advertisement


Turntable speed analysis

Status
Not open for further replies.
Hi , fascinated by the speed discussion...
3kHz , the peak aural sensitivity frequency for flutter, FM modulation of pitch.
Should be as clean as possible but the measurement is usually psychoacoustically weighted by pulse shape and frequency filtered factors. see Wiki on wow and flutter.

I have published some of this material in Vol 3 no 3 HIFICRITIC in trying to chase down how the the Radiakal motor alters/improves the residual noise spectrum from the LINN LP12. That noise may contains many components some amplified, even misrepresented by the connected mechanical structure.
There will in any case be some of: induced electrical noise, RF , electrical hum components, 50Hz , often 100Hz second harmonic dominant, plus many higher harmonics, the chassis resonances in the 4Hz region, the arm-cartridge resonance. Then come eccentricities for the drive pulley, the platter, and the disc, which may also have warp wow components , never mind thickness, diameter variations for the belt or idler pulley. And finally the contributions from the motor itself , bearings, cogging , rotor balancing, electrical frame vibration. All grist to the mill . Martin C
 
Martin,

I think it would be straightforward to produce a standardised W&F measurement from the output of my tools, but I'm more interested in seeing what is going on rather than just producing a number.

What really started this process was when someone on here posted two short clips, both made on an LP12, one using a Lingo the other using a 'Mose', which is a Valhalla clone in a box, but with added 45rpm. When these recordings were subtracted from one another with 'Audio Diffmaker*' there was a clear 17Hz in the residual. As this is the difference in drive frequency between 33 and 45rpm I wondered whether one of the PSUs was allowing the 45rpm generator to pollute the 33rpm output.

The FM demodulation technique should show this directly.

Paul

*If you haven't seen it, Audio Diffmaker is at http://www.libinst.com/Audio DiffMaker.htm
 
The main annoyance is I've never seen a perfectly concentric test record. I've got a copy of HFS69, HFS75 and the current Hi-Fi Choice and they all are eccentric to some visually and sonically noticeable degree, and in the latter case warped too. A nice quick way of assessing speed (or how eccentric your test record is) from a pitch perspective is to use a chromatic guitar tuner. They are capable of indicating the slightest pitch deflection. I just plonk mine on the floor in front of the speaker and it clearly indicates just how crap my test records are!

Tony.
 
Here is my EMT930 playing The Ultimate Analogue 3.15kHz testone:
Click to expand...
I'll get onto it tomorrow.

The main annoyance is I've never seen a perfectly concentric test record.
Click to expand...
The eccentricity generates a distinct 0.55Hz wobble, which can hopefully be ignored as we know where it's coming from.

I am wondering if it would be possible to use software to de-eccentrize needledrops. Possibly completely automatically.

Paul
 
phixphi,

I've forgotten where I was with some tweaks to the software, so the numbers may be askew. I'll re-run my LP12 recordings when I can find them.

Anyway I see speed variations at 4.38Hz and rather stronger at 5.56Hz. Generally it looks pretty good. It would be interesting to know what the motor rpm and idler rpm are.

More later.

Paul
 
Had a play with looking at polar plots. So the distance from the origin represents the speed at that point in the rotation. The plots cover about 5 consecutive revolutions. Quite interesting. The numbers do not mean anything, the curves are very exaggerated. Here are two plots, for the EMT and for my SP10. Due to the different sampling rates and some obscure DSP stuff the scales differ, so this isn't absolutely comparative. The red circles represent perfection. Anyway SP10,

SP10_Polar.jpg


and EMT,

EMT_Polar.jpg


The EMT looks a little 50p piece. I wonder whether that's the idler? It may well be within spec of course.

Paul
 
Interesting idea Paul, but surely if it was the idler, the variation would be at a higher frequency and not synchronous with the platter? It could be imperfections in the platter rim though.
 
More likely in retrospect. Perhaps the lathe in the factory had a slightly dodgy ball in its bearing.

I don't know what the ratio of idler to platter is. It would be nice to connect something to physical reality. The SP10 is very interesting, and worth further investigation. I'll rerun Guy's samples.

Paul
 
Well the idler on my 927 is approx 50mm diameter and I think it's the same part on a 930. Obviously the platter diameter is different, but I suppose we could guess at 300mm for a 930.

So that would be about a 6:1 ratio.

It's a synchronous motor (of some description) so motor speed must be a multiple of 50Hz.
 
If the ratio was very nearly 1:6 and the idler wasn't perfectly circular then that might fit.

I have another EMT recording from Puresound that I'll exhume.

Paul
 
Paul, thank you for the analysis!
Would it be possible to take a look
at a conventional spectrogramme?
What do you mean by "The EMT looks a little 50p piece"??

According to one of the EMT experts, the idler frequency should be 2.72Hz,
so 5.56Hz is close to 2x that. Then, according to the same guy the motor is 27.77Hz,
but I seem to see it at almost exactly 25Hz.

Patrick: I started to have the same thought and blame the acrylic subplatter.
Anyway, the deck is on the way to HM for a checkout as I had an accumulation of
problems (including freshly rebuilt TSD15) I could not resolve.

Cheers,
jk
 
Would it be possible to take a look at a conventional spectrogramme?
Click to expand...
I'll put it up tomorrow.

What do you mean by "The EMT looks a little 50p piece"??
Click to expand...
The picture shows the speed against platter angle for 5 rotations, if you squint at it it looks a bit like a 50p piece, although I think it is more 5 sides than 7. 5 sides fits a 2.72Hz idler quite nicely.

There's no obvious sign of a rotational problem. In my limited opinion.

Paul
 
Paul, I'm sorry for my weak English, but what is a "50p piece"?

The problems I've encoutered were:

- seemingly excessive drag in the main bearing;
- 2x w&f, but here the test record may be the reason, or the cart-arm resonances, or the subplatter, or....
- freshly rebuilt TSD15SPH mistracks and distorts on louder complex passagies, something similar
to what is described here:

http://www.audioasylum.com/audio/vinyl/messages/79/793955.html

cheers,
jk
 
A 50p piece is a 7 sided UK coin. http://en.wikipedia.org/wiki/Fifty_pence_(British_coin) Sorry for the assumption of local knowledge...

Anyway here are the spectra of the speed variations,

Against my SP10, narrow band,

EMT_fft.jpg


The 0.55 and 1.1 bumps are eccentricity, and common to both traces so some of that is in the record. The big bump at 2.77 is presumably the idler.

If we look deeper, and wider,

EMT_fft_wide.jpg


Now using a log scale, so small bumps are exaggerated... Anyway there's something at 24.84 but I'd be very wary of assuming this was mechanical. It also appears in a spectrum of your recording.

Paul
 
The other EMT test tone I supplied came from a 938 which is direct drive rather than idler.
 
Paul, you are the guy!
Thank you! The 25Hz: when I lift the
tonearm, it disappears. Why do you think
it's not mechanical? I've been suspecting the motor,
I see it on a scope either pure (empty grove)
or modulation (test signal). Your method clearly shows that my audiotester is not good at few Hz range. On the other hand I see some complex bump
6-12Hz, probably the arm resonance which seems missing at your plot?
If the deck survives OK the there-and-back shipping,
I can record again the same track after the checkout. I can also
record the test track from the Clearaudio Trackability LP to compare the both.
Will your software be available in some form?
Adding DIN-weighted w&f calculation would be great, as this is what
most manufacturares state.

Cheers,
phixphi
 
I think the 25Hz is unlikely to be a true component of the wow. I might be wrong.

The arm resonance doesn't appear in the plot because this is looking at the rotational speed implied by the recorded signal and not the signal itself. In fact the 3150Hz signal is bandpass filtered before further processing so anything as far away as 10Hz is eliminated.

Computing a W&F figure involves some maths to scale everything correctly and then a weighted sum to get the figure. The trouble is checking that it's right. I was interested in the mechanics of the drive and how that might be seen in a signal. So when you 'upgrade' the power supply was there are measureable benefit. That sort of thing. We shall see.

Paul
 
Status
Not open for further replies.


advertisement


Back
Top