graph does not make that look obviously better):

20-20kHz sweep without and with (bottom) added mass

Also the frequency response sweeps show no low frequency dependency on antiskating force level.  Once again I thought the highs sounded better in the 3.1g antiskating condition, but this is not obvious in the measurements.

Frequency Response vs Antiskating level (2.2, 3.1, and 0g)

These results suggest to me that the added mass should be somewhat smaller than 6.1g, perhaps 4.5g or so.  With 6.1g, the horizontal resonant peak is 9 Hz which is fine but has legs reaching down to 7 Hz, which is undesirable.  Also a lower mass might reduce the 9Hz or so increase in amplitude of the vertical resonance.  But without the mass, there is resonant garbage as high up as 16 Hz, which is clearly undesirable.

(*The photo is to show what the 6.1g mass looks like.  It is not fully authentic, because by the time I remembered to take a photo I had already removed the mass again and had started replacing it with a stack of hockey tape, which should work better.  I didn't remove the beginnings of that stack to take the photo,  I simply stuck the mass back onto the headshell again, so it shows quite a bit more hockey tape underneath the roll than there was before.  Prior to this week, I had never even measured the weight of the hockey tape roll either.   When I came up with the idea of rolling up a bunch of hockey tape and squishing it on top of the headshell 18 months ago, I had conceived it as only a temporary measure.  It would enable me to quickly figure out exactly how massive the mass had to be.  Or so I thought.  Somehow all the tests I did back then seemed inconclusive.  Especially just listening to the "warbling" as the test record instructs you to do is quite ambiguous, and then to compare one such ambiguous listening session with a much earlier one is just confounding.  I knew the mass was an improvement but it would be better larger or smaller?   I couldn't tell, and I didn't dare to make it much larger.  I had never gotten around to the much more precise visual analysis in this post, until now.  And so I like to think that in some ways I'm getting better.  And of course I rebalanced the arm whenever I either added or removed the mass.  The mass is to increase the tonearm+cartridge Effective Mass, not the tracking force.  The Linn Ittok is a low to medium mass tonearm.  For low compliance cartridges such as the Dynavector 17D3 moving coil, a high mass (most importantly, high lateral mass) tonearm like the Dynavector DV-505 or Fidelity Research is called for.  One can fudge this a bit with tonearms that have adjustable damping, like my medium mass Jelco 750L, which I haven't tested yet.)