Reviewing the Furutec Demag

[amirm]

The tracks would seem quieter, like the volume had been turned down.

Following up on this, I ran a simple test:

I first took the after track and copied its Left track to right so now it is a mono 2-channel signal. Played it on my headphones. As I expected, I got a true mono image in the center of my head (confirming my PC is playing back back at equal levels).

I then took the same track and merged it with the before track creating a new mono 2-channel signal. This time, one channel is before and the identical channel after. I then manually sync'ed both tracks by cutting out extra leader. Playing it shows a clear shift to one side.

So there is a clear level shift. I did this blindly and just now looked up which one is louder and it is TrackX or before treatment.

Playing these tracks now one channel at a time shows much better dynamics and impact of course in favor of the before track due to change in level and maybe fidelity. Note that this was a sighted test so the above fact may have impacted my impression. Still, there is no denying that level dropped post treatment.

Next would be to try the same thing again on the previous tries without treatment.

 

[garylkoh]

Following up on this, I ran a simple test:

I first took the after track and copied its Left track to right so now it is a mono 2-channel signal. Played it on my headphones. As I expected, I got a true mono image in the center of my head (confirming my PC is playing back back at equal levels).

I then took the same track and merged it with the before track creating a new mono 2-channel signal. This time, one channel is before and the identical channel after. I then manually sync'ed both tracks by cutting out extra leader. Playing it shows a clear shift to one side.

So there is a clear level shift. I did this blindly and just now looked up which one is louder and it is TrackX or before treatment.

Playing these tracks now one channel at a time shows much better dynamics and impact of course in favor of the before track due to change in level and maybe fidelity. Note that this was a sighted test so the above fact may have impacted my impression. Still, there is no denying that level dropped post treatment.

Next would be to try the same thing again on the previous tries without treatment.

Fascinating!! I would never have thought of this - or have the skills to implement it. Thanks, Amir. You might want to try the same thing on Cut2 and Cut3, and then Cut 2 and Cut4. Cut1 was the first play of a brand new LP, and it might be clearing some manufacturing junk out of the groove. I had to clean the stylus before playing Cut2.

I do have clear vinyl and black vinyl copies of the same thing. MikeL was kind enough to loan me two "reviewer sets". Unfortunately, I have not had time to even take them out of the bag let alone look at them.

 

[amirm]

Confirmed the level shift between track 2 and 3!

Unless there is a test problem with how these are captured, I think we have very conclusive data that levels do shift and shift substantially enough between play to play. Such shift could change one's perception of the fidelity.

 

[garylkoh]

Confirmed the level shift between track 2 and 3!

Unless there is a test problem with how these are captured, I think we have very conclusive data that levels do shift and shift substantially enough between play to play. Such shift could change one's perception of the fidelity.

Between Cut2 and Cut3 - all I did was lift and move the arm. I think that we now have conclusive evidence that levels do shift substantially, and this shift large enough that it could cloud one's perception of fidelity.

This also means that the usual way that we tweak turntables - listen, change VTA/Azimuth/VTF, listen again is invalidated..... which makes it all the more important that we use proper tools to calibrate (like a test LP) to do it properly.

However - given that such play to play inconsistency is present..... is the pre- to post- demag difference less or more than the play to play inconsistency. Does play-to-play account for ALL the differences in the pre- to post- demag?

 

[Bruce B]

Confirmed the level shift between track 2 and 3!

Unless there is a test problem with how these are captured, I think we have very conclusive data that levels do shift and shift substantially enough between play to play. Such shift could change one's perception of the fidelity.

Wouldn't the level shift be because of timing variations between cuts?

 

[Ethan Winer]

I've done some analysis and added some screenshots to show everyone what is happening.

Investigative journalism at its finest. Major kudos Bruce.

--Ethan

 

[Bruce B]

I wonder how hard it would be to advance or retard the one track and then see if the center image shifts still.

Tom

This would be fine if it was happening in just one place. The timing variance starts just as soon as the needle hits the vinyl and it slowly gets worse or better as the track is being played. It's not linear either, that's the problem. It's not predictable.

 

[amirm]

Wouldn't the level shift be because of timing variations between cuts?

Let's hope not. Or else, no one would listen to LPs .

Note that I got rid of the extra lead in noise which would impact numerical summary of the tracks as you performed.

BTW, can you try to repeat my experiment? I want to make sure there was not something I did wrong.

 

[Bruce B]

Let's hope not. Or else, no one would listen to LPs .

But Amir... there are timing differences between the 4 cuts of between 1 and 3ms of each time played.

 

[garylkoh]

But Amir... there are timing differences between the 4 cuts of between 1 and 3ms of each time played.

Yeah - I tried using the Liberty Instruments Audio Diffmaker and got nonsense - then I read in the original introductory AES presentation that for it to be accurate, you need an external clock to sync the source and the recorder.

A 1ms to 3ms difference over 20 secs might not be audible, but it sure is going to mess up our investigations. Anyone know how to clock a turntable?

 

[amirm]

I am not seeing how this slight timing variation causes level shifts.

I checked it very carefully and it both speeds up and slows down over the course of the track. So correction without intelligence is impossible.

What is really strange is that I tried to get the levels equalized but could not make it so. Not sure if I am deaf in one ear or what . That's why I asked Bruce to try and listen for level shift as I did and confirm my findings before going too far.

 

[Bruce B]

What is really strange is that I tried to get the levels equalized but could not make it so. Not sure if I am deaf in one ear or what . That's why I asked Bruce to try and listen for level shift as I did and confirm my findings before going too far.

When the tracks null... the level goes down. When they don't, the level goes up... that's why the level is variable because of speed/timing going in and out of sync

 

[fas42]

I'm curious about these level difference findings ...

I've also been playing with the takes in my corner here, and getting very little difference between the X and D versions so far, I can't nail it to something significant yet. The differences I'm seeing are even less than they were with the original Ella takes posted.

A 1ms to 3ms difference over 20 secs might not be audible, but it sure is going to mess up our investigations. Anyone know how to clock a turntable?

Yes, there is a speed difference, but there is more to it than that. I can do a cut of the lead-in, time shift and speed change one version so the beginning and ends synchronise, but the differencing still gives nonsense. Why? Because nasty old wow raises its head, within the takes the speed is hunting around, back and forth; one moment in synch, the next moment out ...

The very physical nature of an LP conspires against getting a good result here: perhaps more comprehensive spectrum analysis will show up more ... ??

Frank

 

[amirm]

When the tracks null... the level goes down. When they don't, the level goes up... that's why the level is variable because of speed/timing going in and out of sync

I see where the confusion is. I am not using diff files. I am simply taking the same channel from each file and comparing them after aligning them, relying on the mono image to be in the middle if the levels are the same.

 

[amirm]

So here are the steps to repeat:

1. Take one of the files and copy Left channel and paste it on top of the right channel.

2. Play the file. The image should be in the middle since the two tracks are identical.

3. Now paste the right Left channel from another take and copy it on top of the right channel.

4. Align the two tracks.

5. Listen again and see if you get an image precisely in the middle as before.

 

[Orb]

Amir,
I think I am missing something here.
Sorry to ask but please can you explain where the Left channel and right channel has to with copying and then playing in point 2?
Is it just copied for visual representation?

I do not want to jump to assumptions so apologies asking for a more lengthy response.
Thinking way off tangent here, I am wondering at moment if the differences are also relating to L+R (lateral) and L-R (vertical) as per outlined by Paul Miller- Jim Lesurf; seems you may read Hifinews online and if so check out PM Opinions page in March 2011.

Back on to what we can do.
Also its really late so do not have the time to go over Jim's page (highly knowledgable-scientific guy) but he may had done something similar already, so including links but apologies not proofing them yet:
Vinyl Solution disc treatment which seems very similar to what we are doing but with before/after he splits it into two different channels: http://www.audiomisc.co.uk/VinySol/vinysol.html
This link shows his measurements of a cartridges and characterstics how left and right differ (not as relevent) : http://www.audiomisc.co.uk/HFN/LP4/NewLampsForOld.html

Hopefully the two links may be useful in relation to this testing process.
Thanks
Orb

Thanks
Orb

 

[amirm]

Amir,
I think I am missing something here.
Sorry to ask but please can you explain where the Left channel and right channel has to with copying and then playing in point 2?
Is it just copied for visual representation?

Whatever difference there is between these recordings, will very likely also hold true in each channel. By going to mono vs stereo, we make the evaluation much easier.

The second advantage of mono is what I mentioned which is the image should be smack in the middle.

 

[Orb]

Thanks Amir.
BTW did you look at the links, or anyone else?

Not sure how one can go mono with stereo LP recording, especially when both channels do not match, nor does the horizontal and lateral modulation (L+R) (L-R) picked up by the cartridge - it is possible but depends upon the cartridge and in their group test only one got anywhere close.
But I appreciate I may be missing something.
However, is this a good reason to use a mono recorded LP?
Could be another item to add to the list of testing
Sounds interesting, although this then bypasses one of the potential parameters impacted by the effects.

Thanks
Orb

 

[ack]

I am surprised that people are having so much trouble understanding what Amir is doing, which is very simple:

He's comparing the _left_ channels from the before- and after-demag recordings, after lining them up, and he finds that the after image is lower in amplitude - plain and simple.

To do this:

1) he first creates a new mono version pseudo recording by duplicating the left channel from the before image into the right channel (thus overwriting the original right channel) - which is the true definition of mono - and confirms it's true mono with his headphones;
2) then creates another pseudo mono recording by using the left channels from the before and after images - the before-demag left channel in the left channel of the new pseudo recording, and the after-demag left channel in the right channel of the new pseudo recording - and he perceives a shift from the smack middle as a true mono should be, i.e. one channel of the new pseudo recording is louder than the other

Don't forget - when he does this he makes sure the timings are *aligned*, and he assumes that any alleged drop in amplitude in the after image would be the same for both channels - thus, he proceeds to use the left channels for his experiment, and he could have chosen the right in the same set of tests as well.

 

[Orb]

I am surprised that people are having so much trouble understanding what Amir is doing, which is very simple:

He's comparing the _left_ channels from the before- and after-demag recordings, after lining them up, and he finds that the after image is lower in amplitude - plain and simple.

To do this:

1) he first creates a new mono version pseudo recording by duplicating the left channel from the before image into the right channel (thus overwriting the original right channel) - which is the true definition of mono - and confirms it's true mono with his headphones;
2) then creates another pseudo mono recording by using the left channels from the before and after images - the before-demag left channel in the left channel of the new pseudo recording, and the after-demag left channel in the right channel of the new pseudo recording - and he perceives a shift from the smack middle as a true mono should be, i.e. one channel of the new pseudo recording is louder than the other

Don't forget - when he does this he makes sure the timings are *aligned*, and he assumes that any alleged drop in amplitude in the after image would be the same for both channels - thus, he proceeds to use the left channels for his experiment, and he could have chosen the right in the same set of tests as well.

Ack,
I am trying to understand why and the implications of creating a psuedo mono recording.
As I mentioned the left and right channels never match due to reasons I mentioned.

I feel if wanting a mono, then a true mono LP recording should be used where those issues associated with stereo recordings do not apply (I think as its only L+R)

Anyway, I am trying to understand how creating a psuedo mono does not add complexity, a case in point is does Jim Lesurf (audio engineer and also a physics lecturer at St Andrews) go through this process in the link I provided?
Furthermore he mentions using higher pass filter as a further analysis.
From what I understand he keeps the two channels seperate for a before/after comparison.
http://www.audiomisc.co.uk/VinySol/vinysol.html

I just wonder if the psuedo mono approach is overcomplicating the issue, and I guess having problems understanding if it does not affect what we are looking for.
EDIT:
Looking at Jim Lesurf's analysis (last 2 graphs), notice how channel 1 and 2 are quite different from each other for both before and after, how do you make a psuedo mono from that behaviour without changing it?
Bear with me as I appreciate I may be missing something here.

Thanks
Orb