Temperature Difference Sensor to Monitor the Temperature Difference in Processor Active Heat Sink Based on Thermopile

Round 1

Reviewer 1 Report

This is the third version of the manuscript and a lot of ground has been covered.

I have two main remarks in this stage:

1. I think that the authors made an error in the graphs of fig. 18 and in the corresponding discussion. In order to be consistent with the previous versions, fig. 18a should correspond to sensor type N and fig. 18c should correspond to type A. Therefore, the range ΔU is wrongly depicted in lines 610-611 and the conclusion that is mentioned in lines 696-699 is also wrong.
2. The purpose of the experiment 3, as stated by the authors, is “to investigate sensors response time with the accuracy better than in experiment 2”. In my opinion the data presented in this section fail to serve the specific purpose. In fact, nothing new is presented compared to experiment 2. In order to study the transient sensor response, the following changes should be made: The pulse duration of each cycle should be increased in order for the sensor to reach the maximum and minimum levels. Thus, the sensor during the rise period should have enough time to reach a plateau and during the fall period should also reach the base line. Then the authors should study in more detail the step response on each sensor type and they should extract the corresponding rise and fall time, as commonly denoted in the literature [the time taken by the signal to change from a specified low threshold (usually 10%) to a specified high threshold (usually 90%)]. In the graphs of the new version of fig. 19, the Y-axis should indicate the corresponding temperature.

Author Response

Dear Reviewer,

thank you for your remarks. Together with the previous ones it improved the paper significantly, we believe. We introduced the first remark in the paper. Second one needs to prepare some additional experiments using different measurement set-up. There are two objects - the sensor and the heat sink. Based on the suggestions given, the authors plan to deal in the future with a more detailed analysis of the response dynamics of a thermoelectric sensor placed on an active heat sink. The measurement is non-trivial both from the point of view of experimental measurements and theoretical analysis (approximation of results with appropriate mathematical relations) and it is not possible to complete it within the time the journal requires to submit a revised version of the article. The authors hope for wider suggestions from the Reviewer on this issue with the possibility of preparing together a paper on this topic in the future. We hope you accept our explanation below.

 

ad 1. That is right, thank you for pointing this mistake. The subtitles in Fig. 18 as well as appropriate conclusions were improved – marked yellow in the text.

ad 2. We agree that the exact response time of the sensors cannot be determined from the results of Experiment 3. This experiment is only intended as a preliminary evaluation of them. An accurate determination would require, in addition to the comments given by the Reviewer, a change in the entire measurement set-up. Currently, the sensors and the active heat sink form a single entity thermally. The rise/fall times of the sensor output signal are strongly influenced by the heat capacity of the heat sink. Moreover, it is complicated by the fact that an active heat sink is used.

The main goal of the presented work was not to compare different versions of the sensors, but to develop such an device. Three variants were fabricated to additional test certain aspects. However, this is not the substance of the matter and is only an extension of the paper. We would not like to extend this paper with further chapters. The basic version of the article is a presentation of the concept, design and realization of one version of the sensor. Please treat the additionally presented experiments in this way.

Reviewer 2 Report

The authors fully took into account my comment in the previous review and the article can be published in the journal "Electronics" in the presented form.

Author Response

Dear Reviewer,

thank you for your opinion.

The Authors

Reviewer 3 Report

The manuscript is properly improved.

But the experimental results should improve by technical revision of the manuscript.

1. The fan was switched on for 10 seconds each time in experiment 3 (see page 19 in line 622). How can control times under decimal point in this experiment. The authors should explain about this.
2. What does it means “(Ag/Ag)” in 436 line and “(3.4)” in line 449 (see table 1)?
3. NiCr and CuNi wires were used as thermoelectric materials of version E. The Temperature coefficient of resistance (TCR)s of NiCr are different according to atomic percentage (%) of Ni:Cr in the NiCr wires (See ref.; Youn-Jang Kim Thin Solid Films, Volume 675, 1 April 2019, Pages 96-102) also this is very important in 'Scientific Soundness’. I recommend the authors explain more detail this.

Author Response

Dear Reviewer,

thank you for your remarks. We introduced it all in the paper. We believe it will increase the value of the paper.

 

ad 1. The Keysight 34970A data logger took the measurements every 0.209 s. One of the monitored parameters was the fan supply voltage level. The exact times of switch on/off the fan are unknown. We only know the time “n” when the fan is still switched off, and the time “n+1” = n + 0.209 s when it is already switched on. The values in Table 3 were “n+1” times. Therefore, the accuracy is 0.209 s theoretically. However, the rotary speed of the fan is rising over some period of time, until it reaches maximum. That’s why we agree with this remark. The times in Table 3 were changed to more rough – rounded to unity.

ad 2. Parameters in brackets are for sensor version A2 (the differences between A and A2 are pointed). To improve the readability, the star (*) with explanation was added below Table 1.

ad 3. The suggestion as well as the literature reference were introduced to the text – chapters 2.2.3 and 3.1. However, due to the small temperature difference during the tests (in the range from 23°C to 62°C) the impact of TCR is only small.

Round 2

Reviewer 1 Report

In this fourth version of the manuscript the error in the results presentation (fig 18) was corrected, but this affected the explanations of the experimental data. In the previous (wrong) version of the manuscript was stated that: “Sensor version N seems to have the highest sensitivity to thermal condition changes.” (line 611) and “The range ΔU over which the output signal changes is much wider for sensor N, as experiment 2 shows. ΔU is about 0.25 mV for sensor A, 0.45 mV for sensor E and 0.75 mV for sensor N. This is the key parameter for described application.” (lines 696-699).

In this version the interpretation was changed and it is mentioned that: “Consequently, the sensitivity of sensor N should be better than for versions A and E. However, the results obtained in experiment 2 do not support the above. The range ΔU over which the output signal changes is much wider for sensors A and E: about 0.25 mV for sensor N, 0.45 mV for sensor E and 0.75 mV for sensor A” (lines 671-674).

These results are related to the core subject of the presented research work and cannot treated superficially. I believe that the authors should repeat their experimental work and study more thoroughly the basic parameters of their devices (evaluation of Seebeck coefficient, detailed thermal response with constantly temperature monitoring etc). Moreover, the experimental results should be related to the theoretical results, that also presented in the manuscript and in the present version are not compared to the corresponding experimental.

I cannot bypass the fact that the current experimental results are not aligned with the theoretically expected device behavior.

Furthermore, I cannot understand why “An accurate determination would require, in addition to the comments given by the Reviewer, a change in the entire measurement set-up”. In order to study the transient behavior according to my suggestion, the only change that needs to be implemented is to extend the time frames reported in table 3. Otherwise, there is no scientific reason for the experiment 3 to be presented since it reports the same results as experiment 2.

Author Response

An explanation why the sensor behaves differently than predicted theoretically is proposed in the paper. It is related to the convective cooling of the thermocouple wires, which are close to the thermoelectric junctions. Since this is an unexpected phenomenon, it requires further investigation. However, this does not mean that the message about the design and operation of the sensor to the scientific community is premature.  The main goal of the paper is to present the concept, design, fabrication and initial, basic testing of the sensor. The comments provided in the review relate to detailed studies that may be presented as results of further work, in subsequent papers. As authors, we assume that a scientific article does not have to completely exhaust the subject it touches. Especially if the subject is broad. Please, consider the research presented in the Experiment 2 and Experiment 3 sections as additional information about the sensor. Requiring further work, but already presented to the scientific community - for evaluation and exchange of comments on the further direction of this research.  For such comments provided in earlier reviews, we thank you. We will take them into account in further investigations.

In order to make the measurements indicated in the review, the active heat sink on which the sensor is mounted must be eliminated. The heat sink has a much higher heat capacity than the sensor. Consequently, it is much slower to heat up/cool down. The suggested extension of the experiment will determine the time constant of the heat sink, not the sensor. The temperature of the sensor in the investigated set up depends on the temperature of the active heat sink. The sensor cannot heat up/cool down faster than the heat sink. The suggested measurements should be made on the sensor itself. Then they will be reliable, they will relate to the sensor. Therefore, a different heating (pulse heating) and cooling system should be provided. Additionally, the cooling should not affect the temperature of the thermoelectric leads, but only the "cold" LTCC substrate. This implies a complete modification of the measurement system and the design of an additional experiment. This is feasible and we will report such results in future papers. The scope of work and amount of text needed to describe the results are too large to add to this paper. Please, consider experiment 3 as an extension of experiment 2 - the measurement frequency of the sensor output signal was significantly increased. This allows to see how fast it responds (already in the next measurement cycle or only after several). This is a bridgehead for later determination of the time constant.

Reviewer 3 Report

The manuscript is properly revised. But the manuscript should improve by technical revision of the manuscript.

Point 1: When preparing your manuscript, please use SI units in the paper. And the temperature “55.75C” (in 535 line page 17) and “4.25C” (in 535 line page 17) can be updated to 55.8C” (in 535 line page 17) and “4.3C”, respectively.

Please change; “5000” to “5,000” Table 1 (page 14), “3000 seconds” to “3,000s” in line 474 (page 15) and “5900”, “3000” to “5,900”, “3,000” in line 505 (page 16).

Author Response

Dear Reviewer,

than you for your comments. All listed corrections have been made.

Round 3

Reviewer 1 Report

I welcome the authors’ reply to my remarks, but I have to underline that no further progress was achieved from the previous version.

Although unexpected behaviors were observed there is no sufficient explanation reported. I have also to mention that the specific phenomenon was not reported in the previous versions, where totally different interpretation was proposed regarding the comparison of the different device versions. After the correction that Ι was suggested in my previous review, the authors changed the conclusions accordingly and they admit that the current experimental results are not aligned with the theoretically expected device behavior. When the reported results are altered and the corresponding explanations are modified during the review process, then this is an indication that the authors should look deeper in their subject.

The authors mention that their message to the scientific community is not premature, but this is not justified by the several changes in the last four manuscript versions, which in many cases are contradictory.

Furthermore, a new issue was arising with the resent authors’ response, where they mention: “The heat sink has a much higher heat capacity than the sensor. Consequently, it is much slower to heat up/cool down. The suggested extension of the experiment will determine the time constant of the heat sink, not the sensor. The temperature of the sensor in the investigated set up depends on the temperature of the active heat sink. The sensor cannot heat up/cool down faster than the heat sink. …. Additionally, the cooling should not affect the temperature of the thermoelectric leads, but only the "cold" LTCC substrate”.

With the above-mention statement the authors agree that the temperature evaluation of various sensors’ response was actually dominated by the response of the heat sink, thus there is no reason to compare the time response of the different sensor designs. However, they still mention in the conclusions “The third effect is an improvement of the sensor time constant (it was replaced by the term “sensor reaction time” after my 1^st review). The sensor N exhibits a slightly faster change of the output signal than versions A or E. The difference may result from the volume (heat capacity) of the thermoelectric junction – in versions A and E it is a solder joint with a large volume. In version N the junction is much smaller (screen-printed legs, 15 μm in thick) and therefore its heat capacity is smaller. This may result in faster sensor response.”

This is one more example of the several contradictory conclusions throughout the five stages of the review process.

I think that the authors will agree that the papers that are reported to the scientific community have to advance the science a little further and all the contents have to report something new. In the case of experiment 3 there is nothing new that cannot be extracted from experiment 2. The increase of the frequency does not contribute anything to the reported results.

 

Thus, I remain to my previous suggestion: “I believe that the authors should repeat their experimental work and study more thoroughly the basic parameters of their devices (evaluation of Seebeck coefficient, detailed thermal response with constantly temperature monitoring etc). Moreover, the experimental results should be related to the theoretical results, that also presented in the manuscript and in the present version are not compared to the corresponding experimental.”.

Author Response

Dear Reviewer,
thank you for your opinion. Your suggestions were very valuable to us. Most of them (nearly all) were introduced to the paper. I’m sure its improved value of the manuscript. Some of your suggestions I will keep in mind and use in my further investigations related to the topic. I believe there is no space for more detailed studies in this paper. Below you can find the answers on your last comments. 

“When the reported results are altered and the corresponding explanations are modified during the review process, then this is an indication that the authors should look deeper in their subject.”
Our goal was to report the concept of a new sensor. Not a comprehensive analysis of all aspects of its performance. We leave that for further work. We do not want to publish a book, but an article. We agree some issues remain unexplained, we may be wrong in some conclusions. Therefore we write in the article that the research will be continued. Some aspects remains not investigated yet, its need more extensive work in the future. And the most important: no one result was altered, but the additional were added. Only the interpretation was slightly improved after Reviewers suggestions.

“With the above-mention statement the authors agree that the temperature evaluation of various sensors’ response was actually dominated by the response of the heat sink”.
YES. But we know it because we made an measurements reported in the paper (experiments 2 and 3). If we didn’t check it, we couldn’t write it. That is the goal of our experiment and we call it “basic tests of the sensor”. We would like to present concept, design, fabrication and basic tests of the sensor. Moreover, the sensors’ response is dominated by the response of the heat sink, but not totally. The results of the experiment 2 and 3 show there are some slight differences in time response. That’s the results. Why we can’t report it?

“I think that the authors will agree that the papers that are reported to the scientific community have to advance the science a little further and all the contents have to report something new. In the case of experiment 3 there is nothing new that cannot be extracted from experiment 2. The increase of the frequency does not contribute anything to the reported results”
If it is the reason of major revision demand we can remove this experiment. It is just extension of experiment 2, as we wrote in the paper. In my opinion it introduces more accurate results, therefore is could be interesting. In my opinion the result can be reported even if it has only a slight impact. It can i.e. confirm or deny the results of experiment 2. In my opinion the whole paper should advance the science a little further, not the single paragraph/chapter/experiment.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

This paper reports on the fabrication and evaluation of a thermoelectric sensor for monitoring the temperature difference at opposite ends of an active heat sink.

 

The work is presented in detail, but there are several issues that need to be addressed and some aspects that needs to be clarified.

 

My remarks in detail:

 

The title “Temperature Difference Sensor to Control the Temperature Difference in Processor Active Heat Sink Based on Thermopile” indicates that control of the temperature difference takes place. However, this is not the case in the present work thus, the term “control” should be replaced by “monitor”.

 

Fig. 2: The thermocouple indicated in the figure is not representative for the specific case. The cold contact should be “open circuit” so as the thermoelectric voltage difference to be obtained, which is the sensor signal for the proposed application.

 

Fig. 4: The second (a) should be replaced by (b)

 

Section 2.2 (Sensor design). In versions A and E there are no thermoelectric elements printed on the substrates since the printed geometries serve as simple electrical contacts, thus the sensor relies on the thermoelectric wires which are detecting the temperature difference T1-T2. The question is why the authors use PdAg-based ink for ver. A and Ag for ver. E. They should have used the same material so as the results to be directly comparable. Even if (theoretically) the intermediate material does not affect the measurement results, the temperature distribution is not the same with various materials, due to variations in thermal conductivity, thus the condition that “both ends should have the same temperature” is not always fulfilled. The authors should comment on this.

Moreover, in the version N there are printed thermoelectric elements on the substrate and also thermoelectric wires. What is the need for using Ag/Ni wires since simple metal wires can be used? In my opinion an extra version should be evaluated, where metallic wires with low thermal conductivity should be used to connect the two substrates instead of the thermoelectric wires. The thermoelectric wires due to their dimension and their high thermal conductivity maintain an efficient thermal coupling that reduces the temperature difference T1-T2.

 

Fig. 12d does not indicate the “Thermoelectric temperature difference sensor” as denoted in the manuscript.

 

From fig. 13a it’s obvious that not all the hot parts of the three sensor types are located in the same thick film resistor, thus “same thermal conditions” are not assured, as it is claimed in the manuscript. The authors should comment on this.

 

Fig. 14: The response is for one thermocouple and not for the thermopile. This should be indicated in the caption.

 

The voltage differences can be extracted from fig. 16, in each case as follows: A: 0.7mv, E: 0.5mV, N: 0.2mV, which according to the related Seebeck coefficients correspond to temperature differences of: case A: 5.5°C, case E: 1.2°C and case N: 1.6°C. These results are not consistent with the 1^st experiment. The authors should provide explanations for this behavior.

 

The temperature variations presented in figs 17b,18b,19b for both CPU_heat_sink and FAN_heat_sink are different in each case. It is not clear why is this happening since the stimulus is the same, thus the induced variations should be also the same.

 

The data regarding the fan switching in 2^nd paragraph of page 19 should be presented in a table.

 

In the 3^rd paragraph of page 19, the level of the voltage response of each sensor is indicated and the authors come to the conclusion that type E sensor present the highest signal, thus it works more efficient. This is not the correct results interpretation, since the authors present only the maximum value of the signal (type A: 1mV, type E: 2.1mV, type N: 1.3mV) which corresponds to the highest temperature obtained. The parameter that is important in this application is the temperature span which is expressed by the range of the sensor signal (ΔV=max-min value). From figs 17-19 the signal range for each sensor type can be extracted: type A: 0,45mV, type E: 0.6mV, type N: 0.5mV. According to these results the authors conclusion that “It should be noted that the responses of sensors A and N differ from each other despite the fact that they were fabricated using the same thermoelectric wires” is not valid.

Furthermore, the corresponding temperature differences are: 3.6°C (for type A), 1.5°C (for type E), 4.0°C (for type N). The authors should comment on these results.

 

In the last paragraph of section 3.3 the authors present the differences in the delay time of each sensor type. The authors should explain how the different types of the sensors presented in this work corelate with different delay time in the response, since all the sensors have the same geometry and the same substrate materials. In my opinion the different time response is most probably due to the slight differences in the fabrication of the LTCC substrates, since they exhibit slightly different total thermal capacities and conductivities. Moreover, the extracted differences are in the order of 0.2sec, which is the same as the minimum measurement interval.

 

In section 4 two contradictory conclusions are mentioned:

“A comparison of sensor N and A shows that distancing the junctions from the wires increases the output electrical signal from 0.4 mV to 0.7 mV. But comparison of sensors N and E shows that the efficiency of thermoelectric material used is even more important.”

“This means that distancing the thermoelectric junctions from the wires is more important than the optimal choice of thermoelectric materials.”

In my opinion the most important parameter is the distancing of the thermoelectric junction since it assures higher temperature difference, which leads to higher resolution. On the other side, the absolute value of the thermopile signal can be amplified by many methods. The authors should clarify and support their opinion based on their results.

Author Response

Dear Reviewer,

thank you for your detailed and substantive remarks. We believe they will improve our paper. Most of them were applied in the text. Others are discussed below and our explanations are given. Moderate English changes were applied in the text.

 

1) The title “Temperature Difference Sensor to Control the Temperature Difference in Processor Active Heat Sink Based on Thermopile” indicates that control of the temperature difference takes place. However, this is not the case in the present work thus, the term “control” should be replaced by “monitor”.

The appropriate changes were made in the paper.

 

2) Fig. 2: The thermocouple indicated in the figure is not representative for the specific case. The cold contact should be “open circuit” so as the thermoelectric voltage difference to be obtained, which is the sensor signal for the proposed application.

The appropriate changes were made in the paper.

 

3) Fig. 4: The second (a) should be replaced by (b)     

The appropriate changes were made in the paper.

4) Section 2.2 (Sensor design). In versions A and E there are no thermoelectric elements printed on the substrates since the printed geometries serve as simple electrical contacts, thus the sensor relies on the thermoelectric wires which are detecting the temperature difference T1-T2. The question is why the authors use PdAg-based ink for ver. A and Ag for ver. E. They should have used the same material so as the results to be directly comparable. Even if (theoretically) the intermediate material does not affect the measurement results, the temperature distribution is not the same with various materials, due to variations in thermal conductivity, thus the condition that “both ends should have the same temperature” is not always fulfilled. The authors should comment on this.

That is right, the structures should be the same and made of the same material so that they can be accurately compared. The oversight resulted from the fact that more than a dozen geometric versions of the sensors were fabricated, only a selected few were tested. The focus was on thermoelectric parameters (according to Law of Intermediate Materials different pastes have no influence on results) and the heat transport aspect of the structure was omitted. This affects the measurements, as the Reviewer noted. However, in the authors' opinion, it is negligibly small and does not change the conclusions of the study. This is due to the following factors:

1. a) the composition of the PdAg paste is 75% silver, so it is materially quite similar to Ag paste;
2. b) the volume of the LTCC substrate is much larger (6.8 ´2 ´ 0.68 mm³ for CPU_heat_sink) than the volume of the PdAg legs (6 legs 5 ´ 0.3 ´ 0.015 mm³). Thermal conductivity of the whole structure is comparable despite the use of different materials.

 

5) Moreover, in the version N there are printed thermoelectric elements on the substrate and also thermoelectric wires. What is the need for using Ag/Ni wires since simple metal wires can be used? In my opinion an extra version should be evaluated, where metallic wires with low thermal conductivity should be used to connect the two substrates instead of the thermoelectric wires. The thermoelectric wires due to their dimension and their high thermal conductivity maintain an efficient thermal coupling that reduces the temperature difference T1-T2.

The sensor version with simple metal wires (instead of thermoelectric Ag/Ni) was considered, but after detailed theoretical analysis (and experimental verification) was refused because it conflicts with thermoelectric Law of Intermediate Materials. The sensor will not measure T1-T2 (nor T3-T4) temperature difference (see Fig. 8) in that version.

That is right, the wires (thermoelectric or simple metal) reduce the temperature difference T1-T2. Our proposal is to distance the thermoelectric junction from the wires (ver. N of the sensor, where T3-T4 is measured).

 

6) Fig. 12d does not indicate the “Thermoelectric temperature difference sensor” as denoted in the manuscript.

I’m sorry but I don’t understand this remark. Please note that the thermoelectric temperature difference sensor (together with thermoelectric wires) is presented in Fig. 12d.

 

7) From fig. 13a it’s obvious that not all the hot parts of the three sensor types are located in the same thick film resistor, thus “same thermal conditions” are not assured, as it is claimed in the manuscript. The authors should comment on this.

All four thick film resistors presented in Fig. 13a have the same geometrical dimensions, the same material was used in screen-printing process. Moreover, they were connected parallel and powered using the same electrical current level. The thermal conditions should be identical for all resistors. However, we accept Reviewer doubt (no measurements of the conditions were performed) and the appropriate text was modified (“the same thermal condition” phrase was removed).

 

8) Fig. 14: The response is for one thermocouple and not for the thermopile. This should be indicated in the caption.

The response for the thermopile (6 thermocouples) is presented in Fig. 14; i.e. ver. N: 6 ´ 21 uV/K ´ 5.5K = 0.69 mV

 

9) The voltage differences can be extracted from fig. 16, in each case as follows: A: 0.7mv, E: 0.5mV, N: 0.2mV, which according to the related Seebeck coefficients correspond to temperature differences of: case A: 5.5°C, case E: 1.2°C and case N: 1.6°C. These results are not consistent with the 1^st experiment. The authors should provide explanations for this behavior.

The temperature variations presented in figs 17b,18b,19b for both CPU_heat_sink and FAN_heat_sink are different in each case. It is not clear why is this happening since the stimulus is the same, thus the induced variations should be also the same.

Thank you for paying our attention to this part. An important information was missed in the experiments description. In Experiment 1 all sensors were characterized simultaneously. Its purpose was to directly compare the sensors under identical conditions. In Experiments 2 and 3 the sensors were measured one at a time (not simultaneously). Identical experimental conditions were not provided because that was not the purpose of the experiments. Consequently:

* The thermal conditions (CPU_heat_sink as well as FAN_heat_sink temperatures) during the tests of sensors A, E, N were not identical. This did not matter because the response time, but not the output signal level was tested in Experiments 2 and 3.

* The system was cooled in free convection conditions in Experiment 1 and in forced convection conditions in Experiments 2 and 3 (fan).

* Figures 17b-19b are intended to show the trend of temperature variations. Measurements were made with pyrometers of limited accuracy. A, E, and N thermopiles were the main measurement tools. 

The appropriate sentences were added to the paper (first paragraph of 3.1, second paragraph of 3.2 and first paragraph of 3.3).

 

10) The data regarding the fan switching in 2^nd paragraph of page 19 should be presented in a table.

The suggestion was applied in the paper.

 

11) In the 3^rd paragraph of page 19, the level of the voltage response of each sensor is indicated and the authors come to the conclusion that type E sensor present the highest signal, thus it works more efficient. This is not the correct results interpretation, since the authors present only the maximum value of the signal (type A: 1mV, type E: 2.1mV, type N: 1.3mV) which corresponds to the highest temperature obtained. The parameter that is important in this application is the temperature span which is expressed by the range of the sensor signal (ΔV=max-min value). From figs 17-19 the signal range for each sensor type can be extracted: type A: 0,45mV, type E: 0.6mV, type N: 0.5mV. According to these results the authors conclusion that “It should be noted that the responses of sensors A and N differ from each other despite the fact that they were fabricated using the same thermoelectric wires” is not valid.

Furthermore, the corresponding temperature differences are: 3.6°C (for type A), 1.5°C (for type E), 4.0°C (for type N). The authors should comment on these results.

Thank you for the remark and suggestion. The results were analyzed once more and the appropriate paragraph was modified. Table 4 was added to the text. Explanation of achieved results:

The averaged ranges of the sensors output signals, extracted from measurement data:

type A: 0.41 mV,                   type E: 0.69 mV,                    type N: 0.51 mV.

The corresponding temperature differences (TEMP_1):

type A: 3.25°C,                      type E: 1.69°C,                      type N: 4.05°C.

The averaged temperature differences extracted from the measurement data of Figures 17b, 18b, 19b (TEMP_2):

type A: 6.8°C,                        type E: 3.8°C,                        type N: 8.8°C.

They are different because the sensors were investigated in separated measurement processes in experiment 3.

The ratio between TEMP_1 and TEMP_2:

type A: 3.25/6.8=0.48,           type E: 1.69/3.8=0.44,           type N: 4.05/8.8=0.46.

In our opinion it proves TEMP_1 should be on a similar level when the measurement conditions are the same.

 

12) In the last paragraph of section 3.3 the authors present the differences in the delay time of each sensor type. The authors should explain how the different types of the sensors presented in this work corelate with different delay time in the response, since all the sensors have the same geometry and the same substrate materials. In my opinion the different time response is most probably due to the slight differences in the fabrication of the LTCC substrates, since they exhibit slightly different total thermal capacities and conductivities. Moreover, the extracted differences are in the order of 0.2sec, which is the same as the minimum measurement interval.

The difference may result from the volume (heat capacity) of the thermoelectric junction – in versions A and E it is a solder joint with a large volume (see Figure 5). In version N the junction is much smaller (screen-printed legs, 15 µm in thick) and therefore its heat capacity is smaller. This may result in faster sensor response. The difference is noticeable but very small, the authors think it is premature to draw sweeping conclusions. The results were at the limit of the measuring instruments performance. The high speed measurement system should be set up to better investigate this in future work.

This paragraph was added in Conclusions.

 

13) In section 4 two contradictory conclusions are mentioned:

“A comparison of sensor N and A shows that distancing the junctions from the wires increases the output electrical signal from 0.4 mV to 0.7 mV. But comparison of sensors N and E shows that the efficiency of thermoelectric material used is even more important.”

“This means that distancing the thermoelectric junctions from the wires is more important than the optimal choice of thermoelectric materials.”

In my opinion the most important parameter is the distancing of the thermoelectric junction since it assures higher temperature difference, which leads to higher resolution. On the other side, the absolute value of the thermopile signal can be amplified by many methods. The authors should clarify and support their opinion based on their results.

Thank you for the remark. Indicated text was imprecise and is improved now.

 

 

 

 

Reviewer 2 Report

In my opinion, the studies conducted are presented in too much detail. I believe that the volume of the chapter "Introduction" is excessive, and the article as a whole will be better from reducing the volume of the text.

Author Response

Dear Reviewer,
Many thanks for your opinion. We think that the topic we investigated is not trivial. The Chapter ”Introduction” contains about 20% of the text. As we compared this with other papers published in "Electronics" many papers have an "Introduction" of a similar length.  

Reviewer 3 Report

In general, this paper were mainly focused on the summary of technical results, therefore needs to update the more scientific discussion about the results. For example, “the thermoelectric junctions from the wires is more important than the choice of thermoelectric materials” in line 606 of page 20.

Author Response

Dear Reviewer,
Thank you for your suggestion. However, the remarks are very general and it is difficult to discuss with them.
The indicated paragraph was improved, more explanation was given. The chapter Discussion and Conclusions was unfolded. The paper was checked once more and some improvements were made.
Moderate English changes were applied in the text.

Round 2

Reviewer 1 Report

The new version is major improvement over the previous one, however I still cannot recommend the publication of the manuscript for the following reasons:

1. The authors haven’t addressed some major remarks. More specifically they didn’t present measurement with devices with the same interconnecting material (comment no 4). Moreover, they haven’t addressed comment no 5. They claim that “The sensor version with simple metal wires (instead of thermoelectric Ag/Ni) was considered, but after detailed theoretical analysis (and experimental verification) was refused because it conflicts with thermoelectric Law of Intermediate Materials”, however they don’t present corresponding experimental evidence.
2. The authors mention (replying to comment no 8) that fig. 14 presents the response of the thermopile, which consists of 6 thermocouples. However, they use the term “Seebeck coefficient of the sensor [μV/K]” in all the evaluations reported in the manuscript, which is very confusing. The Seebeck coefficient is defined for materials, as denoted in eq. (1) of the paper and for a combination of two materials, that consist a thermopile. In that case the electromotive force (Et) is defined as: Et = a ΔΤ , where a is the Seebeck coefficient of the thermocouple. For a thermopile with N thermocouples the electromotive force is expressed as: Et = N a ΔΤ. Obviously the Seebeck coefficient doesn’t change with N, but the output voltage is multiplied by the number of thermocouples that consist the thermopile.
3. In several points of the new parts that added to the manuscript, I have the impression that the authors treated the results superficially, without understanding the impact of each parameter. For example:

Lines 548-549: “It should be noted that more important parameter is the temperature span. It decides about the sensor resolution”. How it “decides” about the sensor resolutions? What is the relation of the temperature span and the resolution of the device?

Lines 617-618: “Consequently, the N-sensor characterizes with better measurement resolution”. What does “characterizes” mean? What is the definition of the term “measurement resolution”?

Line 619: “The third effect is an improvement of the sensor time constant”. What are the values of the time constant of each device type? How the “sensor time constant” is defined?

 

Finally, there are some expression that are not usual in the literature, eg: “answers of the sensors” (lines 489, 550).

Reviewer 3 Report

Three sensor versions have been designed with thermoelectric materials and fabricated. And the authors have improved the manuscript. But the experimental results are insufficient and should improve by technical revision of the manuscript. For improving scientific results, they should show the results at same thermopile leg material with different thermoelectric wires (for example; Ag/Ni, CuNi/NiCr) and/or at same thermoelectric wire with different thermopile leg materials.