Impedance Comparisons between different measurement systems (question)
I set up an impedance measurement. Same jig, same amp, same computer. Quiet environment. I measured with ARTA(LIMP) and with Soundeasy (EasyLab) in both Pink Noise and Sine Modes. I did it in the following order to account for any heating up of the coil; LIMP-pink, LIMP-Sine, Easylab-MLS, EasyLab-MLS, LIMP-Sine,and then LIMP-Pink. I set levels to account for voltage outputs at approximately 850 mV peak. This was adjusted via the software, the amp was never touched. The Fs was consistent (within a few tenths of a hertz) within each given signal but I found that the Sine wave produced the lowest Fs followed by the Pink Noise signal and then MLS. The Sine wave was centered at 100 Hz.
The differences were by 2-3 Hz between approaches.
Let me stress that there were no environmental variables and signal strength was comparable. The curves were clean and consistent with the phase shift.
I welcome any comments or suggestions as to how to best identify which is most accurate.
I would think that the changes of Fs (Qms as well, btw?) you saw, within that order of difference magnitude, might dominantly come from different large signal conditions near Fs during the measurement. The history of the large signal conditions a few seconds and less before the measurment is also important.
Bass/mid drivers tend to shift Fs (and change other T/S parameters also) with different excursions, and things like DC-offset, creep of the suspensions and all the other nonlinear factors change the linear approximations (=T/S-params) at the operating point given by the setup and method (for example; current drive vs. voltage drive).
You might try to measure impedance with the sine method at various signal levels and look if you find changes of Fs etc in the same ballpark. With very low signal some drivers with high and unstable internal friction also show shift of parameters.
If I am right in the assumption that the highest signal level at Fs will be with the sine, pink noise being lower and different in nature (more like averaging a range of levels) and finally MLS with a white spectrum will be lowest, then the increase of Fs that you saw in that order could indicate that the driver resonates lower when pushed harder (or higher with very low signal).
Or the other way round, take an LRC-network that mimics a driver impedance and is linear. If the shift of Fs is still there, then you have made a really interesting discovery.
Thanks, Klaus. I will try the LCR and measure it as I did the driver.
Regarding large signal history, I did not expose the driver to any large signals. It was exposed only to the signals of 850-875 mV (peak).
I am aware that more excursion will drive the apparent Fs lower; it sounds as though you are suggesting that though the voltage was standardized, the concentration of power would be greatest for the sine wave (since each wave would have an equal concentration of power), then for the pink noise (since there would be more "noise power" at the lower octave ranges, and then for white noise, with equal "noise power" in those ranges. If this is true, then I would expect to see similar results using the LCR circuit; and if I do, how does one really know the Thiele Small values as they will be variable based upon the type of signal used.
My suspicion is that the Sine based signal would be most consistent and accurate as long as it is not pushing the driver out of a linear mode.
Yes Jay, I also would think that the sine will be the most stable and acurate method (arbitrary precision of frequency resolution), also because the math involved in the non-direct methods will only allow for a limited, finite frequency precision at LF (the low FFT-size and high sample rate effect).
I'm looking forward to the result of the LCR test as it answers the most basic question; is it the driver or the setup/method?
From a practical standpoint, I don't think it is overly important to measure T/S with less than 5%..%10 systematic tolerance except for real scientific work and QC. Of course it is always good to know your error terms and system dependencies, no question.
Well, I measured an LCR circuit with all three set ups. The MLS method, Sine and Pink noise methods. They all identified roughly the same point of resonance (within 0.05% of one another). The impedance at resonance, however, indicated that methods of LIMP (Sine and Pink) were right on target (2.7 and 4% variance from my multimeter) while Soundeasy's MLS was 30% off. This suggests to me that my calibration process may have been flawed for Soundeasy and I will try to re do the measurements tomorrow. If they are consistent, however, it does raise some concern regarding the MLS measurements with SE.
I went back and did some more comparisons.
White Noise (set to max levels as recommended by Soundeasy):
Fs 51.4 done once Fs 51.6 with 10x averaging
Pink Noise (set to -3dB, which was the max prior to the "red zone"):
Fs 51.3 done once Fs 50.6 with 10x averaging
at -3dB 48.7 at -20dB 50.2
There is an interesting discussion regarding this going on at Partexpress Tech Board, as well. But this is suggesting that some of the difference may well be the focus of energy at a particular frequency with a greater amount leading to a lower Fs reading. Obviously, white and pink noise will both be a little more variable based upon where in the packet the energy is directed.
|All times are GMT. The time now is 12:30 AM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Copyright ©1999-2016 diyAudio