A Chopper Stabilization Audio Instrumentation Amplifier for IoT Applications

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

 

Review JLPEA- 766970

A Chopper Stabilization Audio Instrumentation Amplifier for IoT Applications

Jamel Nebhen, Pietro Maris Ferreira, Sofiene Mansouri

The work represents an interesting contribution in the emerging field of M&NEMS based systems, combining the state of the art in NEMS (silicon nanowires) and such widely accepted and effective technique of decreasing the dc offset and 1/f noise in linear amplifiers as chopper stabilization. Generally, the work presentation is fluently articulated. Nevertheless, some inconsistencies and minor inherent contradictions were revealed and specified below.

First of all, as being logically related to and stemmed from the works: 

1. Nebhen, J., Savary, E., Rahajandraibe, W., Dufaza, C., Meillere, S., Kussener, E., … Lhermet, H. (2014). Low-noise CMOS amplifier for readout electronic of resistive NEMS audio sensor. 2014 Symposium on Design, Test, Integration and Packaging of MEMS/MOEMS (DTIP).doi:10.1109/dtip.2014.7056676;
2. Savary, E. & Rahajandraibe, Wenceslas & Meillère, Stéphane & Kussener, Edith & Barthelemy, H. & Czarny, Jaroslaw & Lhermet, Helene & Puers, Robert. (2015). High resolution NEMS smart audio sensor based on resistive silicon nano wires for hearing aids. 2014 21st IEEE International Conference on Electronics, Circuits and Systems, ICECS 2014. 558-561. 10.1109/ICECS.2014.7050046.

the paper is better to possess these in the reference list, providing Figures 2 and 3 (See lines 107, 117) with the corresponding references as well.

Secondly, as, e.g. the title of the second reference presumes (“…for hearing aids”), it would be better to choose more precisely the area of application. Whether IoT or Cochlear implant/biomedical application. Despite stressing the IoT term in the title, the sentence in lines (131-136) deviates towards biomedical application again, while absolutely rightly noting the excess of the power dissipated for this kind of target application. Please avoid this type of uncertainty providing the appropriate text correction.  

Another point is the final current consumption. Throughout the article, the current/power consumption is declared to be excessive several times. But in the end, in the Conclusion section, the value is just accepted as it is without even an attempt for any kind of justification. And this point is quite crucial when the final application is considered. So, please, pay more attention to this moment and fix these questions correspondingly; providing a detailed analysis of the power consumption in terms of concrete application constraints/requirements, and, as a conclusion, favoring the most appropriate one.  

Regarding the table of performance comparison, it is better to include the bandwidth information as well. Ideally, it would be reasonable to provide a comparison with the IAs for the same application (audio)/close bandwidth and not for biopotential recordings. Please, note that in the references you have provided (17-21) the performance comparison tables do follow this peculiarity. And as you have noted yourselves, the gain and bandwidth are strictly related.

The selection of the chopping frequency remains quite mysterious. Note, that the chopper amplifier bandwidth doesn’t go over the 10-20% of chopping frequency. Provided that the bandwidth you play around is 20 kHz, 6 kHz may seem insufficient. Consequently, the results reported in the measurement section may be questionable. From this point of view, the authors are kindly invited to consult the following works:

1. Kwon, Yongsu & Kim, Hyungseup & Kim, Jaesung & Han, Kwonsang & You, Donggeun & Heo, Hyunwoo & Cho, Dong-il & Ko, Hyoungho. (2019). Fully Differential Chopper-Stabilized Multipath Current-Feedback Instrumentation Amplifier with R-2R DAC Offset Adjustment for Resistive Bridge Sensors. Applied Sciences. 10. 63. 10.3390/app10010063.
2. Han, Kwonsang & Kim, Hyungseup & Kim, Jaesung & You, Donggeun & Heo, Hyunwoo & Kwon, Yongsu & Lee, Junghoon & Ko, Hyoungho. (2020). A 24.88 nV/√Hz Wheatstone Bridge Readout Integrated Circuit with Chopper-Stabilized Multipath Operational Amplifier. Applied Sciences. 10. 399. 10.3390/app10010399.
3. Ahmed, Moaaz & boussaid, farid & Bermak, Amine. (2017). An Ultra Low-Power Capacitively-Coupled Chopper Instrumentation Amplifier for Wheatstone-Bridge Readout Circuits. 10.1109/ISCAS.2017.8050330.

Below the table of minor inconsistencies noted is given.

Line	Comment
12	To be more precise the accelerometer is better to call “piezoresistive”
32	Eliminate “even” before “professional designers”
59	Change the reference. 20Hz – 20kHz – is better to be referenced to a book on biomedical engineering.
70-86	Eliminate the paragraph as excessive, the reference 10 will be enough.
127	Adjective missing after “quite”. From the context seems to be “high”.
144	Correct the index in Vbisa
147-148	Avoid double “However” use, and unify the sentences.
165	Add “where” before “a Wheatstone bridge”
172	What do you mean by “several configurations”? Specify.
180, 185	According to the function performed, and to avoid the confusion with differentiator which is not used in your circuitry at all, the circuit S1 should be called “subtractor”.
233	There is only 1 integrator present in the circuit (at least as it can be seen from the schematics provided). So eliminate “first”. Otherwise, specify what is the “second” integrator (probably the one included into the low-pass-filter?).
255-268	Unify the sentences
260	“CMFB” should be spelled out in the text.
267-289	Strictly correlate the passages of the narrative with Figure 10.
310-311	Avoid “we perform now”, add “was performed and described below” instead.
325	Magnify the image
347	Control the words’ order “the well”
379-381	Eliminate the phrase “To design…”
385-386	Reformulate the phrase, taking into consideration the guidelines provided in the beginning.
388	“It was conducted” eliminate one.

 

 

 

Author Response

1-Figures 2 and 3 in lines 107 and 117 are with the appropriate references [11] and [12] as

[11] Nebhen, J., Savary, E., Rahajandraibe, W., Dufaza, C., Meillere, S., Kussener, E., … Lhermet, H. (2014). Low-noise CMOS amplifier for readout electronic of resistive NEMS audio sensor. 2014 Symposium on Design, Test, Integration and Packaging of MEMS/MOEMS (DTIP).doi:10.1109/dtip.2014.7056676;

[13] Savary, E. & Rahajandraibe, Wenceslas & Meillère, Stéphane & Kussener, Edith & Barthelemy, H. & Czarny, Jaroslaw & Lhermet, Helene & Puers, Robert. (2015). High resolution NEMS smart audio sensor based on resistive silicon nano wires for hearing aids. 2014 21st IEEE International Conference on Electronics, Circuits and Systems, ICECS 2014. 558-561. 10.1109/ICECS.2014.7050046.

2-We have provided the precisely application which is the IoT application. Therefore, the appropriate text correction in line (131-136) is “the target IoT application”.

3-Regarding the table of performance comparison, we have included the bandwidth information.

Reference [17] in the comparison table is replaced by the reference “Kwon, Y.; Kim, H.; Kim, J.; Han, K.; You, D.; Heo, H.; Cho, D.; Ko, H. Fully Differential Chopper-Stabilized Multipath Current-Feedback Instrumentation Amplifier with R-2R DAC Offset Adjustment for Resistive Bridge Sensors. Applied Sciences 2019, vol. 10, no. 63, pp. 1-9.”

Reference [18] in the comparison table is replaced by the reference “Butti, F.; Piotto, M.; Bruschi, P. A Chopper Instrumentation Amplifier with Input Resistance Boosting by Means of Synchronous Dynamic Element Matching. IEEE Trans. Circuits Syst. I Regul. Pap. 2017, vol. 64, pp. 753–794.”

4-We have corrected the chopping frequency, fch=12kHz. The bandwidth is 20kHz. The low-pass filter has a cutoff frequency of 30kHz, therefore, it eliminates the third and the fifth harmonics.

5-Line 12, piezoresistive sensor

Line 32, “even” is removed

Line 59, reference [9] is replaced by a book on biomedical engineering “Kitchin, c.; Counts, L. A designer’s guide to instrumentation amplifiers. Analog Devices, 2006.”

Line 70-86, we have eliminate the excessive paragraph “The sensor is composed by the MEMS component, which is an inertial mass suspended on a hinge, and by the NEMS component, which is a suspended silicon nanogauge. The silicon nanogauge has a specified situation. Moreover, it is designed along the line between the hinge’s rotation axis and the seismic mass center inertia. Therefore, it is suspended from the anchor to the seismic mass in silicon substrate. If an acceleration is detected, then the seismic mass acts immediately by a rotational movement. This rotational movement is made by respecting the rotational axis. Then, the resulting inertial force is applied to the nanogauge section generating longitudinal or tensional stress. Due to the piezo-resistive effect, the nanogauge stress allows generating a resistance variation proportional to the inertial force. The resistance variation is amplified by a lever effect factor. The lever effect represents the resulting distance which separates the rotation axis and the seismic mass center divided by the resulting distance, which separates the same rotation axis and the nanogauge position.”

Line 127, we have added “ quiet high”

Line 144, we have corrected “ Vbias”

Lines 147-1478, we have avoided the double “However”, and we have unified the sentences. The new sentence becomes “However, the current flow through the sensor implies a power consumption, which is a crucial parameter in battery-powered systems.”

Line 165, we have added “where” before “a Wheatstone bridge”.

Line 172, “several configurations” means the current configuration or the voltage configuration. Therefore, it means the current amplifier or the voltage amplifier.

Lines 180-185, the circuit S1 is called “Subtractor”.

Line 233, we replaced “first integrator” by “instrumentation amplifier”. There is no other integrator in the circuit.

Lines 255-268, sentences are unified “where τ = R × C_in with R denotes the input resistance and C_in denotes the input capacitance of the amplifier. The major weakness of this technique is the τ itself, which not only depends on the sensor’s source resistance R, but also on the amplifier’s input capacitance C_in.”

Line 260, we have added “The instrumentation amplifier with its common-mode feedback circuit (CMFB)”.

Lines 267-289, we have removed “Modulator1 is the first…around zero”. Figure 10 is well described by lines 168-278.

Lines 310-311, we replace “we perform now” by  “was performed and described below”.

Line 325, the image is magnified.

Line 347, word “well” was replaced by the word “proper”.

Lines 379-381, the phrase “To design…” was removed.

Lines 385-386, the phrase is reformulated as “From measurement results over a signal bandwidth of 20-kHz, the low-noise instrumentation amplifier achieves an SNR of 77-dB and it has a great potential of being used in IoT applications.”

Line 388, we have removed “It was conducted” and updated the Acknowledgment by “This work was conducted in cooperation with Prince Sattam bin Abdulaziz University, Alkharj, Saudi Arabia, and Sorbonne Université, Paris-Saclay,CentraleSupélec,CNRS,Lab. de Génie Electrique et Electronique de Paris, France.”

Reviewer 2 Report

This paper presents a instrumentation amplifier for resistive microphone.
I have some comments as below.

1) In Fig. 6, how is the VCM detected? Is the VCM detected from the center node of RG in Fig. 7?
2) In Fig. 6, please indicate where the choppers are applied.
3) In Fig. 7, please add detailed implementation including transistor level.
4) In Fig. 16, the y-axis unit may be V/rtHz.
5) The delayed chopper is widely used. Generally, the demodulation chopper is delayed for the phase lag at the chopper frequency.
Why the LPF is adopted before the demodulation?

The LPF can reduce the high frequency harmonics before demodulation,
however, the effectiveness over additional size and power consumption should be proven.

The chopper frequency is 6 kHz and cut-off frequency of LPF is 40 kHz.
Then, the odd harmonics, third (6*3=18kHz) and 30 (6*5 kHz), are still remained,
and demodulated into the baseband.

Please add the simulation and analysis results for delayed chopper with low pass filter.

 

Author Response

1- Vcm=(VA+VB)/2=Vbias/2[(R3/(R1+R3))+(R4/(R2+R4))]

Vcm is detected from the center node of the Wheatstone bridge (VA+VB)/2.

2- we have added the Chopper stabilization circuit in Figure 6.

3- We have added the detail implementation including transistor level of amplifiers A1, A2 and A3 in Figure 8. We have also added “The architecture of each amplifier is based on a two-stage operational transconductance amplifier (OTA) as shown in Figure 8.”

4- We have changed in Figure 16 the Y-axis is in V/√Hz.

5- Using an amplifier with gain A, a delay ∆t between the modulating clock signals m1(t) and m2(t) can be introduced. The introduction of the delay ∆t causes a chopping of the spike signal itself and hence the dc content of Vout(t) is minimized. An optimum delay ∆topt = ln 2 × τ exists, where complete cancellation of the residual output offset is effective. The major weakness of this arrangement is the τ itself, that not only depends on the sensor’s source resistance R, but also on the amplifier’s input capacitance Cin. Neither of those parameters is well controlled and depends on the sensor application and on process tolerances. Therefore, the basic delayed demodulation scheme is of little use in practical circuit designs. To solve this shortcoming, shaping of the spike can be introduced by the addition of a first order low-pass filter with time constant τc after the amplifier. Provided that T >> τc >> τ, the shape of the time response Xfspike(t) of the filtered spike is primarily determined by τc and independent of the impedance of the connected sensor.

Round 2

Reviewer 1 Report

The work presentation has been essentially improved. Some minor corrections to be done are related to the figures. Eliminate duplicates and provide each with an appropriate legend.

Reviewer 2 Report

The manuscript is well revised, and I recommend this paper to be accepted in the present form.