Turntable speed analysis

Paul R · #1 18-10-09, 06:27 AM

A couple of weeks ago I came across this blog post and my interest was piqued.

In essence if a turntable plays a recording of a pure tone then any variations from a constant speed will manifest as a change in the frequency of the tone. If the tone is demodulated then what comes out is a moment by moment measurement of the instantaneous rotational speed of the platter.

It turned out that this book contained everything one needed to know to implement digital FM demodulation.

A few evenings later I had a couple of crude programs. One applies FM demodulation to a mono wav file, the other applies quite large FFTs to the result and writes a text file suitable for plotting.

Given that numbers are coming out how does one know they are sensible? I generated an FM modulated test signal that appeared to demodulate plausibly and then moved on to making some recordings of a test record.

As no record is completely concentric, nor completely flat, we would expect modulation at the rotation rate and at some multiples. For 33rpm this is 0.56Hz and multiples, for 45rpm 0.75Hz and multiples.

At present I only have an LP12/Lingo. It occurred that I have two ways of making it got at 45rpm, either with the Lingo driving the motor proportionally faster or using the original big pulley adaptor and having the motor go around at the standard rpm.

With the standard 50Hz drive the motor turns at 250rpm/4.17Hz. For 45rpm this has to rise to 337.5rpm/5.625Hz.

After making recordings in the two 45rpm states, processing the data and plotting the results produced,

which rather gratifyingly meets the expectations.

More later. And if anybody is prepared to make a recording of a tone it would be interesting to add.

Paul

sq225917 · #2 18-10-09, 08:52 AM

Any chance of a larger version.

per-Sony-fied · #3 18-10-09, 09:48 AM

What does it all mean? I can see the variation of the test signal reduces with frequency & is very stable between 5 - 10Hz & to some respects using the 45rpm adaptor produces better results but what is the actual variation in tone & how does this compare to other TT's?

Paul R · #4 18-10-09, 12:09 PM

Quote:

Any chance of a larger version.

Click on the thumbnail.

Paul

Paul R · #5 18-10-09, 12:24 PM

Quote:

What does it all mean? I can see the variation of the test signal reduces with frequency & is very stable between 5 - 10Hz & to some respects using the 45rpm adaptor produces better results but what is the actual variation in tone & how does this compare to other TT's?

What we're looking at is the spectrum of the speed variations. A Wow and Flutter instrument adds all these together and gives you a number, this approach perhaps allows identification of causes of variations. So we can see the motor pulley causing a small wow, and the Lingo version is quite a lot better. Whether this is audible is moot, and in this case not very interesting.

What would be interesting is Basik/Armageddon against Valhalla, Lingo and Radikal. Or base Technics SL12x0 against 'TimeStep' modified. Or various settings of the Rega phase adjustment.

What would be especially interesting would be to find a way of recording the speed signal while also playing real music. So we could look at the detail of transient wow, amongst other things.

Paul

PigletsDad · #6 19-10-09, 02:22 AM

A simple way to simulate the dynamic wow is to add extra drag - for example, get one of the horrid cleaner brushes, and drop that down on the running record, while monitoring the tone frequency.

What is the normalisation of the vertical axis?

sq225917 · #7 19-10-09, 05:11 AM

paul i have to download the file from mediafire, i can't view it in IE or chrome

Paul R · #8 19-10-09, 10:18 AM

Quote:

A simple way to simulate the dynamic wow is to add extra drag - for example, get one of the horrid cleaner brushes, and drop that down on the running record, while monitoring the tone frequency.

I think something like this could work. But it would be nice to find out the magnitude of the real effect.

Perhaps record a saturated signal on a tape, then wrap the tape around the platter and use a head in close proximity to pick it up...

Quote:

What is the normalisation of the vertical axis?

The numbers that come out of the demodulator are proportional to the actual frequency. For 3150Hz it's about 0.81. So I've been removing the offset and working directly with what remains. This avoids a huge value at 0Hz in the FFT. What the numbers actually mean is a whole other thing. I'm away from home for a day or so and I don't have access to the raw data.

Paul

Paul R · #9 19-10-09, 10:20 AM

Quote:

paul i have to download the file from mediafire, i can't view it in IE or chrome

I've temporarily lost access to my web space, but the Mediafire option seems to work quite well. The graph is just a screen shot and I don't think the subtleties are significant.

I can email you the original on Weds if you like.

Paul

pure sound · #10 19-10-09, 11:04 AM

3 kHz tone played by an EMT938.

http://www.mediafire.com/?tfdmztm3num

Ideally you need a deck with 2 arms, one playing the test tone the other playing some heavily modulated material elsewhere on that side of the disc. Its difficult to imagine how you could do that to a suspended deck like the LP12 though. There isn't room for the 2nd arm. I suppose a second cartridge could be held by an outrigger attached to the headshell of the arm that is used and the necessary adjustment be made to the c/weight to get the tracking force right for both cartridges. Certainly the HFN test LP I have has other modulated bands that could be played alongside the test tone.

Paul R · #11 19-10-09, 01:29 PM

Ideally we want a recording of someone playing a piano loudly, and this doesn't seem popular with test record manufacturers.

There is a possibility to observe subchassis wobble and other wo(w/e)s at the start of the tone when the drag jumps.

I'll process your tone tomorrow night and let's hope something interesting happens.

Paul

Robert · #12 19-10-09, 01:49 PM

Quote:

Originally Posted by pure sound

3 kHz tone played by an EMT938.

http://www.mediafire.com/?tfdmztm3num

Ideally you need a deck with 2 arms, one playing the test tone the other playing some heavily modulated material elsewhere on that side of the disc. Its difficult to imagine how you could do that to a suspended deck like the LP12 though. There isn't room for the 2nd arm. I suppose a second cartridge could be held by an outrigger attached to the headshell of the arm that is used and the necessary adjustment be made to the c/weight to get the tracking force right for both cartridges. Certainly the HFN test LP I have has other modulated bands that could be played alongside the test tone.

I've an Ortofon direct cut test disc full of test tones for various things.
I think some are on the outer grooves, so you could place a 45 rpm single over the top and play it with another cartridge.
Bodge but it might work.

Paul R · #13 20-10-09, 01:26 PM

Here is a straight comparison of an LP12 file and the EMT, between 0 and 200Hz.

Green is LP12, red EMT. The EMT has a lump at 6-8Hz and big spikes with sidebands at 25 and 100. Very interesting. (I'm not sure the LP12 50Hz spike is real, I think it may be electrical. Neither of my 45rpm recordings show it.)

Speculating, I'd not be surprised if the 25 and 100Hz modulations were audible. The pitches will get fuzzed up.... I'd also predict that the SP10 will be different.

Paul

pure sound · #14 21-10-09, 01:22 AM

I'd add that I'm not convinced this HFN record is centred perfectly. When playing, a cyclical component is audible with a resultant 'fuzziness' at some points. Whether that lack of centricity contributes to the peaks in these results, I'm not sure. I'll see if I have another record with tones on it anywhere.

I'll also try it on the SP10 when I get my MC transformer back from a loan.

Paul R · #15 21-10-09, 02:36 AM

I had a look at the EMT938 service manual. Mostly in German. Nothing is much clearer...

100Hz is 180 pulses per platter revolution, so 25Hz is 45. I think it's a 2 phase motor. So perhaps 22 poles/2 phases gets us near enough.

It would be interesting to go through the service/test procedure and see if anything changed.

Paul