LP Wow and Flutter program
jim lesurf (2082) 1438 posts |
Not sure if there are many RO users who also still play LPs. But just in case: I’ve just put up a new program, !LP_WowAndFlutter, which can be obtained from my software page http://www.audiomisc.co.uk/software/index.html Given a recording made from playing a tone on a test LP the program will evaluate the wow and flutter behaviour. It can be used to plot the speed variations versus time, rotation, etc. And will work out the wow and flutter percentage value in accord with the recommended Audio Engineering Society method. Source code included. Jim |
Dave Higton (1515) 3534 posts |
I recently got a turntable working again. (It reminds me of how awful vinyl is, compared with CD… I’m too mean to buy the old material again.) But is there such a thing as an inexpensive test disc with frequency response tests, i.e. bands several seconds long at discrete frequencies? I had access to one 40 years ago, but not nowadays. I’m using a Shure M75EJT2, which has a pronounced mid range suckout. I could design and build a compensator. |
jim lesurf (2082) 1438 posts |
You can sometimes find test discs second-hand. Of course that may mean they have been damaged by past misuse. However my wow and flutter program measures frequency by a ‘cycle counting’ method. Hence it doesn’t mind if the tone has harmonic distortions. Like a true hifi geek I still do have various test LPs, mainly bought mumble decades ago. Beyond that, personally I use the “Analogue Productions” test LP. The snag being that this is produced in the USA. I bought a small batch for my own use from them. I think some dealers in the UK may import them. I assume they may not be “inexpensive”. though. However you raise a question I can ask on a forum I’m using. That does have some dealers posting, so they may be able to tell me of alternatives. Stay tuned… :-) |
Clive Semmens (2335) 3276 posts |
I’m sure you’ve thought of this extension to your program, Jim, but in case you haven’t – if you can analyse wow and flutter, surely you could correct it? |
jim lesurf (2082) 1438 posts |
The difficulty is the usual one. The theory is easy, the practice is a nightmare. 8-] Yes, you can analyse the cyclic variations which are rotation phase angle locked to the turntable when you play a test tone. The challenge would be how to know the rotation angle at any instant for a recording of a normal LP, simply from the output. Easier would be to develop a ‘correction’ for the turntable drive system. However in practice a good direct drive will be trying to minimise this anyway, and react to what is actually happening, rather than what happened six weeks ago when you did a test. For turntables that use belt or idler wheel drive the phasing of the flutter won’t be tied to the deck rotation angle, and will vary as the drive ‘slips’, etc. So its a nice idea, But I won’t hold my breath waiting for someone to do it. 8-] WRT test LPs, I’ve not had any reponses as yet. I only know of two recent and probably current examples. 1) The Hi Fi News disc which may still be available from ‘moth’ in the UK 2) The Analogue Productions one. Which you may well have to buy from the USA. A snag with [ 1 ] is that the test tones on it aren’t ideal. They are mainly 300 Hz which is a bit on the low side. 3150 Hz is the standard. I know my program does fine with a 1 kHz tone, but I’m not sure about 300 Hz. When I get a chance I’ll test that and other frequencies to see what I get. |
jim lesurf (2082) 1438 posts |
BTW if you look at http://jcgl.orpheusweb.co.uk/temp/polar.png you can see a plot typical result. This would be a series of samples all a distance ‘1’ from the origin if the rotation was perfectly uniform. They represent the speed (frequency) versus turntable rotation angle. The distance from the origin scales with the speed (frequency). Further than ‘1’ means faster, Closer to the origin means slower. You can see from the per-cycle ripples the effects of things like cogging’ or other cyclic variations like the drag being angle dependent. Note the scaling is expanded so that a change of 0.1 in radius from the center means 0.1 percent change in speed. The plot shows 9 sec of measurement at 33.333 rmp |
Clive Semmens (2335) 3276 posts |
Ah. I’ve probably missed things what with not having read the whole thread…I was assuming you were doing a clever analysis from a music disc – much harder than doing it from a test disc, and dependent on assumptions about the music on the disc as well. But if you’d done it from a music disc, correcting that particular disc shouldn’t then be too terribly hard I wouldn’t have thought. How often is the error more in the disk than in the turntable? It’s surely not that difficult to engineer a turntable that has the centre post accurately in line with the axis of rotation – and if the problem is that the drive is inconstant, rather than a centring error, then the variations won’t be phase locked…(as you said, of course)… |
jim lesurf (2082) 1438 posts |
Simply from experience, I’d say that many musical LPs are pressed off-center and/or have an over-sized central hole making them prone to being placed off-center by amounts that give more wow than the rotation speed. However ‘flutter’ tends to arise for other reasons. e.g. if you look at the polar plot from my Technics direct-drive you can see the ‘cogging’ variations in speed which give a number of cycles of variation per rotation. Flutter can be more audible than wow, but has a more complicated effect on the sound. Disc ‘warps’ also cause variations within a rotation, but are usually ‘one bump per rotation’ effects. So the point of the measurement is to give a general ‘health check’ to the turntable and to see if it is within spec, or find how much out of spec it may be. (My Technics is 30+ years old, so is doing quite well given how much use it has had!) |
Clive Semmens (2335) 3276 posts |
That was my suspicion. And a test record result is irrelevant to that, of course. The only thing you could do about that would depend on that pitch variation analysis of the music on the affected record – difficult, as you say!
Yup. Understood. That cogging effect – direct drive, so obviously not actual gear teeth cogging; presumably the poles of the drive motor passing each other! Unless it’s ball or roller bearings deteriorating, but that would probably be less regular & probably not n cycles per rotation. I’m used to flutter from worn rubber tyres in non-direct drive trains, and they’re fairly but not exactly regular, and definitely not n cycles per rev… 8~) …or from slightly imperfect capstans on tape recorders… |
jim lesurf (2082) 1438 posts |
For my direct drive I think there are more ‘poles’ than I see ‘cogs’. So it may be some kind of imbalance in the sequencing of the drive signals. IIRC it is a single point thrust bearing, so no ball races. You’ve set me thinking about using a normal music disc. In principle if you had an LP and a CD of exactly the same recording it might be possible to do a cross-correlation to extract the relative timing. I did something like that to measure BBC iplayer playout rates and propagation delays years ago. But it is a real booger of an analysis to do. And it needs a rare combination – and LP and a CD where they haven’t been significantly differently treated along the way. |
Clive Semmens (2335) 3276 posts |
Even with just one copy of a music disc, one can make some assumptions (not necessarily justified) about what pitch variations are musical and what are errors. But as you say, it’s a real booger of an analysis to do – only quite a bit more of a booger even than your suggestion. I was pretty impressed for a moment when I first saw this thread and thought you’d done it… 8~) Actually, turntable errors could conceivably be corrected by doing multiple runs with the disc rotated a few degrees between runs. But you can’t fix disc errors that way. |
jim lesurf (2082) 1438 posts |
IIRC Denon did once make a deck that auto-adjusted the center spindle location to reduce any ‘swing’ due to the LP being off-center. But it was a bit of a monster, mechanically. I doubt anyone ever bought one. |
Clive Semmens (2335) 3276 posts |
8~) I think I’d do it in software if I was going to do it at all today! But at least Denon’s machine would presumably have been registering by the edge of the disk? If so, why not simply omit the centre spindle altogether, and locate the disk on the platter by its edge? If the pressing is misregistered relative to the edge, you’re really stuffed either way. I rather like the idea of measuring the swing of the needle arm and zeroing that – then the first few seconds will wow, and all will be well thereafter. |
jim lesurf (2082) 1438 posts |
The Denon did use the arm movements to ‘center’ the spiral being played. I recall it being shown on “Tomorrow’s World”. But beyond that I have no idea how well it worked, or if anyone actually bought one! |
Clive Semmens (2335) 3276 posts |
Well, there’s no magical software required to do that – it could be entirely electromechanical with analogue electronics and servos, all stuff that was very well established yonks ago. No reason why it shouldn’t have worked extremely well if it was decently engineered. But unless it ran round a couple of tracks with amp off to centre it, then lifted the needle and went back to the beginning with amp on, you’d get wow for the first couple of rotations. |