Measuring complex impedance of an ultrasonic transducer

[The_Dutchman]

Hello all,

For my thesis i'm designing a measurement device for measuring complex(real and imaginary part) impedance of an ultrasonic transducer.
The impedance should be measured from 1MHz to about 10MHz. The expected resonance frequency is about 5MHz where the sensor has an impedance of 150 Ohms.

For signal generation I am using an AD9834 DDS chip that can generate sine waves up to 75MHz.
My idea is to use the LM7171 opamp as driver. The ultrasonic transducers have a maximum average power of 0.125W.
I was planning to have a measurement range of about 50 Ohms => 1000 Ohms.
So if the impedance is 50 Ohms, and I use a drive voltage of 3Vampl => Iampl = 0.06A => Pampl = 0.18W => Pavrg =127mW
However I am neglecting phase shifts here. As the impedance increases I can also increase my voltage to the LM7171 its limits.

The ultrasonic transducer needs to be grounded, so the only method to measure the current through the sensor is with a shunt.
Herefore I would like to use 3 LM7171 amplifiers in an instrumentation amplifier circuit. Hower the voltage may not be to high, else the LM7171 cannot withstand the common mode voltage.

Now my questions, can I use 3 LM7171 amplifiers in an instrumentatin amplifier circuit to measure the current with the above specs?
Lastly, I am planning to use an AD8302 gain / phase detector, but I am wondering how I should connect the instrumentation amplifier to the input ? The input seems to be very low ? -74dBV = 223µV ???

Thanks in advance

 

[FvM]

For the current measurement, you should also consider differential amplifiers like AD8131 and AD8132. They achieve a common mode rejection that won't be easily achived by regular instrumentations amps.

I see that AD8302 is a real wideband and high dynamic device, at the expense of only moderate accuracy. I guess that it may be suffcient for your application, but you'll need at least a good calibration.

A regular vector network analyzer from a lab will be able to perform the intended measurements by pressing just a button, by the way.

 

[The_Dutchman]

Yes, but the purpose is to make a small portable device. Can you give any advice on connecting the signals to the AD8302 ?
I've also found something like this:
USB VNA

Maybe it is possible to adjust this for my purpose ?
Will the complex impedance be dependent on the excitation power?

 

[mtwieg]

Yes, but the purpose is to make a small portable device. Can you give any advice on connecting the signals to the AD8302 ?
I've also found something like this:
USB VNA

This would likely suit your needs, as long as you don't need great accuracy and dynamic range. The designer was clever enough to find a way around the poor phase measurements of the AD8302 near 0 and 90 degrees... the accuracy of the whole system is problably limited by his directional coupler as much as the detector itself.

Will the complex impedance be dependent on the excitation power?

It's possible for mechanical transducers... but if you only want to find the resonant frequency of a transducer, then I doubt it will matter much. Only one way to find out though....

 

[The_Dutchman]

With everything I read everybody is working in dBm, how do you know when you are in range of the AD8302 ?
Can I measure the real and imaginary part with only one AD8302 ? Or do I need to use 2 AD8302's with each time a reference signal ?
I'm not familiar with the theory about s-parameters yet. But do I need this ?
So yes wich S-parameters do I need to measure, and can this imply simplifications ?
Also do I need a directional coupler to interface the AD8302 or are there other solutions ?

Accuracy is not very very important, but better is better ofcourse.
As I understand it right now, the VNA's are using the mismatch in impedance to determine the load impedance by doing calculations with the s paramters from the reflection ? I can understand that for RF signals f > > > that you need this. But with a bandwith of 10MHz ? That's not RF right? Aren't there other solutions I am missing?

Thanks in advance

 

[The_Dutchman]

Hey,

I'm working out a schematic to use the AD8302 to measure the VSWR and the phase. With these 2 parameters it should be possible to determine the complex impedance. I am working with a 50 Ohm system, but to measure impedances of 1000 Ohm's, the VSWR will be 20:1. Is this possible with a directional coupler ?

Thanks alot,
Greetings

 

[FvM]

In my view, the VNA approach would be reasonable if you have one already in your lab. The 1:20 impedance ratio won't be a problem at these rather low frequencies with a standard open/short/load calibration.

But i you want to design an impedance measurement instrument for frequencies up to 10 MHz, I would rather go for a standard current source + vectorial voltage measurement.

 

[The_Dutchman]

A current source at 10MHz seems not so simple to me. How would you measure the magnitude and phase then ?
I would at least need an FPGA to do this, because sampling a 10MHz signal (at about 40MHz then) with a standard 8-bit controller is not done.
I would also measure the reflections because of impedance mismatch.

 

[The_Dutchman]

Hello,

I just calculated some values for the following schematic ( I don't know if I forgot some matching?) Also, what do you think of the calculations/schematic?


These are the calculations. As I can read from the datasheet, it has an accuracy of 0.1dB within 20dBm under -30dBm and 15dBm above.
I calculated the return loss(dB) and substracted it from the incident power on the load and than made a plot off impedance vs difference in dB.

Z0 = 50 Ohm Incident 9,25
Zl Reflection coefficient VSWR Return Loss dB Reflected power
50,0000 0,0000 1,0000 Inf delta
75,0000 0,2000 1,5000 13,9794 -4,7294 4,4370
100,0000 0,3333 2,0000 9,5424 -0,2924 1,5980
125,0000 0,4286 2,5000 7,3595 1,8905 1,3389
150,0000 0,5000 3,0000 6,0206 3,2294 0,9151
175,0000 0,5556 3,5000 5,1055 4,1445 0,6685
200,0000 0,6000 4,0000 4,4370 4,8130 0,5111
225,0000 0,6364 4,5000 3,9259 5,3241 0,4041
250,0000 0,6667 5,0000 3,5218 5,7282 0,3278
275,0000 0,6923 5,5000 3,1940 6,0560 0,2715
300,0000 0,7143 6,0000 2,9226 6,3274 0,2286
325,0000 0,7333 6,5000 2,6940 6,5560 0,1952
350,0000 0,7500 7,0000 2,4988 6,7512 0,1687
375,0000 0,7647 7,5000 2,3301 6,9199 0,1472
400,0000 0,7778 8,0000 2,1829 7,0671 0,1296
425,0000 0,7895 8,5000 2,0532 7,1968 0,1150
450,0000 0,8000 9,0000 1,9382 7,3118 0,1028

There is a 0.1dB difference until an impedance of 275 Ohm.
This is a very narrow impedance range. Are my calculations correct?
Are there any solutions?

Thanks in advance

 

[FvM]

I find it difficult to estimate the expectable impedance accuracy limits imposed by the AD8302 specification. I'm under the impression, that a linear detector wuld achieve better results. As an aditional point, how do you determine phase sign?

Your assuming, that phase sensitive measurement would imply FPGA logic. But there are different methods to implement it.

 

[The_Dutchman]

With the AD9834 (DDS) I also have a register for setting the phase. As known, the phase in the region near 180° and 0° are not very accurate. So if my ADC reads voltages within the range of 20° near 0 or 180° I will shift 40° further. And will use that phase substracted by 40° to do the measurement. The phase sign is also determined by shifting. Because if you shift, the voltage can go up or down corresponding to the phase sign.
I find it very good explained in this document:
www.wetterlin.org/sam/PhaseShift_AD8302.pdf

If I use a AC voltage source(then the voltage shouldn't be measured).Then I need a shunt and an instrumentation amplifier to measure the current. The current will be out of phase and have a different amplitude according to the connected impedance. Important here is to have an instrumentation amplifier with a bandwith from 1-10MHz and with a high common mode rejection ratio and high common mode inputs. LM7171 ?
If I use a AC current source (then the current shouldn't be measured) but over a range of impedances, there also should be a good compliance. Also high linearity over 1-10MHz. For measurment accuracy.

If I want to do this, I need something to measure the phase and magnitude between V and I and output an analog voltage. Measuring phase might be doable by using zero cross detection and shifting ? How about magnitude ?
Maybe I could still use the AD8302 to do this job?

Maybe for the the magnitude detection I could use a precision rectifier to make the RMS => DC as I am always using sine waves with hopefully not to much harmonics

 

[FvM]

I don't see how shifting the DDS phase would change anything in your setup, because it's not modfying the relation of forward and reverse wave seen by the directional coupler.

An impedance measuring setup must not necessarily use an ideal current source. It can also use a series resistor of e.g. 500 ohm and eliminate the source output impedance in the calculation. You're doing basically the same when calculating impedance from S-parameters. As previously said, I would refer to phase sensitive detection, using quadrature (0 ad 90°) reference signals. That's the standard method of vector network analysis.

 

[The_Dutchman]

Yes you are right, the phase shift won't do anything. Cause everything shifts.
The only S-paramater I need is S11, but if I understand correctly it is like the USB VNA, where I use reference signals so it is possible to shift the reference phase 90° to what the coupler sees. But will my range of measurable impedances increase? My impedance is grounded, so it is a one-port measurment so I don't need the RX side in my opinion ?

EDIT: **broken link removed**
At page 2 my calculations get confirmed that its only any good for impedances close to 50 Ohms. Need to use an other pinciple I think. I-V method or RF-IV method.

 

[FvM]

An industrie standard VNA achieves e.g. +/-2.5 % magnitude and +/- 2° phase uncertainty at low frequencies for |S11| near 1. This translates to 25 ohm uncertainty for 1000 ohm real impedance. Phase errors are considerably higher. Higher accuracy may be possible by making reference measurements with high impedance calibration loads and correcting the results. I think, the accuracy may be sufficient for an overview measurement, but it's surely not mindblowing.

A dedicated RF LCR meter or impedance analyzer like HP/Agilent 4275A achieves in contrast 2% relative accuracy for impedances up to 100 kOhm at 10 MHz