advertisement


Turntable speed analysis part II

The Linn motor obviously does perform worse under light load, hence the bearing mod making the difference to smoothness. You can feel this by holding the motor with standard vs modded bearing.
 
If someone took a recording from a Xerxes we wouldn't have to rely on hypothesis...

The lower the load on the Linn motor the closer to 0 torque it runs - when up to speed, the load on the motor is extremely small - fluctuations in that load are only a small percentage of the average load....

It seems to me that the problem with an eddy current brake is that the resistive load is itself governed by the velocity of the platter and is therfore prone to being modulated by the very thing it is supposed to be governing.
 
Try again. I didnt realise on my previous attempt the output from the fm demodulation had clipped. Also after some reading I found a better bandwidth should be around 700hz.

I have also been experimenting with gnu plot.
spectrum5.jpg


Some peaks not shown at 60 and 120hz. Probably the power supply.
 
And my attempt at a polar plot. Not as clean by anymeans. But the 16 peaks corospond very nicely to the 9hz peak on the spectrum.

polar1.jpg
 
sq225917 said:
The Linn motor obviously does perform worse under light load, hence the bearing mod making the difference to smoothness. You can feel this by holding the motor with standard vs modded bearing.
Click to expand...

More load means more torque is required to achieve steady state which means a bigger magnetic field which means more current is needed from the power supply. I can't think of a reason why the motor would function better under those circumstances, so maybe it is the power supply that is able to function more precisely?
 
That is an FFT of the fm demodulated signal. Horizontal is the frquency in Hz, verticle is in -db, but is probably meaningless unless comparing a signal fromt he same process.
 
YNWOAN said:
It seems to me that the problem with an eddy current brake is that the resistive load is itself governed by the velocity of the platter and is therfore prone to being modulated by the very thing it is supposed to be governing.
Click to expand...

A drag (viscous fluid) based load will also be velocity dependent
 
If I do a FFT on edd9000's wave file, I get this frequency spectrum. The fundamental appears to be at 3145hz (Hanning window, width 1024) and there is a spectrum of harmonics (multiples of 3145hz) and a noise floor which drops with frequency. All of the wav files submitted for speed analysis look like this with some variations in the level of the noise floor and harmonics. The question is, what causes the harmonics? Is it harmonic distortion from the cartridge or due to speed variations?

edd9000_yamaha_ypd71.jpg
 
Those are just the harmonics of the tone. Cartridge, tracking, the orginal tone all conrtibute to those. I cant imagine how speed variance would alter the harmonic distortion.

What we are doing is FM demodulating the signal, using the 3150hz as a carrier. This give a waveform of the speed variance, which can than be looked as a spectrum to locate the peaks of modulation.
 
edd9000 said:
Those are just the harmonics of the tone. Cartridge, tracking, the orginal tone all conrtibute to those. I cant imagine how speed variance would alter the harmonic distortion.
Click to expand...

I was surprised at the strength of the harmonics, I was expecting a purer 3150hz sine wave.

edd9000 said:
What we are doing is FM demodulating the signal, using the 3150hz as a carrier. This give a waveform of the speed variance, which can than be looked as a spectrum to locate the peaks of modulation.
Click to expand...

I understand the process. Your carrier tone though is not precisely 3150hz, do you account for that? What affect do the higher order harmonics have on demodulation?
 
I use a narrow window fft to find the peak frequency, then use that as the demodulation center, I think I used 3149, with a bandwidth of 700hz. I assume that means the demodulation filter bandpasses the frequency so higher order harmonics wont have much effect.

As Understand it, if its slightly off in speed, and you fm demodulate around 3150, it would just shift the amplitude of the waveform, and not change too much on the fft or polar plots.
 
John Channing said:
More load means more torque is required to achieve steady state which means a bigger magnetic field which means more current is needed from the power supply. I can't think of a reason why the motor would function better under those circumstances, so maybe it is the power supply that is able to function more precisely?
Click to expand...

Actually after a bit more reading on this

"as the load on the synchronous motor increases, there is no change in its speed. But what gets affected is the load angle 'δ' i.e. the angle by which rotor axis retards with respect to stator axis."

This certainly changes the shape of the interacting magnetic fields (between the stator and rotor) and therefore could make rotation smoother.
 
Your carrier tone though is not precisely 3150hz, do you account for that?
Click to expand...
It turns out that FM demodulation doesn't care what the carrier frequency is. But I cannot remember if I had to pay attention to it to do the Hilbert Transform. Things seem to work fine regardless of the sampling frequency (if you didn't know then a 48k sample of 3150Hz is indistinguishable from 96k of 6300Hz.)

What affect do the higher order harmonics have on demodulation?
Click to expand...
Empirically sensible results come out regardless, but I usually narrow band filter the signal before demodulation to remove noise and put the various sources on a level playing field.

Paul
 
Not just could, but seems like it does. At least in my and Ynwan's cases it certainly does because our power supply is the wall, a transformer and a couple of phase caps, no active components.
 
John Channing said:
"as the load on the synchronous motor increases, there is no change in its speed. But what gets affected is the load angle 'δ' i.e. the angle by which rotor axis retards with respect to stator axis."

This certainly changes the shape of the interacting magnetic fields (between the stator and rotor) and therefore could make rotation smoother.
Click to expand...

Yes, I agree. When considering the electrical theory one must also consider how a motor actually mechanically works.

Assuming there are load variations within the drive system, consider how those variations may interact with a motor running at near zero torque.
 
A fresh copy of the Ultimate Analogue Test LP arrived today so here is my 3150hz file. I had a quick look at the output and my deck appears to be running very slightly fast, but I would be very interested to see the polar plot.

Turntable Details
Linn LP12 Black Ash 1989, Cirkus + Trampolinn
Linn Lingo mkI 1991
Linn Akito mkI 1991
AT OC5
 
I'd be interested to see which is more linear, viscous drag or eddy current, I'd expect the eddy current brake is far less linear in terms of drag/speed. I'd also presume viscous drag is closer to being near constant for a fixed rotational speed, but I'd like to see some numbers.

But I do think some drag is better than none, and not by a small margin.
 
You must log in or register to reply here.


advertisement


Back
Top