Fun with vinyl measurements

[Thomas_A]

Hi Thomas, sorry for the delay in replying—I hadn’t had a chance to check the forum.

Thank you very much for sharing your results. There’s definitely something wrong with the frequency counter in my script, which is becoming evident with the Denon disc recording.

As @Balle Clorin pointed out, the script calculates frequency using the zero-crossing method, which is the one recommended by Virtins. However, I’ll review the filtering chain using your file, as it’s possible that noise is affecting the measurement—just as you and Balle have already suspected.

As soon as I have some time, I’ll revisit the implementation and share an update.

Best regards!

I really like the script as such since it works on OSX, so if you could develop it, I would appreciate it.

Do you get similar plots and results from Virtins and the script using the below files?

Denon file:

Dropbox

www.dropbox.com

Tacet file:

Dropbox

www.dropbox.com

 

[alvaro-oliver]

Do you get similar plots and results from Virtins and the script using the below files?

Here are the results. I think the measurements are pretty close.

Tacet	Unw. W&F	Weig. W&F	Freq (zero-cross.)	Freq (FFT)
Virtins	0.0777	0.0391	3150.206	3150.0877
WnF Script	0.0885	0.0426	3150.206

And the frequency counter actually seems to be working well after all.

Denon	Unw. W&F	Weig. W&F	Freq (zero-cross.)	Freq (FFT)
Virtins	0.4071	0.0459	2214.87	2215.1173
WnF Script	0.3295	0.0519	2215.093

2.15 KHz is quite a bit off from the nominal 3 KHz, and it roughly matches a 3 KHz tone recorded at 45 rpm but reproduced at 33.33 rpm, so it’s worth double-checking that.

Also, when checking the audio file, it’s recorded at a very low level. Try adjusting the tone level to around -6 dB on the monitor before starting the capture.

I hope this helps!

Tacet :

Denon:

 

[alvaro-oliver]

The frequency can be found by other methods, I cannot recall exactly but AI/coplitot has written some Python scripts for me - calculating wow and flutter and so on… it may give you suggestions too.
Zero crossing would need heavy filtering to remove low frequency garbage and clicks and pops filtering that affect the zero crossing

Of course, the most intuitive way to estimate the average frequency is by taking the peak frequency from the FFT analysis.

Another method, which I used before and find more efficient for this application, is based on the instantaneous frequency signal. This comes from the Hilbert transform applied to the same FFT analysis, so the results are quite similar.

That said, the zero-crossing method also has its advantages: it’s a more direct time-domain measurement with less systematic error than frequency-domain analysis. It doesn’t need pre-filtering and isn’t affected by sample duration, window size, or sampling rate.

Just to clarify, this method is only used to determine the central frequency of the modulated signal. That frequency is then used to tune the filters and report speed. Wow and flutter are analyzed using the instantaneous frequency signal from the Hilbert transform.

Cheers!

 

[Thomas_A]

Here are the results. I think the measurements are pretty close.

Tacet	Unw. W&F	Weig. W&F	Freq (zero-cross.)	Freq (FFT)
Virtins	0.0777	0.0391	3150.206	3150.0877
WnF Script	0.0885	0.0426	3150.206

And the frequency counter actually seems to be working well after all.

Denon	Unw. W&F	Weig. W&F	Freq (zero-cross.)	Freq (FFT)
Virtins	0.4071	0.0459	2214.87	2215.1173
WnF Script	0.3295	0.0519	2215.093

2.15 KHz is quite a bit off from the nominal 3 KHz, and it roughly matches a 3 KHz tone recorded at 45 rpm but reproduced at 33.33 rpm, so it’s worth double-checking that.

Also, when checking the audio file, it’s recorded at a very low level. Try adjusting the tone level to around -6 dB on the monitor before starting the capture.

I hope this helps!

Tacet :

View attachment 477984

View attachment 477985 View attachment 477986

Denon:
View attachment 477988
View attachment 477989 View attachment 477990

Just a comment; the Denon record is a 45 RPM and I played it at 33,33 RPM. So it should be 2222 Hz. I set the speed using the Tacet record though, and records do differ a bit in f.

 

[Thomas_A]

Here are the results. I think the measurements are pretty close.

Tacet	Unw. W&F	Weig. W&F	Freq (zero-cross.)	Freq (FFT)
Virtins	0.0777	0.0391	3150.206	3150.0877
WnF Script	0.0885	0.0426	3150.206

And the frequency counter actually seems to be working well after all.

Denon	Unw. W&F	Weig. W&F	Freq (zero-cross.)	Freq (FFT)
Virtins	0.4071	0.0459	2214.87	2215.1173
WnF Script	0.3295	0.0519	2215.093

2.15 KHz is quite a bit off from the nominal 3 KHz, and it roughly matches a 3 KHz tone recorded at 45 rpm but reproduced at 33.33 rpm, so it’s worth double-checking that.

Also, when checking the audio file, it’s recorded at a very low level. Try adjusting the tone level to around -6 dB on the monitor before starting the capture.

I hope this helps!

Tacet :

View attachment 477984

View attachment 477985 View attachment 477986

Denon:
View attachment 477988
View attachment 477989 View attachment 477990

Hm, I don't really get the same visual plots from the script when I run them; yours are scaled differently. Numbers same though. Do you set the scale manually?

 

[alvaro-oliver]

Hm, I don't really get the same visual plots from the script when I run them; yours are scaled differently. Numbers same though. Do you set the scale manually?

That’s because I adjusted the zoom on the instantaneous frequency and histogram before exporting the image. To change the zoom, just click the magnifying glass icon in the results window.

 

[dougi]

DENON DP-50M SORTED
Despite recapping my 1978 but new to me tt, it's slight stopping problem remained. I replaced all the c458, power 2sc2023 (nos) and c1162 transistors today. All these ones were "black legged". Issue finally solved.
W&F still vv low. No need to buy a new sl-1300g, despite these being on heavy sale here ATM.

 

[morillon]

DENON DP-50M SORTED
Despite recapping my 1978 but new to me tt, it's slight stopping problem remained. I replaced all the c458, power 2sc2023 (nos) and c1162 transistors today. All these ones were "black legged". Issue finally solved.
W&F still vv low. No need to buy a new sl-1300g, despite these being on heavy sale here ATM.

View attachment 478799

These are indeed very comfortable ( and audible) wf values that even modest dd s from the 70s/80s allow...
is the speed just right or is it a measurement aberration? Adjustable?
;-)

 

[dougi]

These are indeed very comfortable ( and audible) wf values that even modest dd s from the 70s/80s allow...
is the speed just right or is it a measurement aberration? Adjustable?
;-)

Speed is a measurement quirk. Spot on using the Waxwing runout click technique.

 

[morillon]

0.2% is a lot... I think it's a good idea to stay below 0.1%... to be quite explicit, for "the musicians in the room"... at 441Hz... 0.1%... 0.44Hz...

It's obviously sensitive...
;-)))
(Explicite- illuminated too wf variations impacts for proportions but at "+-" 0.05% etc etc)

 

[dougi]

0.2% is a lot... I think it's a good idea to stay below 0.1%... to be quite explicit, for "the musicians in the room"... at 441Hz... 0.1%... 0.44Hz...

It's obviously sensitive...
;-)))
(Explicite- illuminated too wf variations impacts for proportions but at "+-" 0.05% etc etc)

As said, actually speed is spot on. The deck also has mechanical strobe stripes to be double sure.

 

[morillon]

As said, actually speed is spot on. The deck also has mechanical strobe stripes to be double sure.

I understood...I just took the opportunity to point out this very general subject.
;-)
(all the more so as this...is added to the upstream deviations...during the design and manufacturing process of the disc! and can measure it by observing the wanderings of our test discs which could be considered a minimum "clean" "qualitatif"!?!)

 

[Balle Clorin]

Phone apps and Even Shaknspin is often off on absolute speed. The reference is the strobe check.

 

[morillon]

Phone apps and Even Shaknspin is often off on absolute speed. The reference is the strobe check.

At its simplest... sure...mains frequency or built in?
(and in general those in situ of the machines should be largely sufficient)


or, but there are still limited precautions because light sources are limited by frequency accuracy... for example, my 50Hz mains voltage is officially at 0.1% in my country and it slips in effect... if we use 300Hz LED sources, we can also check the theoretical accuracy of the quartz crystals, etc. Otherwise, if it has been validated on, for example, a Denon and its tape system* or other approaches, using a serious 3k or 3k150 record, a simple tuner indicating the frequency or frequency meter in an Android app, etc., with a nice AVG, allows this to be done easily, etc.... ;-))

Comparing approaches, as always, allows us to do this over time...
(especially since it is affected by... the WF of our turntables...!)

(hence the interest in validating on turntables with very low WF)


(*maybe the safest in the end)

Personally, I generally always check if my 50.00hz is ok at the time of measurements as a precaution especially since currently an AC motor is not frequency regulated...


;-)

It will be interesting to understand the speed precision of the different test disks currently available on their 3k 3h150 despite all their clumsiness in production

 

[Balle Clorin]

I use a quartz controlled strobe light of course..
But my power hz is good too

 

[morillon]

I use a quartz controlled strobe light of course..
But my power hz is good too

View attachment 479023

My point is just to say that despite everything, all approaches are fraught with uncertainty and can be challenged... (I don't think there's an ultra-stable power supply anywhere in the world... I imagine that the one in my country is actually quite stable (+-0.1% and is easily observed), I fear that there remains serious observation to be made at the time of the measurements)
the wf is a real, real problem...but respecting the speed is in my opinion a good indicator of seriousness on the manufacturer and engineering side...and going below the +-0.1% bar is a minimum

;-)
(If you are sure of your speed and its precision by making the distinction with the wf... the observation of your numerous discs tested at 3k 3k150 becomes interesting.... the frequency meter method can "return to service" ;-) )

 

[Thomas_A]

Another way of looking at frequency response.

My Shure V15V with JICO SAS, 100 pF+140 pF/36k. Original CD vs. original LP, R.E.M. Losing my religion, spectrum.

Same with 330 pF + 140 pF/36k

 

[Balle Clorin]

You can user trace arithmetic and show the difference , easier to se the difference then

 

[Thomas_A]

You can user trace arithmetic and show the difference , easier to se the difference then

Yes,

I was just want to compare with the loading setting based on the CA-TRS1007. A "linear" response at the first track correlates with a drop of ≈1 dB in the HF range, which is track 2 on the LP. It is probably a good idea to use the tracks in the middle to get closer to the real world.

 

[melloncolliecat]

Did you specifically choose an LP that you know is from the same master as the CD? @Thomas_A