advertisement


Turntable speed analysis

Status
Not open for further replies.
In that relation 'omega (F)' is the instantaneous phase, which should be in radians per period. Which can be converted to instantaneous frequency by multiplying by the sample rate and dividing by 2Pi and hey presto, the desired output.

OK, that makes sense.

I optionally bandpass the input signal at the measured centre frequency, use an FIR Hilbert Transformer to generate Q,..

This is confusing to me - why 'optionally' - the primes are the bandpassed quantities, right? So shouldn't be optional. I was thinking the bandpassing of the input signal (I') would be done in frequency space using FFTs to go back and forth, i.e., f'(M(f(I))), but I think you are saying it can also be done with Q, i.e., Q'(M(Q(I)). Is that right. Finally, since Q produces a complex number, are you just looking at the real part for omega? I realize I'm using ' to denote both bandpassing and inverse - sorry about that.

cheers,

Dan
 
AFAICT the maths makes no bandwidth requirement of the derivatives, so the comment in the source blog may be something to do with his use of LabView, a product I know nothing about.

IIUC 'Q' isn't complex, I and Q are the real and imaginary parts of an analytic signal, the relation you quoted is a clever way of avoiding an inverse tan operation on the ratio between the real part and imaginary part. 'i' gets dropped out somewhere in a way a real mathematican could justify. So the output from the equation is a real number representing the instantaneous phase.

Originally I was bandpass filtering in Audacity before processing, which is very boring, I found that sensible results are produced without any filtering, perhaps because the modulation depth is very small. But I put a bandpass filter into my code for completeness and I think (although I need to go back and check) that it reduces the crap in the region of output that interests us.

Sorry for being vague, I used to have the capacity to properly understand this stuff, now I tend to grab and go.

Paul
 
Hi Paul, I wonder if you could perform another plot of my PL-71?

http://www.mediafire.com/?oo7uuedttsd022z

I've tweaked the PSU some more and added filtering and better pre-regulation of the supply. Subjectivley the deck sounds a whole lot better than with my previous effort. Would be interesting to see if there is any visible change in the plot.

Cheers.
 
Neal's modded PL71, red is the original file from some time ago, green the new one,

nealg_PL71_NewPSU.png


and

nealg_PL71_NewPSU_polar.png


Looks to be an improvement. Need to figure out what's what, everything on a DD is (by definition...) sync'd to the platter speed. I have to look back at the earlier posts to see if we worked out how many poles the motor had.

Paul
 
sq's Kuzmaesque with Geddon update and drag,

sq_geddon_recap.png


and polar,

sq_geddon_recap_polar.png


It's nice to see the recording's flaws show up consistently. I think the demodulated spectrum shows an improvement in the artifacts around the motor speed. The three red peaks between 4Hz and 6Hz are at effectively 250rpm, 300rpm and 333rpm(!). The polar looks smoother or at least more consistent. The speed has dropped by about 0.4%.

Paul
 
Status
Not open for further replies.


advertisement


Back
Top