If one looks at how Q is measured (actually computed, not directly measured), in most quality modern bench type LCR meters, it's no surprise that high Q values are questionable.
The mentioned LCR measurement usually involves some form of "synchronous sampling" to reduce the effects of noise and unwanted signals, also with some averaging, and utilize high resolution accurate ADCs (even the handheld DER-5000 has a equivalent 4.5 digit ADC, the TH2830 and IM3536 have 5.5~6.5 digit eqv. ADCs). The actual measurement is from the voltage and current of the DUT, where the current is deduced from a ~ virtual ground port Lcur, and sensed by Lpot, which scales and computes the DUT impedance with Hcur and Hpot values.
With Q the absolute value of Tan of the DUT Impedance Z angle, and the sensitivity of Q can be accessed as the derivative of Q wrt the DUT Impedance angle or dQ/dtheta = 1/[Cos^2(angle)], and the Cos function is most sensitive (slope) to the angle at +-90 degrees, and it's squared
Then a Q of 1000 results in an angle of 89.9427 degrees, and a measurement/computational error of 0.1% results in a Q assessment of ~390 (sensitivity of 10^6), not 1000, and a simple linear error assessment would assume a Q of 999 or 1001, not 390!
Using this same conditions a Q of 100 would result in an angle of 89.42706 degrees, and a 0.1% error would be a Q of ~86.5 (little better, sensitivity of 10^4), and a Q of 10 would have an angle of 84.28941 degrees and with a 0.1% error a Q of ~9.854 (much better, sensitivity of 10^2).
This is why we don't use "Q" much to define or represent a single component, but for filter use seems to be a much better parameter as the
BW/Fc Fc/BW.
Anyway, as we older "seasoned" folks have learned, "Know Thy Instrument", or "THI" for quality, reliable, and repeatable measurements
![ThumbsUp :-+](https://www.eevblog.com/forum/Smileys/default/icon_smile_thumbsup.gif)
Best,