Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: An analysis of turntable rumble (Read 12258 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

An analysis of turntable rumble

The following is an analysis of the rumble response of my turntable, with a needle drop of Autechre's Untilted (a record I've been using quite often), against the FLAC version of the same album I downloaded off of Bleep. Other reports have not convinced me of the exact parameters necessary for a proper filter (and are wildly conflicting at that), so hopefully this will put some sort of bound on what to use.




Analysis details

The reference WAV is a FLAC of the first song "LCC" purchased on Bleep, considered equivalent to the CD version. The vinyl WAV is a 24/96 recording of the very first play of the side containing LCC. Recording chain is Goldring 1012 -> Music Hall MMF-5 -> TCC TC-750 -> RME DIGI96/8 -> MAGIX Samplitude 6. The recording had no significant pops and was normalized to peak volume prior to analysis. No calibration was done to the turntable prior to recording and all antiskate/tracking force settings were at dealer defaults.

Both WAVs were downsampled to 8khz with foo_ssrc in Ultra mode. Analysis was done in LabVIEW with the first 3 million samples (over 6 minutes) from each file fed into an amplitude spectrum plot. The left and right channels of both WAVs were mixed into L+R and L-R channels during analysis to better isolate the effects of vertical vs horizontal groove/cartridge distortion. The LP WAV was delayed by a fraction of a second due to the leadin groove which was not removed.

An analysis of turntable rumble

Reply #1
L+R channel

Here is a complete amplitude spectrum plot of the summed L+R channel, corresponding to the horizontal groove modulation, from 0 to 4khz. The CD signal is clearly larger than the LP signal, by roughly 3-5 db, for most of the spectrum. This does not significantly affect the analysis.
http://www.hydrogenaudio.org/forums/index....pe=post&id=1750

At around 1760hz, note that the frequency response of the LP is slightly shifted up with respect to the CD. The shift is roughly 0.4% of the original signal. The large peaks in the CD version do not appear in the LP version. I suspect this is because wow and flutter in the recording tend to smooth out and heavily reduce these components over the course of 6 minutes of data, but I have no concrete evidence to back that up.
http://www.hydrogenaudio.org/forums/index....pe=post&id=1749


At 30hz, this frequency shift and lower amplitude are still visible on the LP signal. The frequency shift is still around 0.4% and the peaks on the LP response are somewhat well correlated to the CD.
http://www.hydrogenaudio.org/forums/index....pe=post&id=1747


At 20hz the correlation is not nearly as evident. The LP is still of roughly the same amplitude with respect to the CD as with higher frequencies, so in my opinion, a rumble filter is not yet necessary. However, it is somewhat debatable if any real information actually exists in the groove here - or, for that matter, if any exists in the CD as well.
http://www.hydrogenaudio.org/forums/index....pe=post&id=1746


At 14hz the rumble begins to dominate the signal and a large harmonic spectrum develops with a fundamental of 0.55hz, or 33 cycles per minute. This definitely would need to be attenuated with a rumble filter.
http://www.hydrogenaudio.org/forums/index....pe=post&id=1745


Finally, here is a frequency response from 0 to 40hz of the L+R channel. The vinyl response peaks around 8hz, which is a very believable resonant frequency of the tonearm.
http://www.hydrogenaudio.org/forums/index....pe=post&id=1744

An analysis of turntable rumble

Reply #2
L-R channel

Here is a complete amplitude spectrum plot of the differenced L-R channel, corresponding to the vertical groove modulation, from 0 to 4khz. Right away the LP signal is quite obviously closer in amplitude to the CD signal, compared to the L+R signal. This may be due to imperfections in the cartridge but should not affect the analysis.
http://www.hydrogenaudio.org/forums/index....pe=post&id=1743


At 140hz the LP has a very clear correlated signal and shows the characteristic 0.4% speed increase.
http://www.hydrogenaudio.org/forums/index....pe=post&id=1742


At 80hz the LP signal is further attenuated, although there is clearly some signal there. This suggests that bass channels are in fact mixed down to mono to some degree, but people who say that rumble filters should cut L-R out at 140hz are nuts.
http://www.hydrogenaudio.org/forums/index....pe=post&id=1741


At 50hz the L-R signal is far, far noisier. Some correlation still exists but the SNR is effectively 0db at this point. Nevertheless, signals are still identifiable at 0db SNR, and the overall LP level is comparable with the CD. Below 37hz the rumble dominates the signal.
http://www.hydrogenaudio.org/forums/index....pe=post&id=1740


Finally, here's a 0-40hz response plot of L-R.
http://www.hydrogenaudio.org/forums/index....pe=post&id=1739

An analysis of turntable rumble

Reply #3
Discussion

The low frequency noise has a markedly different character between the sum (L-R) and difference (L+R) channels. L+R seems to not require a filter down to at least 20hz; L-R really needs one below 34hz, and the SNR even at that frequency is so high that filtering at 50hz may be recommended.

Both channels show large harmonic series spaced at 0.55hz, which corresponds to the 33.33rpm rotation rate of the record; the L+R channel shows these harmonics to around 20hz and the L-R channel shows them to 26hz. A comb filter may be useful in eliminating this.

More specifically, I was able to count the spikes and their frequencies to obtain a turntable speed of about 33.43 rpm, or ~0.3% fast, which agrees well with the other evidence for the table being about 0.4% fast.

As far as recommended rumble filter settings: I personally am considering a 5th-order Bessel highpass filter with a corner frequency between 16-20hz for the L+R channel, and a 6th-order Bessel highpass filter with a corner frequency between 34-40hz for the L-R channel.  I may also experiment with an adaptive comb filter to better remove the highly harmonic rumble components.

Please note that this can only be considered a single data point on what kind of rumble to expect; it cannot be used as a lower or upper bound. A better turntable configuration, combined with a better quality pressing (the LP has a warp on the order of a millimeter) may yield a lower rumble spectrum. OTOH, a severely warped disc will may show a phenomenally large L-R rumble spectrum, and more filtering may be necessary. I don't have a good example of a warped LP coupled with a CD version of the same LP to test this out.



Open issues

The FLAC version has low level harmonics from at least 18hz to 30hz, and possibly as wide as 8hz to 60hz, with a harmonic spacing of 0.75hz. This corresponds to a (possible) record playing speed of 45rpm. Coincidence? Digital effect? Crazy mastering decision? I don't know what to think.

The exact level difference, or whether or not that difference changes between L+R and L-R channels, has not been clearly identified. It may be due to improper cartridge alignment, or perhaps it may be an inherent flaw in the cartridge itself, or it's just an illusion resulting from interactions between the frequency spectrum and the way I'm plotting it.

An analysis of turntable rumble

Reply #4
I'll cut right to the chase: 60Hz used to be a common cutoff for rumble.  If you have a serious problem with rumble in your turntable I would look for another.  Your point that your cartridge might be misaligned could possibly make the rumble worse, but could not cause it.

What I would do is not use the turntable.  They were superceded for a really good reason: CD's are better.  No rumble, no hiss, no groove distortion, no print-through from adjoining grooves, no dirt and grit problems, no high-end distortion or attenuation at the inner grooves of the LP; there were many, many drawbacks to LP's.  But, they were all we had at the time.  Yes, their brand of distortion sounds "warm."  Cool, if you like distortion.   
Nov schmoz kapop.

An analysis of turntable rumble

Reply #5
Even 20-some years into the CD age, there are still LPs or LP mixes that haven't been rereleased on CD.  For those, you will want to archive them to digital, and unfortunately a turntable's the only way to do it (though I seem to recall a JAES article that used something like confocal or atomic force microscopy to reconstruct wax cylinder and vinyl grooves...)

An analysis of turntable rumble

Reply #6
There are a few other good reasons to still keep a good vinyl system. For me, the overriding reason is that my parents gave me 200+ LPs and replacing them with CDs would be prohibitively expensive. Furthermore, it is arguable that the lower amounts of compression applied to vintage LPs as opposed to modern CDs may make them sound objectively better, although I have little objective data to back that up (yet). Decent vinyl is also dirt cheap - $7 for a VG+ is the norm, and $1/disc is not entirely uncommon (although new releases can still run you $20). Finally, a lot of *very* good electronic music only exists on LP.

An analysis of turntable rumble

Reply #7
yeah.  and nothing whiles away a quiet weekend like going to an op-shop and getting the cheesiest records you can find for 50c and less, then playing them and lauging your arse off

op-shop records are so fun.

a friend did something extremely cool at a party the other day - got 16 LP covers and turned them into a "twister" mat.  with a spinner made of a CD with the record labels written on it, you can go "right foot on boy george" or "left hand on hammond-a-gogo" all night.

oh, and my limited vinyl collection shows much less compression than their CD counterparts.  good examples are Portishead (self titled) and Beastie Boys - Ill Communication.

interesting stuff with the rumble btw - my amp has a little "subsonic filter" button on it which i believe is a highpass that kicks in at 18Hz, and rolls off at 18dB per octave.  but i usually don't hear the rumble anyway - it's more for the sake of the speakers, as often i miss the line-out on my computer and feed the amp with solid DC current from the line-in... once it even made my amp's protection circuit kick in (i'm glad that it works!)

An analysis of turntable rumble

Reply #8
Quote
Finally, a lot of *very* good electronic music only exists on LP.


And MP3.  Don't forget www.AnimaReco....  Never mind ;-)



:-D

An analysis of turntable rumble

Reply #9
Axon,  I am finding these test results very illuminating !  Just learning that you surmised that your tone arm resonance is probably the 8 hz peak you see in the plot,  has helped me immensely.  Now I'm wondering,  is the fundamental resonance of most tone arms somewhere below 20 hz?  If so,  that seems like good news.

I  totally expected the slightly more prominent leading edge transients from the digital version,  VS the vinyl sourced one.  But the rest is quite a surprise.  It's my understanding that Vinyl Mastering engineers will use EQ while cutting the mother disc,  not to editorialize,  but to compensate for the limitations of the vinyl medium itself.  And some use their ears for this task, "goosing" things in such a way that it may not measure the same as the source master,  but with the intent of it sounding correct when it's played back on a "real world" turntable.  Which in the end,  cannot help but be a form of editorializing. Like tuning an SU carburetor on a Jag,  there is quite a bit of skill and craft to the process that isn't learned overnight.  And each cutting engineer eventually a develops a "sound" to his work that production engineers will seek out.  Perhaps this is not true anymore,  and cutting vinyl has improved signifigantly.  What year was the LP pressed?

***THUSLY,  I was expecting far more radical swings in the EQ between the digital and the Vinyl,  but I have yet to do such a comparison myself.  Can you please speak to any noticeable differences you heard listening to the version that came from vinyl,  compared to the one that didn't? If so,  are these differences dramatic?

I've been slowly assembling a signal chain to transfer some LPs myself. The last piece missing, is to get a decent record cleaning machine before I really get busy.  In general,  at record purchase time for years now,  I  favor any music recorded and released on vinyl that was pressed before digital recording showed up.  This means it has never touched the numbers;; analogue.  Post 1990,  I'm happy with the CD,  figuring it has always been all numbers,  but of course many recordings still use analog tape during production. 

If I'm looking at the plots right,  both mediums  are remarkably close to each other in so many ways,  even with your qualification of only modest  mechanical set up.  Perhaps I should toss my turntable and start over !! // * > * \\


An analysis of turntable rumble

Reply #11
Quote
The FLAC version has low level harmonics from at least 18hz to 30hz, and possibly as wide as 8hz to 60hz, with a harmonic spacing of 0.75hz. This corresponds to a (possible) record playing speed of 45rpm. Coincidence? Digital effect? Crazy mastering decision? I don't know what to think.

Air conditioning. Building resonances. I've measured air conditioning many times that way.

Just like I've measured 15735 Hz peaks from monitors.

Of course, my answer is not definitive, but I have measured the background in my office in several locations, and that kind of spectrum (different peaks, different base frequencies, of course) appears very commonly in buildings.
Quote
The exact level difference, or whether or not that difference changes between L+R and L-R channels, has not been clearly identified. It may be due to improper cartridge alignment, or perhaps it may be an inherent flaw in the cartridge itself, or it's just an illusion resulting from interactions between the frequency spectrum and the way I'm plotting it.
[a href="index.php?act=findpost&pid=330924"][{POST_SNAPBACK}][/a]


Mastering for vinyl is nothing like transparent, it involves controlling L-R signals because they throw (literally) the stylus out of the groove if they are too big, or too fast.  There is also a bunch of frequency shaping, etc, applied so that the whole process of master->reverse-master->stamper actually comes out only a fair distance from good. (without it, things would be worse)  Finding differences in levels between L+R and L-R in CD vs. vinyl would not surprise me even a tiny bit. I'd be more surprised if there weren't large differences.

Of course, then there's the issue of pivot points in stylii, that create interaction between the L+R and L-R  (or left and right, take your choice, but the real mechanics is L+R and L-R) having to do with pivot points, sensing points, and stylus-bar resonances.  That's mostly at high frequencies, though.
-----
J. D. (jj) Johnston

An analysis of turntable rumble

Reply #12
Quote
Axon,  I am finding these test results very illuminating !  Just learning that you surmised that your tone arm resonance is probably the 8 hz peak you see in the plot,   has helped me immensely.  Now I'm wondering,   is the fundamental resonance of most tone arms somewhere below 20 hz?  If so,  that seems like good news.
The common wisdom is that tonearm resonance should be 8-15hz.
Quote
What year was the LP pressed?
2005.
Quote
***THUSLY,  I was expecting far more radical swings in the EQ between the digital and the Vinyl,  but I have yet to do such a comparison myself.  Can you please speak to any noticeable differences you heard listening to the version that came from vinyl,  compared to the one that didn't? If so,  are these differences dramatic?
I really haven't listened too critically to the LP version yet, but the differences are a little puzzling. For the most part, the mixes are indistinguishable. However there are one or two instrument tracks that were far, far more prominent in the LP version than the FLAC version. I usually listen to the LP with a different set of headphones compared to FLAC, so quite a few unknowns are currently present.
Quote
I've been slowly assembling a signal chain to transfer some LPs myself. The last piece missing, is to get a decent record cleaning machine before I really get busy.
FWIW, I just scored a KAB EV-1 record cleaner from kabusa.com for $160. This is for all practical purposes a Nitty Gritty 1.5 and so is probably the cheapest RCM available, new or used.

An analysis of turntable rumble

Reply #13
Quote
Air conditioning. Building resonances. I've measured air conditioning many times that way.
BINGO! My turnable is on the 2nd floor of my house, on a windowsill. The core HVAC is mounted in a closet about 15 feet away  I really should try comparing waveforms with/without the AC running. I suspect what I'll wind up doing is just turning it off while recording. Turntable isolation is a dreadfully expensive and snake-oil-dominated enterprise. Mana, anybody?
Quote
Mastering for vinyl is nothing like transparent, it involves controlling L-R signals because they throw (literally) the stylus out of the groove if they are too big, or too fast.  There is also a bunch of frequency shaping, etc, applied so that the whole process of master->reverse-master->stamper actually comes out only a fair distance from good. (without it, things would be worse)  Finding differences in levels between L+R and L-R in CD vs. vinyl would not surprise me even a tiny bit. I'd be more surprised if there weren't large differences.
Indeed, one of the more shocking things about this pressing was just how tame it is compared to the FLAC version. No huge lowpasses, no major frequency deviations in the audible band.

I guess those Brits sure know how to cut their laquer. Or.... something like that.

An analysis of turntable rumble

Reply #14
Quote
What's wrong with numbers?

The whole 'analog vs. digital' thing is rife with bad preconceptions.  Here's a great pair of essays
concerning how digital 'analog' can be, specifically as it relates to LP:


http://www.st-andrews.ac.uk/~www_pa/Scots_...rt12/page1.html

http://www.st-andrews.ac.uk/~www_pa/Scots_...rt12/page2.html
[a href="index.php?act=findpost&pid=331616"][{POST_SNAPBACK}][/a]

Actually,  I don't hate digital audio.  On this one,  I'm on the fence.  I don't hate analog tape or vinyl either.  I like each medium for what it does best,  and hate each one for what it does worst. 

***Digital is GREAT for the Test Axon has done here. Accuracy. I think this shows both a very decent phono signal chain and a decent cutting engineer who mastered the LP version as well.  There is bandwidth out to the extremes.  Real 20-30 hz !None of the very moderate peaks and valleys in the EQ is very alarming.  Bravo !!


The transition from all analog recording over to digital in the 80's was nasty.  Old Catalog was slam dunked over to CD's,  and often transfer FLAT without even much listening,  because the medium was judged intellectually to be "Perfect".  The early AD converters were far cruder than today.  CD's were assembled in 16 bit,  and released in 16 bit,  but any DSP along the way took a little of the bit depth resolution away. (Now everyday edits at 24 bit or higher,  so there is some headroom before you get to the 16-bit CD)  This meant the first wave of CD's seemed bright,  and the bottom a little lean.  Audiophiles bitched about CD's (I was not one of them at the time...but I had a truly sad turntable) . The entire methodology to the playback monitors had to change,  and slowly,  everybody learned how to EQ and adopt to the new medium.

Why is analog  tape so different?  Retaining anything above 10khz is tough,  but getting a rock solid 200hz  a piece of cake. And a built in instantanious euphonic peak limiter, if you hit the tape hard.  Why is digital do different?  So little coloration to the medium itself,  and no built in compression.  Welcome to the brick wall. Just unnerving at first,  that's all.

The other coin,  was that LPs were still on the shelves,  right along with the new medium,  CD's.  But many eager producers began to record the multi-track masters in digital.  DIRE STRAITS "The Sultans of Swing",  seemed a watershed moment to me,  because suddenly it was proven to one and all,  that digital didn't have to be "Brittle"  or "Hard" in the right hands.  And then transfering from this digital master to LP had it's own set of headaches to figure out.  I was still buying Vinyl as well,  and whenever I read a record label that proudly proclaimed "Recorded to digital",  the sound often disapointed me,  with some sort of weird "glare" in the upper octaves.  So I decided to split it down the middle,  pre and post digital age,  and buy mostly old classic rock and jazz catalog on LPs,  and new stuff to CD

But as I dabble in hip hop and NEW just pressed jazz vinyl releases, they impress me to no end.    It seems a digital master pressed to vinyl is a wonderful combination !!  The cutting engineers are in the ZONE.  I've had my turntable out of the closet for a year,  and CD's just bore me.  I'm not saying that CD's sound bad.......it's just that.........I'm just not that motivated to buy them any more !!.  I don't need the conveniance,  because I listen sitting with my face in front of the speakers,  and not in my car,  or in headphones,  etc. 


***Does Axom's test here prove that vinyl is better pressed today than ever before?  It would not surprise me.  I don't know,  what do you think?  His turntable seems fine !!  I'm trying to decide looking at the data how far off of linear,  is enough to be alarmed, or thrilled.  As much as I love the sound of vinyl,  I've never  really considered it to be clean or LINEAR,  just euphonic.    Was it always this good, and I just never bought a good turntable to hear it untill a few years ago?

An analysis of turntable rumble

Reply #15
Addendum: Inner groove measurements
I reanalyzed the needle drop on its last 3 million samples, against the equivalent 2nd track "Ipacial Section". The results turned out to be far different. (This time I'll actually have working images.)

L+R

Wideband view, 0-40hz.

The crossover level where the LP rumble dominates is far lower in frequency, more like around 12hz.This is much quieter than before. A lot of this is expected due to the turntable clamp flattening out the disc a bit.  There's also a mild lowpass effect evident around 20hz, as if a rumble filter was in fact being applied during pressing. Resonance is still 8hz.

8-20hz.

The broadband rumble noise starts ar around 14hz but it only starts being an issue around 11hz or so. Given the 33rpm spikes in the outer groove plot, you'd expect the spikes here to be about the same thing. You'd be wrong.

6-9hz.

These spikes are, in fact, correlated with the FLAC signal and are thus real information! Moreover, these spikes are occurring right on top of the tonearm resonance of 8hz.

The frequency shift associated with the LP signal comes out to be 0.8% faster than normal speed - this is an expected effect of DC servo belt-driven turntables, that the inner tracks play faster than the outer tracks.

L-R

Wideband view, 0-50hz.

Again, the LP signal is a bit cleaner. Rumble starts to dominate at 26hz.

20-40hz.


Discussion
There are no particularly easy and good filters that handle both the outer groove and the inner groove of this single album without filtering out real subsonic information. I would thus opt to be as conservative as possible in my filtering, and only focus on attenuating the tonearm resonance and lower frequencies - and only then simply taming them to the same level as the CD, and no lower.

I guess the moral of this story is that I probably need a better turntable if I want to eliminate rumble.

An analysis of turntable rumble

Reply #16
WaveFiler, as I mentioned in my first post, it's hard to generalize from this one disc, recording, turntable or pressing. Vinyl mastering, like tube construction, is a black art that is gradually becoming extinct, and mistakes can still be commonly made during mastering due to inexperience. That said, the usage of high quality digital masters nowadays instead of tape (especially 15ips tape) can only be a good thing.

 

An analysis of turntable rumble

Reply #17
Quote
yeah.  and nothing whiles away a quiet weekend like going to an op-shop and getting the cheesiest records you can find for 50c and less, then playing them and lauging your arse off

op-shop records are so fun.
[a href="index.php?act=findpost&pid=331200"][{POST_SNAPBACK}][/a]


Oh thank goodness - I thought it was just me!

Strangely, when I say to my wife "listen to how cheesy this record is", she just says "Oh great, you've bought another crap record!"

Cheers,
David.