[SOLVED] Impedance measurement with microcontroller

[asimkumar]

Hi friends,

i have assigned a job to measure complex impedance with a microcontroller and my guide restricted me to use no easy way like AD5933 SOC impedance analyzer. So i tried to design a similar thing like AD5933 with some extra facilities like multisine.
but the problem is that i am not getting a good result . The impedance vs measured impedance graph is exponential instead of straight line.
if anyone done something like this please help me.

 

[FvM]

The first point would be to determine if it's a case of bad measurement (hardware problem) or bad data processing (software problem).

I'm quite sure, that you did neither implement the exact hardware of the AD vector impedance analyzer chip nor the same software (which is actually unknown). So you would want to tell some more details.

At present, I can't imagine what kind of exponential graph you got.

 

[asimkumar]

yes, i have modified all the stages but similar to ad5933 the datasheet gives all the technique hardware and software

i have designed the DDS stage with same as it is in ad5933 except the sin and cos vectors that are later to be used for dft of the current signal.

i just produce a single sine or multisine wave. pass it through the subject. record the signal from the trans-impedance converter, the signal is recorded with ADC in DMA mode so at 256Ksps far above nyquist (signal freq 20KHz). the adc resolution is 12 bit(it may be a factor but ad5933 has same resolution).

later adc recovered data is sent to a pc through serial port , which is processed with a 1024 point DFT
i have tried dft in two ways, one is as people normally use, which gives bad frequency resolution.
so i tried the same with sin and cos vector as it is given in ad5933 datasheet.
the output of dft is a inversely propotional to the impedance. So am trying to calibrate it.
This is what i have done yet.

Thanks

 

[FvM]

the datasheet gives all the technique hardware and software

It's describing the "theory of operation" but you can't be sure, that it's exactly implemented this way. What you described as not getting a linear impedance relations sounds like a serious fault either in the hardware signal chain or in the data processing. You should be able to detect it by looking at the waveforms respectively data tapped at different points of the signal chain.

Regarding signal processing methods, I think that uisng a single frequency data processing (e.g. Goertzel algorithm) or simple synchronous demodulation of the input signal with quadrature DDS signals will give the impedance measurements with less arithmetic effort.

 

[asimkumar]

thanks for the information, i will try the Goertzel algorithm.