The Courteous Vinyl Playback Discussion

[Deleted member 50971]

If your resulting digital signal goes to -1db (or higher) you're in trouble

For this kind of thing, the "mere" 120db we have in the better ADC's is often marginal...

Get to know your cartridge and its max output under extreme circumstances (pops/clicks/scratches)

Once your gain is set based on that, you are likely to find that your actual musical peaks end up down around -80db, -90db if you are lucky...

In a perfect world we would be seeking out true 24bit resolution dacs or even better 32bit - but the state of the art isn't there yet....

Yes it is a big rabbit hole right now, although it need not be - this could all be automated into a single component, which I have been saying for more than a decade now.

More recently, Technics have demonstrated it, by releasing their TOTL Integrated with a built in digital phono stage, and a test record - you play the test record, and the phono stage "sets itself up".... then you get on with playing music. Trouble it, it is priced well above most people's budgets!

I am waiting for Technics to release the phono stage as an external component....

Now that’s different story, and that would be one cool phono preamp. I’m sure Technics (and probably others) have discussed that, and I could see one or two people sitting around the table saying… how many people are going to buy this? When a company could sell 100 DACs to 1 of those, although Technics is in the turntable business.

 

[dlaloum]

Now that’s different story, and that would be one cool phono preamp. I’m sure Technics (and probably others) have discussed that, and I could see one or two people sitting around the table saying… how many people are going to buy this? When a company could sell 100 DACs to 1 of those, although Technics is in the turntable business.

not going to be cheap.... but between $1k and $2k would be a bargain.... Under $3k - assuming top level performance - would still be good value for many people - beyond $3k and people will just look at it as an academic curiosity.

Consider that in terms of the core technical capabilities, the Puffin phono stage has all the smarts needed to do the job.... it would just need an appropriate software upgrade, and a test record - so it should be achievable and marketable at under US$1000

 

[dlaloum]

That’s extraordinary, and counter to anything I’ve seen running DSP RIAA for nearly a decade.

Depends on how well adjusted your input gain is - the clicks/pops can jump your peak to +20db

When you start out with an unknown record, not knowing whether it is a quiet or loud one, and an unknown cartridge - trying to adjust the gain to match your ADC's "window" - every db counts.

In digitising some of my records, I often struck issues - it came down to needing to do an initial play through to set levels optimally, before doing the recording... (and cleaning, cleaning, cleaning)

Or you can choose not to be a perfectionist, and accept the distortion on the occasional pop/click - they are after all, distortion themselves (the battle between "Better and best" is real!)

 

[Holmz]

If your resulting digital signal goes to -1db (or higher) you're in trouble
…

…
For this kind of thing, the "mere" 120db we have in the better ADC's is often marginal...
…

Huh?

…
Get to know your cartridge and its max output under extreme circumstances (pops/clicks/scratches)

Once your gain is set based on that, you are likely to find that your actual musical peaks end up down around -80db, -90db if you are lucky...

In a perfect world we would be seeking out true 24bit resolution dacs or even better 32bit - but the state of the art isn't there yet....

Yes it is a big rabbit hole right now, although it need not be - this could all be automated into a single component, which I have been saying for more than a decade now.

More recently, Technics have demonstrated it, by releasing their TOTL Integrated with a built in digital phono stage, and a test record - you play the test record, and the phono stage "sets itself up".... then you get on with playing music. Trouble it, it is priced well above most people's budgets!

I am waiting for Technics to release the phono stage as an external component....

Really it would be better to just blank out the clicks, or interpolate through them.
(In s/w.)

One can get the RMS value and the crest-factor.
Keeping the RMS value at -20 to -30 dB full scale should be more than enough.
One does not need to record the scratches at full scale, as they are not part of the signal.

 

[dlaloum]

One can get the RMS value and the crest-factor.
Keeping the RMS value at -20 to -30 dB full scale should be more than enough.
One does not need to record the scratches at full scale, as they are not part of the signal.

When the ADC hits its distortion zone (varies a bit with different ADC's - but anything above -1db is often troublesome) - it tends to start throwing out a bunch of hash and grabage right through the frequency bands.

You can edit out the scratch/pop/click - but if the ADC clipped, you will also have a bunch of added distortion, which will extend several ms from the actual problem peak.... and you cannot edit that out.

So you need to "absolutely" keep the signal away from the clip zone (and I absolutely admit to this being a purist perspective.... will the additional distortion artifacts be audible? don't know, have not made those experiments... I know they are present, and I know that the type of artifacts thrown out, vary depending on the architecture/type of chips used, and the associated filters etc... but they are of a level that should be audible, their short time period, may obfuscate them though).

Soft Clipping vs. Hard Clipping | Analog Devices

Can 0.1 dB really make that much difference?

www.analog.com

 

[Holmz]

When the ADC hits its distortion zone (varies a bit with different ADC's - but anything above -1db is often troublesome) - it tends to start throwing out a bunch of hash and grabage right through the frequency bands.

You can edit out the scratch/pop/click - but if the ADC clipped, you will also have a bunch of added distortion, which will extend several ms from the actual problem peak.... and you cannot edit that out.

So you need to "absolutely" keep the signal away from the clip zone (and I absolutely admit to this being a purist perspective.... will the additional distortion artifacts be audible? don't know, have not made those experiments... I know they are present, and I know that the type of artifacts thrown out, vary depending on the architecture/type of chips used, and the associated filters etc... but they are of a level that should be audible, their short time period, may obfuscate them though).

Soft Clipping vs. Hard Clipping | Analog Devices

Can 0.1 dB really make that much difference?

www.analog.com

That is all true, but it is like going out with a field recorder to get the cricket sounds for a movie background, and deciding that if there is a nuclear blast that one needs to record at a lower volume level. Or if a flightier plane flies over at Mach-x then the sonic boom need to not run the recorder beyond full scale.
Ideally though we do not insert a sonic boom, or a nuclear blast into track that is conveying a bunch crickets.

Let’s agree that the scratch is not there on purpose, and is not part of the intended musical signal.

If the signal level goes some factor above the musical maximum, then do we care if it is clipped and making harmonics?
We could just say, “that is an anomaly, so lets squelch it.”
Blanking might be another term we could use.
Then we can argue whether we blank it to zero, or interpolate between a before and after value… or something else entirely.

This is easier in a causal system with some delay. Doing it in truely real-time, would be difficult.
It’s much easier with some buffering to allow it to be identified and then, removed before output.

If it also upsets the arm, and causes a resonance that lasts milliseconds, then that is harder to deal with.

If we take this to the limit, where the scratch is infinitely narrow and infinitely high in amplitude, then we can get to the point where we need infinite resolution bits.
But we know that normal crest factor is 10-20dB, so if we take 25dB or 30dB then the 90dB audio hearing level, gets us back to a 120dB DAC.

 

[JP]

Keeping the RMS value at -20 to -30 dB full scale should be more than enough.
One does not need to record the scratches at full scale, as they are not part of the signal.

Exactly, and it’s more than enough with engineered ADCs.

 

[dlaloum]

That is all true, but it is like going out with a field recorder to get the cricket sounds for a movie background, and deciding that if there is a nuclear blast that one needs to record at a lower volume level. Or if a flightier plane flies over at Mach-x then the sonic boom need to not run the recorder beyond full scale.
Ideally though we do not insert a sonic boom, or a nuclear blast into track that is conveying a bunch crickets.

Let’s agree that the scratch is not there on purpose, and is not part of the intended musical signal.

If the signal level goes some factor above the musical maximum, then do we care if it is clipped and making harmonics?
We could just say, “that is an anomaly, so lets squelch it.”
Blanking might be another term we could use.
Then we can argue whether we blank it to zero, or interpolate between a before and after value… or something else entirely.

This is easier in a causal system with some delay. Doing it in truely real-time, would be difficult.
It’s much easier with some buffering to allow it to be identified and then, removed before output.

If it also upsets the arm, and causes a resonance that lasts milliseconds, then that is harder to deal with.

If we take this to the limit, where the scratch is infinitely narrow and infinitely high in amplitude, then we can get to the point where we need infinite resolution bits.
But we know that normal crest factor is 10-20dB, so if we take 25dB or 30dB then the 90dB audio hearing level, gets us back to a 120dB DAC.

Well, we have several different issue - first there are the mechanical ones... arm settling etc...

Not much we can do about those... I will assume that the arm and cartridge are already optimised... the arm is properly damped, and the cartridge/stylus compliance and arm effective mass are well matched....

If it is clipped and making harmonics & hash - you are not getting the desired signal - you can't squelch it at that point - removing that peak from your recording, won't remove the harmonics & hash generated by the ADC - it's now on the recording... It is not a transient consequence of the peak, and therefore removing the peak removes the noise (that would be the case if it was a DAC issue - at the replay end... and yes that can also be a problem) - it is happening at the input end of the chain - you can only "squelch" it by adjusting the gain before the ADC to ensure it doesn't clip.

I do use software to remove clicks/pops - but these solutions work best if you have captured the entire clock/pop peak - then all you have to delete is the click/pop - you don't have added hash and harmonics ... which aren't removable.

Yes click/pop removal is really only relevant with recordings where you can process it... it takes time, CPU processing time, and is not a real time process.

In real time listening, if you are concerned with getting the best possible result, you still need to keep the peak level below clipping so the ADC doesn't "misbehave" - at which point it would start sounding harsh.... nasty.

In the vast majority of cases, a 20db margin will be plenty... and especially if we have pristine records, or thoroughly cleaned ones (or wet play, etc...)

 

[Newman]

I do use software to remove clicks/pops - but these solutions work best if you have captured the entire clock/pop peak - then all you have to delete is the click/pop - you don't have added hash and harmonics ... which aren't removable.

I think the Puffin declicks 'live'... so one could simply rip/record what comes out of it with a low headroom.

Even without that, perhaps someone can rip a song from an LP that has clicks/pops, and do the following test: rip it with plenty of headroom eg your suggested 20 dB, then compare that file by ear with the same file that has been boosted 18 dB then cut 18 dB. The latter file should be 'damaged' with the added hash and harmonics that you are describing, but at the same playback volume.

If the difference is trivial to the ear, then we might as well rip vinyl without headroom for clicks and pops.

Has anyone tried this?

cheers

 

[Holmz]

Well, we have several different issue - first there are the mechanical ones... arm settling etc...

Not much we can do about those... I will assume that the arm and cartridge are already optimised... the arm is properly damped, and the cartridge/stylus compliance and arm effective mass are well matched....

If it is clipped and making harmonics & hash - you are not getting the desired signal - you can't squelch it at that point - removing that peak from your recording, won't remove the harmonics & hash generated by the ADC - it's now on the recording... It is not a transient consequence of the peak, and therefore removing the peak removes the noise (that would be the case if it was a DAC issue - at the replay end... and yes that can also be a problem) - it is happening at the input end of the chain - you can only "squelch" it by adjusting the gain before the ADC to ensure it doesn't clip.

I do use software to remove clicks/pops - but these solutions work best if you have captured the entire clock/pop peak - then all you have to delete is the click/pop - you don't have added hash and harmonics ... which aren't removable.

Yes click/pop removal is really only relevant with recordings where you can process it... it takes time, CPU processing time, and is not a real time process.

In real time listening, if you are concerned with getting the best possible result, you still need to keep the peak level below clipping so the ADC doesn't "misbehave" - at which point it would start sounding harsh.... nasty.

In the vast majority of cases, a 20db margin will be plenty... and especially if we have pristine records, or thoroughly cleaned ones (or wet play, etc...)

Are we talking about clipping from a scratch?

If the musical signal is below full scale, then the scratches hitting full scale and clipping, would only affect the sound during the scratch.

The word for this is CLIPPING, it is not “HASH”, and any harmonics are the consequence of clipping so as to form a square wave, so it is NOT harmonics in the common usage of the term. It is introducing a square wave, where one should not exist.
That only gets worse as the square wave get taller, say from a lot of extra headroom in the first place.

I agree that we want to not have the musical signal clipping. Whether that is 1dB below full scale, or 20 dB doesn’t matter a whole lot.
But we do not want it 60dB below full scale, as we start loosing resolution even with a 120dB SINAD DAC. Especially in the quite parts. We end up with 1 or 2 bits, where we should have had 12-16.

 

[dlaloum]

I think the Puffin declicks 'live'... so one could simply rip/record what comes out of it with a low headroom.

Even without that, perhaps someone can rip a song from an LP that has clicks/pops, and do the following test: rip it with plenty of headroom eg your suggested 20 dB, then compare that file by ear with the same file that has been boosted 18 dB then cut 18 dB. The latter file should be 'damaged' with the added hash and harmonics that you are describing, but at the same playback volume.

If the difference is trivial to the ear, then we might as well rip vinyl without headroom for clicks and pops.

Has anyone tried this?

cheers

That won't have the same effect - an overdriven ADC adds additional distortion to its output, it doesn't merely clip off anything above the 0 point.

If you take a digital file and raise it up so the peaks are above 0db - it depends on the software, but at its simplest it will clip the tops off - and the audible effect will be a result of how your DAC handles the resulting data.

It won't sound the same as a clipping ADC.

If you want to experiment, simply make a recording, and intentionally adjust the gain to ensure clipping - it doesn't take much "hard" digital clipping for the sound to get seriously harsh.

If you like record once clipped, and once with the gain turned down, so there is NO clipping.

Then adjust both recordings so they have the same AVERAGE level (don't target the peak levels as you have cliped them on one version!)

Now they will be at the same level, ie: level matched - listen to the clipped version vs the unclipped version

If you process a digital file, and raise its level so it clips, the resulting output will be a clipped file - which will cause similar distortion at the DAC level - just as nasty.

Whether the Puffin suffers from ADC click related gain / clipping issues will depend on whether its de-clicking is done in analogue or digital - you can only cleanly de-click a digital signal AFTER it has been cleanly recorded - which means the click needs to be recorded "in full" without clipping, and then the de-clicking software can do its bit ... without the result having additinal clip related distortions.

 

[JP]

@dlaloum do you have any actual examples that you can provide?

 

[BDWoody]

Whether the Puffin suffers from ADC click related gain / clipping issues will depend on whether its de-clicking is done in analogue or digital - you can only cleanly de-click a digital signal AFTER it has been cleanly recorded - which means the click needs to be recorded "in full" without clipping, and then the de-clicking software can do its bit ... without the result having additinal clip related distortions.

Looks like it is done digitally:

http://parksaudiollc.com/magic.jpg

 

[dlaloum]

Here is an interesting photo comparing conical/spherical stylus with eliptical and line contact types.

I have mentioned elsewhere that they read above and below the groove wear patch - this makes it obvious how and why...

Elipticals have too short a patch to really take advantage of this on word records...

I have this one labelled conical-eliptical, but the right side needle looks a heck of a lot like a Shibata or HyperEliptical / FineLine / etc... design - see below for more on the shapes...

Another thing to note - look at the way that needle rides... none of them go "lower" in the groove - it is never about getting "deeper" into the groove - in fact all the designs avoid that bottom zone - the groove garbage patch!

The other nice thing - the longer the contact patch and the greater the resulting surface area of the contact patch, the lower the aereal pressure on the vinyl, and the lower the wear on both stylus and vinyl... typically a line contact will have a lifetime of at least 4x that of a spherical/conical as a result - and vinyl wear will be reduced accordingly as well...

Another collection of stylus shape information:

This one is from Audio Technica.... in Italian... I am sure there are English versions of this, but this is the one I have!

Mostly self explanatory but...:

A contact patch from above
B Front view of stylus in groove
C Horizontal section of stylus
D Side view of styus and contact patch
E Total areaof contact patch
F Ratio length to width of contact patch

 

[linear diver]

(Is this the right spot for this question? -- I DO respect the TT deeply - that's why I put this here

Hi! l am music and gear lover in the 'hello boomer' age category.

I have been learning a lot at ASR over the years, and I am very happy with my digital system. Now I would like to start using my 46 year old Linn LP12. Mine is the earliest model without the electronic power supply. It's in excellent cosmetic shape. Now I have a Linn Basik LVX arm with a new Grado Black cartridge. I don't think the TT bering has ever been serviced. The suspension springs and rubber washers are original are original.

I have spent some time researching TT setup on ASR, but have not yet found a quick outline of what the current best practices are. An example is azimuth. Some sources recommend comparing L and R crosstalk measurements. Others claim eyeballing physical setup is enough. I'm about to order a digital stylus scale, and I notice a lot of new, low cost alignment accessories available - are any of these worth it?

What I have done so far ...
Checked that all components of the TT are appropriately square and level.
Downloaded an alignment template for my tonearm.
Physically installed a new Grado Black cartridge. Mounting Screws finger tight.
Aligned cartridge and headshell using the downloaded template
Adjusted tracking weight with an old-school mechanical balance (going to get digital soon!)
Adjusted anti-skating with clear band on test record.
Adjusted pivot height to give SRA of 92 - 93 degrees (using microscope)

It sounds pretty good! But where can I go from here?

 

[mike70]

(Is this the right spot for this question? -- I DO respect the TT deeply - that's why I put this here

Hi! l am music and gear lover in the 'hello boomer' age category.

I have been learning a lot at ASR over the years, and I am very happy with my digital system. Now I would like to start using my 46 year old Linn LP12. Mine is the earliest model without the electronic power supply. It's in excellent cosmetic shape. Now I have a Linn Basik LVX arm with a new Grado Black cartridge. I don't think the TT bering has ever been serviced. The suspension springs and rubber washers are original are original.

I have spent some time researching TT setup on ASR, but have not yet found a quick outline of what the current best practices are. An example is azimuth. Some sources recommend comparing L and R crosstalk measurements. Others claim eyeballing physical setup is enough. I'm about to order a digital stylus scale, and I notice a lot of new, low cost alignment accessories available - are any of these worth it?

What I have done so far ...
Checked that all components of the TT are appropriately square and level.
Downloaded an alignment template for my tonearm.
Physically installed a new Grado Black cartridge. Mounting Screws finger tight.
Aligned cartridge and headshell using the downloaded template
Adjusted tracking weight with an old-school mechanical balance (going to get digital soon!)
Adjusted anti-skating with clear band on test record.
Adjusted pivot height to give SRA of 92 - 93 degrees (using microscope)

It sounds pretty good! But where can I go from here?

a better cartridge

if you don't want to spend a lot, let me say the AT vm95ml or the vm540ml are spectacular.
tracks like a train, zero IGD or difference between inner and outer tracks
lots of detail ... 180 degrees from a Grado Black

my vm540ml is a permanent occupant in my set of cartridges.

 

[Bob from Florida]

(Is this the right spot for this question? -- I DO respect the TT deeply - that's why I put this here

Hi! l am music and gear lover in the 'hello boomer' age category.

I have been learning a lot at ASR over the years, and I am very happy with my digital system. Now I would like to start using my 46 year old Linn LP12. Mine is the earliest model without the electronic power supply. It's in excellent cosmetic shape. Now I have a Linn Basik LVX arm with a new Grado Black cartridge. I don't think the TT bering has ever been serviced. The suspension springs and rubber washers are original are original.

I have spent some time researching TT setup on ASR, but have not yet found a quick outline of what the current best practices are. An example is azimuth. Some sources recommend comparing L and R crosstalk measurements. Others claim eyeballing physical setup is enough. I'm about to order a digital stylus scale, and I notice a lot of new, low cost alignment accessories available - are any of these worth it?

What I have done so far ...
Checked that all components of the TT are appropriately square and level.
Downloaded an alignment template for my tonearm.
Physically installed a new Grado Black cartridge. Mounting Screws finger tight.
Aligned cartridge and headshell using the downloaded template
Adjusted tracking weight with an old-school mechanical balance (going to get digital soon!)
Adjusted anti-skating with clear band on test record.
Adjusted pivot height to give SRA of 92 - 93 degrees (using microscope)

It sounds pretty good! But where can I go from here?

What are you using for a phono preamp - need to know before making a suggestion.

 

[Holmz]

…

It sounds pretty good! But where can I go from here?

May I suggest Thom Makris of Galibier Design in Colorado.

 

[linear diver]

Thanks for the replies! I'm processing away ...

@mike70, It makes perfect sense to upgrade the cartridge. I used Grado's 30 years ago and it does not look like they have changed much. The vm95ml is cheap enough to be a no-brainer. And the vm540ml is well within reach.

@Bob from Florida, I have not seriously listened to records in quite a while. Right now, I'm using an old Musical Fidelity V-LPS mm/mc preamp module into a Behringer MX602A mixer and from there balanced line level into my Genelecs. I also have an E1DA Cosmos ADC and one APU, so I can try that as well.

Great suggestion @Holmz, I had no idea someone like Thom was local. I'll give him a call.

 

[dlaloum]

The Grado Black is a very basic cartridge - but it has substantial scope for upgradeability, as it shares it's "Chassis" with the Gold, or the much more upmarket 90's Grado exchangeable needle cartridges eg: MCZ / MCX / 8MZ / XTZ / TLZ - these styli will upgrade a Black through Gold chassis to similar performance to the very upmarket "Woody" cartridges from Grado.
Difference from Black through to Gold in the current range of bodies, is basically a process of cherry picking the best matched (left/right) bodies at the factory before sorting them into price grades, and then fitting the relevant stylus to them. (that means you can get lucky with a very well matched black body, but you are more likely to get a well matched body in the gold range)

Moving to something like an AT95 family cartridge will be a sideways move... as a family the AT95 and Grado's are of similar levels of quality (depending on the needle fitted!) - but they do have very different "sound". (Clearaudio Maestro is a 95 family chassis, fitted to a custom shell, with a special needle fitted...)

Having said that - getting the best from a Grado, is achieved with a relatively low mass arm (they are high compliance) - the Linn Basik LVX is at the heavier end of the acceptable range...

Also you should check the rubbery bits on the counterweights, as they have a reputation for degrading over time ... you might need to get some replacement bits.

As an aside - optimal performance will be achieved by optimal matching of cartridge to arm (effective mass / compliance) - and performance will be driven by the cantilever type (mass mostly) more than anything else (more than whether it is MM or MC).
Typically a top grade cantilever will come with a top grade needle on the end of it (the needle is cheaper than the cantilever... so you can get great needles on not so great cantilevers)