Turntable Details

Audio-Technica AT-LP2D-USB
Street Price: $100 (US)
AT3600L Moving Magnet Cartridge
Belt Drive
Fully Automatic
Wow and Flutter: <0.25% (WRMS)
S/N: 50dB unweighted
Platter: Aluminum
Tracking force: 4.54g (Measured personally)
More in turntable guide

Photos

The 1 kHz test sample and Dum Diddly (LP) don't work for me

Also the links for the Gorillaz and Black Eyed Peas samples are all kinds of messed up.

I don't have time for a full analysis, but I did just do a quickie frequency response comparison of the Aja samples. Read this graph like a graphical equalizer. Window length is 300ms, overlap is 66%, Blackman-Harris window, L+R and L-R channels (black is L+R, ie the mono signal, red is L-R aka stereo).

Unfortunately I have no idea on how to link to an attachment anymore (mods?), so:

http://www.hydrogenaudio.org/forums/index....showtopic=69618

Off the cuff, the bass boost is pretty shocking, to the point that I'm not exactly sure if it could even be caused by a tonearm resonance problem - it might be a mastering change on the LP. It's around 30db@50hz ish. I'd need a far longer sample (like several minutes) to be sure one way or another. (PM me if you're interested in that sort of thing.) The rising L-R response at the upper treble is probably going to be tracing distortion. I'd guess the drop around 200hz and the generally attenuated L-R signal is a mastering change, but that's always hard to gauge.

Hey: I looked at one of your close-up pics and that reminded me. While you're at it, could you also be sure to clean the stylus with a Magic Eraser before each play, if you can? Just lightly press the Eraser to the stylus in an up/down motion to get all the gunk off. It's the cleaning method du jour nowadays and it's pretty cheap. It really can improve some of the results.

If you don't have a ME then a useful alternative is a Q-Tip soaked in rubbing alcohol, but you have to be sure not to get the cantilever too wet as it can wick up to the suspension, and some stylus adhesives do not like getting wet (but that's usually a problem with the high-end carts more than the entry-level ones).

Q-tips were too scary for me for stylus cleaning back in the day. I used to use a small, closely-packed brush pad, dosed with alcohol, that I could run the stylus through (like moving a spike through tall grass). I think it was made by the same folks who made the Discwasher fluid...I remember it has a classy wooden handle.

Anyway, a small artist's or touch-up paintbrush could probably serve as well or better today. Every once in awhile I would also check out the stylus under a microscope/loupe -- amazing how much gunk it could pick up from just one play, depending on the disc. (FWIW this was/is a Shure V15TypeV MR, at the recommended tracking force)

I only listened to the Steely Dan clips so far, and it made me not miss vinyl. ;> (To be fair, I think the LP version might sound better with some level-matching to the CD version )

Bad news, krab... they're already level-matched.

I apologize for not double checking the links. It was approaching midnight as I was wrapping up the post. They are fixed now so listen away!



That was my impression too just by listening to the Gorillaz LP sample vs. the CD. These should be the same masters since the album was put out in 2005.

I will PM you the link to the full Steely Dan song since I know you own both the CD and LP.



Yes, I noticed that piece of debris when looking at the close-up of the cartridge. It's not there anymore. I will take your suggestion and pick up a ME marker and apply it before each play.



These are all new LPs, so there should be minimal gunk.

That reminds me, though, I still have one more music sample- a dusty, scratched Fleetwood Mac LP. I won't touch the stylus to this record until I finish the test LP tracks.



The original plan was to keep match the levels as best as possible. Then I discovered that the USB turntables I'm testing don't allow me to set the levels!

I could normalize after the fact but I don't want to give the impression that the samples have been tampered with.

I only own Aja on LP, not CD. On The Run is a better sample for me.

Magic Eraser is a sponge, not a marker - it's going to be in the cleaning aisle.

Don't normalize - I can normalize the plots myself so that the response at 1khz is set to 0db.

Wow.

Well ultimately, that's OK -- I have the CD, not the LP.

EDIT: -- hold on -- the Dark Side clips yield the same replaygain value (+1.24 dB) ..but the Dan clips do not

Home at Last LP -0.22 dB
Home at Last CD -2.88 dB

It's not unusual for LP versions to be mastered differently than CD versions. Even if the same master tape was fed into the cutting process, EQ can be applied during cutting (and then if that EQ'd feed is saved, that become the 'LP production master' for subsequent cuttings). Or two masters may be made from the mixdown master -- one for modern CD (e.g. highly dynamically compressed) and one for LP.


.

Vinylphiles IME commonly recommend a thorough vaccuum/clean of even brand-new LPs. Maybe it's just one of their rituals, but isn't there supposed to be some kind of 'releasing agent' on new LPs (artifact of the stamping process)?



That's OK, if I want to I can apply replaygain at the playback end, without altering the actual data.

New pics up for Pink Floyd and Steely Dan samples, check the upload thread for the images and details. I'd go this far with the plots:

• Between 1k-9k the response rolls off by 6-9db. This is a huge, huge difference.
• The response at 20k is about 15-20db below at 1k. Not as bad, but still crummy in the grand scheme of things.
• The tonearm resonance is now pretty concretely in the 20-30hz range, which is not as bad as it was looking before, but still not good. It should be in the 8-15hz range, and on many home entertainment systems this may cause audible problems, especially during live playback of vinyl (although many people will like it because it provides a bass boost)
• The Pink Floyd needledrop has a large noise peak at 19khz, but it doesn't exist in the Steely Dan recording. So it might be cut into the record. Probably not audible to begin with.
• L-R narrow peak around 21k - this is in both recordings and is probably electrical noise in the turntable, but is certainly inaudible
• Large (5db) response discontinuity in both tracks at 9.5khz, probably a resonance of some kind? Maybe in the cartridge? Probably audible.
• ~3db bump at 14k, not audible for most people
• Large-to-huge peak just after 20k followed by a huge rolloff - possibly the stylus tip-vinyl resonance, or else it could be the LCR resonance of the preamp

It's hard to derive any more from these plots without probably equivalent masters and/or the test record sweeps. Plus, I've already spent way too much time on this on a weekday. The plots look really ragged overall, but it's hard to gauge how much of this is the turntable and how much of it is the mastering, especially with the raggedness goes in opposite directions for different records.

Mold release agents are generally added to the vinyl formulation directly, and I remember a quote (Stan Ricker?) that actually spraying MRA onto the presses almost never happens. That said, there are lots of stories nowadays about brand-new record with chaff and crap inside the seal.

I'd also like to point out that even though an analysis of the frequency sweeps is going to give an authoritative picture of the frequency response of the turntable, the analyses of the recorded music show things that may not appear on the frequency sweeps, such as the impact of stylus shape on the response. They can help explain what you hear on the vinyl, while the sweeps are much less direct in their interpretation.

However, it's important not to read too deeply into it. Excessive distortion may only show up as heightened treble response, and may mask potentially unrelated (but major) linear distortion issues. A certain amount of interpretation is always required.

Yeah I was confused. nm.

Got ya.

You can tell how much cleaning I do around the house. When I brought a couple boxes home, the wife tells me we already have a bunch of boxes in the garage!



I seem to have misplaced the LP attachment to my vacuum cleaner.

Seriously, though, I looked around at acquiring vacuum system. I'm not a handy person, so the DIY route is not a viable option. Buying an assembled system means forgoing another USB turntable or two.

I did pick up the old, reliable Discwasher system, though. I'll clean the test LP, ME the stylus and then post another sample of the silent groove.





The graphs look great. Thanks for that.



I'm working on test record samples after this post.

That brings up a good question: Which specific tracks make for the best samples? You can read the track list and details for Ultimate and Hi-Fi News are online.

Here's what I'm considering:

Ultimate

• Side A, Track 1: 1kHz tone (Done)
• Side A, Track 10: Wow & Flutter test Signal
• Side B, Track 1: Anti-skating Test, 315Hz sweep to +12dB
• Side B, Track 4: 1kHz vertical, out-of-phase signal
• Side B, Track 5: 1kHz to 10Hz sweep at -20dB, vertical
• Side B, Track 6: Silent groove (will re-do after cleaning)

Hi-Fi News

• Side B, Track 7: 20Hz to 20,000Hz sweep

Will that do it or are there any other important tracks? Are there any I shouldn't bother with?

Take that sh*t back and surreptitiously suggest to your wife that y'all really, really need a $30 pressurized steam cleaner from Walgreens. Look on several other vinyl forums (Audiogon, Audio Circles, Vinyl Asylum etc) on suggestions on how to use it. Combine that with some decent quality microfiber cloths and you'll probably have RCM-quality cleaning right there, just with some more legwork. Without spending any money on vinyl-specific equipment, too! Use a $10 carbon fiber brush to clean off superficial dust.



First of all, IMHO, don't clean the test records unless you can see specific issues with them that require cleaning. You're putting a lot of trust into the cleaner to not negatively affect the test record results, and there are several (admittedly theoretical and unobserved) ways that can happen. Brush scratches record, dirt embedded in brush scratches record, dirt in brush is left on record, cleaning agent leaves residue on record, cleaning agent + brush moves the dirt around without picking it up, etc. For most other records, there is a good enough potential benefit of reduced noise that cleaning makes sense, but most test records are very well manufactured and maintained, and are also more expensive. That said, it's not the end of the world if you clean the records - no results are compromised - but if we hear a hiss in one of the channels on the silent bands and varies based on the position of the record, I'm going to wonder.

Second, just post the entire needledrops of Ultimate and HFNRR. I can use almost all of the former and most of the latter. And when you're done with all the tables, please re-record each test record with the first turntable you played it on, to establish how much the results changed as a result of all the plays.

Good thing you were online and stopped me before I did anything stupid!

For the record, I did clean Side A of DSOTM. It was the most crackly of the bunch and I wanted to see if it would make a difference. It sounds exactly the same. The crackles and pops are still there.

Since all the records I'm using in this test are brand new, I will refrain from cleaning any of them and continue simply handling them carefully. I'll touch the Magic Eraser to the stylus a few times before each needledrop.



Will do, with a caveat: We talking about a half GB or more in FLAC files. I will have to monitor my server's bandwidth usage closely.

If it becomes too popular, I may need to make it available on a per request basis (you PM me, I'll give you the URL). Though based on the downloads I've made available so far, we're a long way from that happening.



You got it!

Don't deal with the server - put up a tracker on Pirate Bay and I'll download it from there. Mods, put the guns down! IANAL, but I couldn't imagine that test record transcriptions would be considered a copyright infringement. They are absolutely not a substitute for the original article, you are adding vast amounts of information by transcribing it on your turntable, it only adds to the demand for test records by defining a larger market for their analysis, etc. BitTorrent is absolutely the correct distribution system for this sort of thing.

But of course, I've never done this before

The Ultimate Test LP FLACs are up. See the original post.

The Hi-Fi News is proving to be a bit trickier. A circular groove follows each track. It forces me to lift the tonearm and scoot it slightly toward the center of the album before I can play the next track. So, when you hear hard breaks in the two HFN FLACs, that's why.

These files are much smaller than my guesstimate for two reasons: The entire album is only 21 minutes and most tracks contain very little audio data- often just a test tone. Side A is 37.2MB and side B is 49.4MB.

Enjoy! I expect to have HFN up sometime tomorrow.

Yeah that was a sucky mastering decision with HFNRR that's been discussed here before. Generally the amount of content on a test record is far less than on a real record to maximize sound quality and make it easier to position the needle on the bands.

I'm downloading the Ultimate FLACs now but I can't promise anything earlier than late next week for results. I'm swamped.

I ran the Wow & Flutter test tone through Spectrogram and it exceeded my expectations in revealing deviations from the 3,150Hz test tone. Or maybe I underestimated just how unsteady the Audio-Technica turntable's playback would be.

Either way, I got some great captures! Click to zoom in.

Those are pretty, but they're not remotely resolving enough. 99% of that wave might be due to the pressing being off-center - that will be easy to spot if all the other turntables show the same thing. Most of the width of the spectrogram is due to FM sidebands which distract from the carrier wave. Measurable flutter (that we care about) can extend well into the 200-300hz range and beyond. Your frequency and time resolution should both be at least a hundred times more greater to get useful results, and that level of detail is simply impossible with spectrograms.

The reference for interpreting wow/flutter is Poul Ladegaard's legendary 1977 paper on mechanical resonances. Specifically look at Fig. 18 on page 7. Those are the sorts of charts you ought to be displaying/interpreting.

The "right" thing to do here is demodulation of the 3150hz tone into a wow/flutter signal followed by DC removal and amplitude spectrum. Don't spend much time playing with the spectrograms for any kind of analytical use - they're good for visualization but not for analysis. I've been meaning to cook up a good quadrature demodulation program for some time, and now is the perfect time.

The Hi-Fi News FLACs are up.


It's never as simple as I hope!



This is quickly getting beyond my limited aptitude for signal analysis. It's really interesting reading, though.



As always, thanks for your contributions and guidance. I couldn't have done it without you.

The last sample is up: Fleetwood Mac's Rhiannon from an abused LP.

The cut starts with a skip. We'll see how well the other two turntables handle it.

The sample ends during the fade-out to help make the clicks and pops more apparent.

I told you the LP was dusty...

Fun times. I told you to get a carbon brush

That's almost guaranteed to be an unfair recording of the AT, then, because it would up cleaning the record so the rest of the tables don't have to. A lot of crud develops on/around the contact points of the stylus that drastically affects the sound. Try rerecording it and see how different it is.

Good point.

Well, fair is fair. I re-recorded the sample following these steps:

1. Removed the debris seen in the photo
2. Used Magic Eraser on the stylus
3. Played Rhiannon twice to make sure the stylus has done all the cleaning it's going to do
4. Used ME again
5. Recorded Rhiannon for the new sample

The first post has the new sample. Here are the two samples side-by-side:

• Rhiannon (Dirty version)
• Rhiannon (Cleaner version)

Just a status update - I got most of the demodulation working over the weekend but it has a lot of warts I need to address. Still, the results are shockingly beautiful so far. It will be most interesting to compare them against other tables.

I haven't had any time to work on this since then - I've been a little distracted by the whole Pleasurize thing, but I'm spending a pretty minimal amount of time on that too.

If this helps, I'm waiting with bated breath to see how your little demodulator works. I'm not nearly as excited about dynamic range. What would be really cool is if it would put out a .wav files of both AM and FM so that each could be analyzed separately.

Thanks for the cheerleading. Demodulating AM is trivial, I just dropped it in and it took 5 minutes. Now, getting the AM demodulation to sync up with the FM will cost you...

I'll post a new thread sometime in the next few days with some preliminary findings, because the software will clearly find use beyond this project. Some of the results may shock the reader. Viewer discretion is advised.

And if I haven't said it enough, thanks for lending your time and expertise to this project.



Sounds great. I can't wait to see it.

I'm needledropping with the Ion TTUSB05 as I write and should have the entire set of samples this Saturday. Lots more data to run through your software! It will be interesting to see the difference (or not) between the two turntables.

Again, waiting with bated breath. ;-)

What you find may shock some, but I think I've been prepared by years of FFT analysis that suggests that massive FM distortion is part and parcel of the usual LP listening experience. It's ironic that ragazines that IMO push vinyl so hard like SP, have been obsessing all these years over the relatively small FM distortion in digital players.

Some suggest that the ear's reduced sensitivity to LF FM distortion is the reason why most tolerate it, but I'm beginning to suspect that years of forced ear training to simply ignore it is also a strong cause.

Primary problem IME - off-center holes, followed by record warps.

BTW for whatever reason, I just can't stand lots of FM distortion. I was sooooo happy when the CD came out, because that was pretty much the cure. I had previously discovered that I really liked music with reduced FM by listening to live music, early digital recordings, and good tape recordings, not that the latter are exactly free of it.

Post is up:

http://www.hydrogenaudio.org/forums/index....showtopic=69961

Please split off general discussions on wow/flutter over to there, and we'll continue focusing on sample analysis here.

The exact nature of the wow/flutter analysis to use here isn't quite obvious to me - is it necessary to post AM spectra? How high a frequency limit is necessary? - and I have not yet implemented the weighting filter, so the truly final wow/flutter results for the LP2D will probably wait until samples from other tables come in, and the measurements that best illustrate the differences between them can be used.

I always wonder what's in the magic erasers that makes them clean well. I'd be kind of nervous about using one on a stylus for that reason. With my luck, the cantilever would rot away or something.

I recently picked up a refurbed AT-LP2D-USB and it works quite well. (I was unable to get the included Cakewalk Pyro to work, but Audacity works fine so I don't care about Cakewalk.)

I've still got an old carbon anti-static brush, so I'm all set to transfer some of those old LPs to CD.

It's nice when a product does what it's supposed to and doesn't cost an arm and a leg.

Strange- I didn't get a reply notification when tiptoe replied. In any case...



I'm curious where you had trouble getting Cakewalk Audio Creator to work.

I reviewed Cakewalk for the USB turntable guide and it did take some time to learn. Two big stumbling points:

1. You need to press the button on the bottom left to hear the turntable audio.
2. After you hit the record button, you need to hit the pause button to start recording.

I produced an illustration (below) showing the steps to recording but never published it (until now).