Multiplexed Impedance Measurements

June 17, 2020 by Meng Li

Introduction

Having more impedance measurement channels is often desirable in many applications. For instance, in electrical impedance tomography (EIT), it is common to have many channels to achieve enough spatial resolution (144 electrodes on the thorax) [1]. Naturally, this can be done with multiple instruments working together. When synchronization is important, the multi-device-synchronization (MDS) feature can further ensure a minimal lag (< 10 ns) between measurements. However, the cost also scales in such a design. A more cost-effective solution is to multiplex the input and output channels via multiplexers (and demultiplexers). The question is, how good is the impedance measurement in such a configuration? This blog post will walk you through a simple multiplexed impedance measurement setup and help you to understand the measurement dilemma in frequency range and impedance matching:

  1. The voltage drop across the DUT can only be accurately measured via the multiplexer if the DUT impedance is lower than that of the multiplexer. In the common case of capacitive DUTs, this requires higher frequencies.
  2. The measurement accuracy is lowered by the parasitic impedance of the cable and the multiplexer. This error is less significant at a low frequency.
  3. Calibration methods such as user compensation are difficult due to the complex measurement path.

Measurement Setup

The multiplexer (MUX) in this blog post was produced by CSEM (Centre Suisse d’Electronique et de Microtechnique) and has been introduced in an earlier blog post and this application note. There, we showed that a 4-1 MUX could extend the number of single-ended voltage input on an HF2LI Lock-in Amplifier to measure 4 photodiodes at different angles. The DIO output on the back panel of the instrument was used to select the channel and switch the input impedance of the MUX. In this blog post, in order to measure the impedance in a multiplexed setup, we use the MFIA Impedance Analyzer to measure the current and differential voltage (via two MUX channels on a 8-2 MUX) simultaneously. A sketch of the setup is as follows:

multiplexed_impedance_measurement.png

Figure 1: Schematic showing the setup used for this blog post. The electrodes (and DUTs) can be connected in series or in parallel around an object of interest (this setup shows a series configuration).

To keep the setup as simple as possible, we connect two DUTs rather than many, only to highlight the working principle and to compare the result to the control measurement without multiplexing. In Figure 1, the two DUTs are in series and share the same current, which is measured by the MFIA using its current input. The voltage drop across the DUT is given by the difference between the voltages measured at its entrance and the exit. A BNC T-splitter is used to split the voltage between the 2 channels on the same row. By measuring in such a 4-terminal configuration, the common mode noise in voltage can be reduced.

Note that the input impedance of the MUX plays a key role in the voltage measurement. As we use the MUX input channels to probe voltage, the input impedance should be higher than the DUT. To avoid a high impedance from the DUT, large capacitances (DUT 1 at 1 nF, DUT2 at 100 nF) are chosen. Looking at the reactance chart (Fig. 2 below), we can be sure that the impedance of the DUTs remains below 1 MOhm for the whole frequency of test. It is also important to be aware that the gain at 1 MOhm input impedance is 2 in this setup. As such, a scaling factor of 0.5 V/V is set in the advanced tab to account for such a difference.

EIT-Reactance-chart.png

Figure 2: Reactance chart showing the impedance accuracy of the MFIA. The impedance range of the 1 nF (blue) and 100 nF (red) DUTs are represented by arrows running from 1 kHz to 500 kHz.

Results and Discussion

To begin the experiment we start with the LabOne Scope module. With the MF-DIG option enabled, the MFIA can read 2 channels, for instance, the current input and the differential voltage input simultaneously. In the blue trace of Figure 3, besides the test signal at 1 MHz, we also see unexpected noise peaks at other frequencies, likely coming from the MUX, as these peaks are not visible in the control measurement without the MUX. When the test signal moves closer to the noise frequencies during a sweep, the FFT data can be more prone to error. This highlights the need for a low-pass filter to obtain an accurate impedance in a multiplexed impedance measurement setup. The MFIA is a perfect tool for achieving this thanks to the lock-in detection principle (explained in detail in this white paper).

EIT_scope.png

Figure 3: Screenshot showing the LabOne Scope Module (with FFT) measuring the voltage (blue trace) and the current (orange trace) on DUT 2 with a 1 V test signal.

To compare the measurements of each DUT with and without the MUX, we use the LabOne Sweeper module and record the voltage drop and capacitance between 1 kHz and 5 MHz. In Figure 4, we see the derived capacitances on the two DUTs. In both cases, the error remains less than 1% from the control group until up to 500 kHz. The error in the multiplexed impedance setup originates from the 1-meter-long BNC cables used to connect the DUTs to the MUX as well as the phase error coming from the active amplification in the MUX. The parasitic inductance not only increases the measured phase from -90 deg (purely capacitive), but also alters the impedance amplitude (hence voltage drop on each DUT) at high frequencies. To verify the measurements we can also test the DUT 1 and DUT 2 in series, by opening MUX 2 In 4 and MUX 1 In 1 together. Figure 5 shows the voltage drops on the two DUTs having a ratio of 100:1, agreeing well with Kirchoff's law. And after connecting the two in series, the voltage stays almost the same as the larger capacitance DUT. But again at high frequencies, the voltage starts to roll off, which should be kept in mind.

Note: To improve the impedance measurement accuracy and to correct the phase error in any setups, normally one can perform a user compensation with a 'short' and 'load' of the same contact geometry as the DUT (see this blog post). For electrical impedance tomography applications, however, finding such components can be difficult as the sample shape can be complex. Fortunately, the measurement precision is less affected by multiplexing, meaning that the color gradient in the post-processed image likely would not change.

EIT_c.png

Figure 4: Capacitance measured in different conditions in the LabOne Sweeper. Red: 100 nF DUT without MUX. Orange: 100 nF DUT with MUX. Blue: 1 nF DUT without MUX. Cyan: 1 nF DUT with MUX.

EIT_v.png

Figure 5: Voltage drop across the DUT(s) measured in different conditions in the LabOne Sweeper. Orange: 100 nF DUT with MUX. Cyan: 1 nF DUT with MUX. Green: 1 nF DUT + 100 nF DUT with MUX.

Automating the Measurement Workflow with Python

The MUX can be manually controlled by the DIO tab in LabOne by referring to the truth table. But a fully automated solution can also be created with any of the five included APIs (MATLAB, LabVIEW, .NET, C, Python).

The Python code below shows an example of setting the measurement on DUT 1, then DUT 2, finally DUT 1 +DUT 2, by simply changing the DIO output on the MFIA. Obviously, the code can be extended to have other combinations of channel selection. The MFIA has 32 bits of DIO output in total, which can control two CSEM 8-2 MUX together. If we stick to the scheme in Figure 1, this gives a maximal number of DUT of 4 and an electrode combination of 12.

daq.setInt('/%s/dios/0/drive' % device, 3) 

#enable MFIA DIO bus 0 and bus 1 (8 bit each) lines as output 

daq.setInt('/%s/dios/0/output' % device, 4081) 

#4081 in binary is 00001111 11110001,  meaning all input 8 channels are switched to 1 MOhm input 
#impedance and open Mux 2 In 2 and Mux 1 In 1, in order to measure V on DUT 1
daq.sync() 

# synchronize the commands to the MFIA 
#do some measurements here 
daq.setInt('/%s/dios/0/output' % device, 4093) 

# this opens Mux 2 In 4 and Mux 1 In 3, to measure V on DUT 2 
daq.sync() 

#repeat your measurements here 

daq.setInt('/%s/dios/0/output' % device, 4085)

# this opens Mux 2 In 4 and Mux 1 In 1, to measure V on DUT 2 and DUT 1 in series 

daq.sync()

#repeat your measurements here   

Conclusion

This blog post describes a simple multiplexed impedance measurement setup with the MFIA Impedance Analyzer and a 8-2 MUX. Using two capacitors as DUTs, we show that the additional measurement error above typical accuracy remains < 1% below 500 kHz, but becomes larger at higher frequencies due to the parasitic impedance of the MUX and the cables. We also show how the test signal frequency and the impedance matching between the DUT and MUX are important factors to consider when building a MUX setup. In addition, it should be noted that other multiplexers than the one from CSEM can also be used, and if more channels are needed, a demultiplexer and external DIO switch can be added to the simple setup presented in this blog post.

If you have any questions or suggestions, please feel free to get in touch.

References

[1] Spatial resolution in electrical impedance tomography: A topical review, Journal of Electrical Bioimpedance, Volume 8: Issue 1.