The W&F Analyzer Project, Uberrimus

[Balle Clorin]

VPI Prime..

My own Michell gyro SE

Not sure why wow and 0.55hz peak is so much larger on the Uberrimus app?

 

[Thomas_A]

I am seriously now thinking of a new bearing of my Axis which is 36 years old. The Karousel. I have checked the inner platter when spinning it manually and it has significantly more friction than other inner platters I've seen tested on videos. It's only the price and if it can be fitted that makes me hesitate, and also whether it will make any difference whatsoever on sound or measurements.

 

[JP]

Taking a break for a couple days from this. Hung out with a friend on Thursday who's in the industry, and one of the outcomes was the idea to simulate W&F profiles for listening tests so people can understand their sensitivity. I made a script that uses the 16 peaks from the FM demodulated plot data from wf_core to build a profile. You can then use that profile to apply simulated wow and flutter at different DIN metrics to audio files.

Here's an SP-10MKII FG profile simulated from 0.05% to 1.5% in 16 increments. The test is to choose which is the reference file vs. the higher W&F files.

acidtest.io listening tests

 

[Balle Clorin]

That will make people hurt.. I have made av W&F simulator too ( have the virtin files too) and as I recall I could ( with some training) hear very small deviation, few turntables will be good enough to be inaudible..on test tones, in music it is much more tolerable. Will be interesting to do your test ( blind I presume)

I think you need to include and 0.02 and 0.01 too .0.05 is quite audible , on test tones.Do you mean 2S or classic DIN ?

EDIT: I see now that WF can be applied to any audio file, very interesting

 

[Balle Clorin]

I am seriously now thinking of a new bearing of my Axis which is 36 years old. The Karousel. I have checked the inner platter when spinning it manually and it has significantly more friction than other inner platters I've seen tested on videos. It's only the price and if it can be fitted that makes me hesitate, and also whether it will make any difference whatsoever on sound or measurements.

I tnink a frictionless bearing will have less damping an let more motor irregularities through,, so some friction may be good..you mat try to dampen you’re springs or even immobilise them, less springiness improve wf on my gyro

 

[Thomas_A]

I tnink a frictionless bearing will have less damping an let more motor irregularities through,, so some friction may be good..you mat try to dampen you’re springs or even immobilise them, less springiness improve wf on my gyro

I will clean out my bearing housing and replace the oil as a first step. I may be dirty oil causing a bit more friction. I checked the bearing tip and cannot see anything that would indicate wear. Otherwise, the 0.5 and 1.0 Hz is a belt issue most likely. The belt probably needs to be changed more often, sad to say for the expensive Linn belt. I will try a Thakker belt now at 1/3-1/4 the cost.

 

[Thomas_A]

Cleaned out the bearing 3x with sewing machine oll and replaced the Linn black oil with the same sewing machine oil. The inner platter now spinned much longer and there was an effect on both speed, drift and W&F. The 1 Hz peak decreased significantly, but still have the 0.5 Hz. Now I need to adjust speed but will wait for a new belt. No need for an expensive "Karousel" bearing...

Before:

After:

 

[JP]

That will make people hurt.. I have made av W&F simulator too ( have the virtin files too) and as I recall I could ( with some training) hear very small deviation, few turntables will be good enough to be inaudible..on test tones, in music it is much more tolerable. Will be interesting to do your test ( blind I presume)

I think you need to include and 0.02 and 0.01 too .0.05 is quite audible , on test tones.Do you mean 2S or classic DIN ?

EDIT: I see now that WF can be applied to any audio file, very interesting

There is a link to the test - you can show me how quite audible

 

[OHtaru]

Since I started using tape walkmans and cassette decks, I've found that my ears are more sensitive to flutter.

My audible threshold obtained in the W&F listening test is 0.2%. But I can hear the flutter of my cassette deck especially the 30Hz component, although the WRMS is only 0.085%.

This has caused a lot of debate among cassette enthusiasts. Most people believe that W&F audibility is proportional to WRMS, but at least my experience tells me that this is not the case. Flutter sounds more obvious and annoying to my ears. I would rather choose a one with a slightly higher WRMS but very little flutter than one with a nice-looking WRMS but high flutter. I have owned several Sony DD walkmans, but without exception I sold them all even though they are claimed to have low W&F. Their WRMS can usually easily reach below 0.08% if properly serviced. However, the Unwtd flutter is around 0.3% to 0.4% from what I remember.

Just to add one more thing, the above experience is mainly more noticeable when listening through speakers. My thought is that the FM'ed signal becomes an FM-AM signal due to the multipath effect caused by room reflections, thereby enhancing its audibility, and this is much better when using headphones.

 

[Balle Clorin]

There is a link to the test - you can show me how quite audible

Will do , but on a road trip now. Post 161 is audible to me 0.05% DIN. With single tones , not music.. Maybe belt wow is more detectable than DD flutter? Would be interesting if you made a belt drive simulation too

 

[Balle Clorin]

I will clean out my bearing housing and replace the oil as a first step. I may be dirty oil causing a bit more friction. I checked the bearing tip and cannot see anything that would indicate wear. Otherwise, the 0.5 and 1.0 Hz is a belt issue most likely. The belt probably needs to be changed more often, sad to say for the expensive Linn belt. I will try a Thakker belt now at 1/3-1/4 the cost.

Talkum on my belt always reduce W&F.

 

[JP]

But I can hear the flutter of my cassette deck especially the 30Hz component, although the WRMS is only 0.085%.

This has caused a lot of debate among cassette enthusiasts. Most people believe that W&F audibility is proportional to WRMS, but at least my experience tells me that this is not the case. Flutter sounds more obvious and annoying to my ears. I would rather choose a one with a slightly higher WRMS but very little flutter than one with a nice-looking WRMS but high flutter. I have owned several Sony DD walkmans, but without exception I sold them all even though they are claimed to have low W&F. Their WRMS can usually easily reach below 0.08% if properly serviced. However, the Unwtd flutter is around 0.3% to 0.4% from what I remember.

It's similar to looking at THD as a number - that works when that number is very low, but when it's on the cusp of audibility you need to look at the spectrum. All analog formats are close enough to audibility that the waveform and spectrum tells far more than the metric ever can.

 

[Thomas_A]

Talkum on my belt always reduce W&F.

Which f? You have a round belt?

 

[mglobe]

Based on my results I’m going to stop worrying about W&F ‍

 

[JP]

Whipped together a motor calculator: https://docs.myhi.fi/tools/motor-frequency-calculator.html

 

[Balle Clorin]

Which f? You have a round belt?

When peak variation /2S reaches 0.12 on phone app i treat the round belt with talcum , it then drops to 0.05-0.08 peak deviation. This app

Shaknspin show same effect , but app is more available

Shankspin experiment with talcum

 

[Thomas_A]

I have to learn more about how to interpret this stuff.

Here's a polar plot from an SP-10MKII with the AP test record:

View attachment 520614

And now the SL-1200GR with the Tacet test record:

View attachment 520613

Something a bit off with those records or something else. Both show too much wiggles.

My AP and Tacet:

 

[Balle Clorin]

I have to put this somewhere..about belt drive and vibration

Belt Passing Frequency Vibration Calculator

Belt Vibration Calculator finds the Belt-Passing Frequency of worn out Belt-Driven Machines.

www.ritec-eg.com

Spoiler: belt drive issues

Vibration Library & Tools

Vibration Analysis Case Studies - Bearing Frequency Calculator - Demos & Simulation - Vibration Calculator - Belt Frequency Calculator - FFT Simulation - Interactive Vibration - Vibration Simulation

www.ritec-eg.com

Belt Passing Frequency Vibration Calculator

Belt Vibration Calculator finds the Belt-Passing Frequency of worn out Belt-Driven Machines.

www.ritec-eg.com

another case on belt

difference between 2 belts..

Different vibration sidebands on 33 and 45 rpm pulley..( elevation and stretch changes)

IN FINALLY A GOOD BELT. Record is CA TRS-1007 1000hz and silent groove between sweeps

 

[JP]

I'm at a bit of a crossroads here. Making this similar to an analog wow and flutter meter is pretty easy because the filters are basic, but that doesn't necessarily make a good diagnostic tool. Something like Multi-Instrument is using FFT brick wall filtering for band separation between wow and flutter, which is much, much tighter, but it also causes issues with low carrier frequencies that motor frequency generators use. I'm not sure which way I want to go with this right now. I have working versions of both, but there are some interesting trade-offs.

Might end up doing two measurement paths, one for low frequency carriers and one for high frequency carriers. The other "issue" with the more advanced filtering being FFT-based is that the processing takes a lot longer. What was sub-1-second processing is now 6+ seconds of processing for a 60-second signal file.

 

[acd]

No. An off center record will give kind of useless and misleading results.depending on the eccentricity,,, A good record can have wow of 0.02-0.05%, a poor off center one 0.2-0.3 % and will overshadow the actual TT speed performance . A very good TT can have wow peak variation 0.01-0.05, an OK one 0.05-0.1 A poor one 0.2. I am referring to 2S or peak wow, which represent peak variation , not DIN W&F value.

Yesterday, I got aware of this script and used it on a couple of wav files I have recorded some time back. Eccentric, and likely warped, test records ruin the results. This script requires (near) perfect records which are very hard to find. I (and Claude) work on a front end filtering mechanism that may happen to make the script less infected by inferior test records. I may explain the idea later if successful.