With the very noisy data recorded with the Bode 100 that I kindly received from Jack, see Cartridge dynamic behaviour, I have compared these data to my LTspice model: (0.46mH + 800 ohm) //800K//3.6pF.
The images below are showing both Phase and Inductance the same way as the HP3577A does.
Results are very comparable, telling that the LTspice model is quite close to the ATM540L.
Hans
The images below are showing both Phase and Inductance the same way as the HP3577A does.
Results are very comparable, telling that the LTspice model is quite close to the ATM540L.
Hans
Attachments
At least we only need 200kHz or so, HP's accessories will all be intended for the full instrument BW.
Jack,
What about my suggestion Cartridge dynamic behaviour.
Are my assumptions correct and is this possible ?
Hans
Hi Hans yes that gets the excitation level on the coil small enough. That leaves whether the phase detector can work well enough at such levels to fish phase info out of the mire. It might but prob wont and is easy to try so why not?
LD
Hi Davidsrsb if it happens at all it will be at small levels by programme standards.
LD
Hi LD,
According to the image below, with 47dB SNR, phase error will be below 1 degree and magnitude error below 0.1dB.
Hans
Attachments
In the table below, one can see what 100% variation of the Cap and 10% variation in L does to the phase at 3kHz and at 10kHz.
But because phase is close to 90 degrees, a bit of noise on the phase doesn't hurt to calculate the imaginary from the complex modulus |Z|.
Maybe it would therefore be even better to measure at 10kHz instead of 3kHz, because at this frequency phase is closest to 90 degrees, giving probably the highest accuracy in calculating L.
Hans
Attachments
Hans with all due respect this statement is nonsense, the modulus is real not complex and you don't calculate the imaginary from something, you measure magnitude and phase Euler's equation gives you the direct correspondence to real and imaginary part.
Scott,
And with much respect from my side, the magnitude that is measured is |Z| and is real, being by definition the complex modulus of Z.
And yes, with the phase you have to calculate from |Z| the re and the im magnitude.
But the magnitude from im1=|Z|sin(88 degrees) differs only 0.04% from im2=|Z|sin(89 degrees).
That's why I meant to take a frequency where the phase is closest to 90 degrees as possible, which is the case at 10kHz, getting L with the highest accuracy.
Hans
Sorry Hans we just have some kind of terminology problem, complex modulus makes no sense (to me). The modulus (magnitude) is a real number there is no imaginary part. It just makes a layer of translation for me to see what you mean.
Hi Scott what Hans is saying is that the real part of Z at 10kHz is << imaginary part. So magnitude of Z tells us what we wish to know ie magnitude of imaginary part of Z.
That's true until we add the loss mechanism we are looking for. Then vector isn't 90 anymore so all bets off. I think we do need to measure phase unfortunately.
LD
Hi Scott, no problem, but obviously I'm not the only one using this terminology Complex Modulus
Complex modulus - Wikipedia
Hans
Hi Jack,
That would be a shame, when resolution turns out to be the deal breaker.
The HP3577 spec's are a bit vague on that, but they mention that at levels between -90 dB to -100dB below Full Scale Input, resolution with 10Hz BW is still +/-0.75dB.
When taking +15dBm or 1.26V as Full Scale input, -100dB would be 12.6 uV.
That is still- above the 22uV that I was suggesting with the -40dBm setting + 40dB attenuation with 5K - 50 Ohm.
So do I make a wrong interpretation of the spec's ?
Hans
Yes I apologize again for mis-understanding. The observed measurements presented sometimes show 10 or more degrees of phase error so direct application of minimum phase theory would possibly miss what we are looking for.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.