advertisement


Wow and flutter 'meter', etc

Jim Audiomisc

pfm Member
Out of curiosity, over the last week I've written a program than can take a recording from a test LP and use it to generate Wow and Flutter measurements to the AES spec for weighting, etc. It also shows the average test tone frequency with an accuracy of about a few parts-per-million to check the basic speed (assuming the test LP is correct!) Plan to make the program and source code public. But am wondering if people will find this of interest/use. It also plots the speed/time.

FWIW although a 3k test tone is usual for the standard tests, it work with any decent tone over a range from about 1k up to about 10k. So most test LPs should have something suitable.

I'm also thinking of having a bash at something similar to check THD, crosstalk, etc, as an aid for checking cartridges and their alignment. Again, is this likely to be of interest as free open software? As above, I'm mainly doing it out of curiosity and to experiment with.

Jim
 
Always good fun to do, but free tools already exist: https://play.google.com/store/apps/details?id=ch.progtec.PlatterSpeed&hl=en

With this sort of measurement, turntables are so good these days, it is hard to know whether what you are observing is speed instability in the record lathe or the deck.

The main 'app' there seems to be Android, but I couldn't immediately see if it or the alternatives were free and open source. I confess I'm slightly wary of the way the term "free" may be used for Android given the tendency for 'apps' to 'track' the user, etc.

Yes, it would be a good idea to use more than one test LP if someone wanted to check accuracy. It also makes sense to use LP tracks at more than one radius for a given LP to correct for the LP being off-center.

FWIW I don't use Android, nor, these days, Windows or Macs. I use Linux and RISC OS and make the source code available for free so others can adapt it as they may wish.

Jim
 
There was an interesting piece of software processing on Vinylengine that took a recording of the 3kHzish test tone and plotted frequency deviation as a polar graph based on 33.33 rpm rotation. This was good for showing platter pulley and motor pole issues
 
A noble effort. Do you not think that most HiFi users are based on something other than Linux/RiscOS? This may limit the amount of use it gets.
 
A noble effort. Do you not think that most HiFi users are based on something other than Linux/RiscOS? This may limit the amount of use it gets.

Afraid I didn't choose my computer platforms purely on the basis of what others prefer. :) Fortunately, one of the points and advantages of making source code free and open is that others can use it on whatever platforms they prefer. So there is actually no need for us to all use exactly the same operating system.

I write programs mainly in 'C', which many programmers can use on Windows, Macs, etc. And when I have versions that work OK, I put the source code on the web, open and free for others to access, copy, etc, as they want.

Hence once I put up the code, others can help themselves to it if they wish. That should apply whatever platform they use or prefer be it Windows, Macs, whatever.

Jim
 
I did something like this a few years ago. But I didn't want to calibrate it to W&F standards, it doesn't seem to be relevant and I wanted to explore the nature of the variations rather than make arbitrary judgements about their magnitude.

I didn't 'open' the code or giveaway the exes because I didn't want to support it or write instructions. And since it wasn't calibrated, what the output means is moot.

What I wanted to do was answer the question 'Is this better than that?' when modifications or maintenance were done. So I can see the 50Hz and 100Hz pulsing of an LP12 AC motor, if you add a little drag to the platter I can see the speed drop as the belt stretches and the 50Hz wobble is reduced.

That sort of stuff. The limitation is having to use a record with a test tone, so what happens when playing a real record remains obscure, and much more interesting.

Paul
 
I did something like this a few years ago. But I didn't want to calibrate it to W&F standards, it doesn't seem to be relevant and I wanted to explore the nature of the variations rather than make arbitrary judgements about their magnitude.

Yes, Paul, it was very interesting and the graphs generated were very instructive. I was trying to find the old thread early, I think this was back in 2012?
 
Out of curiosity, over the last week I've written a program than can take a recording from a test LP and use it to generate Wow and Flutter measurements to the AES spec for weighting, etc. It also shows the average test tone frequency with an accuracy of about a few parts-per-million to check the basic speed (assuming the test LP is correct!) Plan to make the program and source code public. But am wondering if people will find this of interest/use. It also plots the speed/time.

FWIW although a 3k test tone is usual for the standard tests, it work with any decent tone over a range from about 1k up to about 10k. So most test LPs should have something suitable.

I'm also thinking of having a bash at something similar to check THD, crosstalk, etc, as an aid for checking cartridges and their alignment. Again, is this likely to be of interest as free open software? As above, I'm mainly doing it out of curiosity and to experiment with.

Jim

All of that sounds fantastic Jim and I would definately encourage you :cool:.

Paul, Si sent you some files for analysis..... Perhaps they found their way to your 'junk' folder?
 
I did something like this a few years ago. But I didn't want to calibrate it to W&F standards, it doesn't seem to be relevant and I wanted to explore the nature of the variations rather than make arbitrary judgements about their magnitude.

I didn't 'open' the code or giveaway the exes because I didn't want to support it or write instructions. And since it wasn't calibrated, what the output means is moot.

What I wanted to do was answer the question 'Is this better than that?' when modifications or maintenance were done. So I can see the 50Hz and 100Hz pulsing of an LP12 AC motor, if you add a little drag to the platter I can see the speed drop as the belt stretches and the 50Hz wobble is reduced.

That sort of stuff. The limitation is having to use a record with a test tone, so what happens when playing a real record remains obscure, and much more interesting.

Paul

You don't necessarily have to actively "support" something if you just put up the sources, etc, with some basic info and a "help yourself" statement. I do get occasional emails asking for changes to what I've made available, and if they seem a good idea, I sometimes do them. But in general my reaction tends to be to encourage others to fiddle about and make their own versions and modifications.

Re drag, etc. One thing I've wondered about is recording from *before* a test tone begins. That would allow the program to track the changes in rate following the onset of the increase in drag caused by modulation. I've not yet tried it, though, as I've mainly been focussed on sorting out aspects like ensuring the AES weighting and 'two sigma' measurement of the wow and flutter value. Early days and I'm just experimenting out of curiosity.

One of my other programs might be more useful for the above as it does an analysis, block by block, though a file and doesn't assume you want a wow and flutter, but just spectra and a nominal 'counter' frequency. However I have a feeling the program has a bug so I should check it. (Last time I used it, it didn't seem to want to let me choose a start time in the file, which it should.) Its called WAV_FFTScan if anyone else was to look at it.

FWIW I generally start by writing a RISC OS version as I find that the easiest platform to get something sorted out. Once happy I then do a GCC version for Linux which others tend to find easier to port and modify if they're interested. But I wasn't sure if there would be much interest. Hence mentioning it here.

Jim
 
I use WFGui for setting speed and W+F tests on tape decks, it can use 3000 and 3150 Hz tones, something that could tell me what the frequency was if not close to 3000 or 3150 Hz would be useful as WFGui starts rounding to the nearest 5Hz if it's far off. I have a half speed cassette deck and reel to reels to set the speed for too.
 
FWIW I generally start by writing a RISC OS version as I find that the easiest platform to get something sorted out. Once happy I then do a GCC version for Linux which others tend to find easier to port and modify if they're interested.

Or you could just write it in Java and have no porting issues at all :)
 
This was the polar output for some test files Dave Cawley created with a proper (IIRC) FM modulation of some form.

DC_Calibrate_polar.png


So that's a pure tone, one at 0.1% deviation and lastly at 1% deviation. Modulation is at 6.66Hz, which is realistic in terms of poles of a DD motor etc.

It all goes a bit more wrong with the real world.

Phix_polar.png


I'd have to research to find out what's going on there, same recording, same TT as I recall. Perhaps an idler involved.

I find this type of presentation quite fascinating, but it's not a W&F meter and it's not real time as implemented. I'm looking forward to seeing Jim's code. However what would be interesting would be a real time analyser for a common platform, even if only as a sanity check on some of the other tools out there. Open Source definitely wins when it gets some peer type attention and some certainty that it is doing what the author intended.

Paul
 
Paul your Polars and the wavelet power analysis are by far the most useful speed tools I've ever seen or used. The polars give so much useful info about deviation, they can show multiple issues at the same time and are incredibly easy to interrogate.
 
I used to have one of those when I was a kid. Great fun, it was.

il_fullxfull.285261139.jpg


With a bit of practice you can make nice patterns, like Paul's.

Joe
 
....It all goes a bit more wrong with the real world.

Phix_polar.png


I'd have to research to find out what's going on there, same recording, same TT as I recall. Perhaps an idler involved.

I find this type of presentation quite fascinating, but it's not a W&F meter and it's not real time as implemented. I'm looking forward to seeing Jim's code. However what would be interesting would be a real time analyser for a common platform, even if only as a sanity check on some of the other tools out there. Open Source definitely wins when it gets some peer type attention and some certainty that it is doing what the author intended.

Paul
I wonder how many direct drive turntables out there have winding balance or even not working faults
 


advertisement


Back
Top