Comparative Analysis among Single-Stage, Dual-Stage, and Triple-Stage Actuator Systems Applied to a Hard Disk Drive Servo System
Round 1
Reviewer 1 Report
This paper explores the triple-stage HDD servo system, and compares it with single and double stage servo system. I believe it is an insteresting topic and promising technology. The most work in this article mainly focued on the high precision position control.
Comments:
1 The introduction gives few words on the development of control method for such system.
2 The control algorithm relating to PID is easy for inplementation.But it is a hard work for too many parameters tunning. Especially, the model of plant is required for decouple. But we can not guarantee the accuracy of the model when existing some uncertainty.
3. An experiment is encouraged.
Author Response
Reviewer # 01
Thank you very much for your valuable comments and suggestions. We have tried our best to revise the paper as per your recommendations. Please see the following responses.
Comment 1: The introduction gives few words on the development of control method for such system.
Response 1: Many thanks for your comment. We have added more following information about the development of control technique.
“Already different control method have been developed to control a servo system. PID controller is an ordinary controller technique for controlling any system. In this research, different combination of PID controllers (like PI, PD and PID) have used to optimize the outcome. After that lag-lead controller for VCM actuator, lag filter for PZT actuator and inverse lead plus PI controller for TPC actuator have designed to reach the goal.”
Comment 2: The control algorithm relating to PID is easy for implementation. But it is a hard work for too many parameters tunning. Especially, the model of plant is required for decouple. But we cannot the guarantee the accuracy of the model when existing some uncertainty.
Response 2: PID is a popular controller that is very simple and can be designed without great effort. But the parameter tunning process of PID controller is a challenge, especially for system with multiple actuators. So in that case, the other controllers such as lag-lead controller, lag filter, inverse lead plus PI controller can be applied. Therefore, in this paper, we have analyzed these different issues and provide a comparative stage.
Comment 3: An experiment is encouraged.
Response 3: Thank you for your valuable suggestion. In this research work, only simulation has done by using MATLAB/Simulink. In our lab, we still don’t have the facilities to implement this project. In future, if I will get any chance to implement this then I must do that. However, we are currently developing our lab and as a part of future work, we will implement our simulation study.
The authors would like to thank you once again for your important guidelines. Besides, thank you very much for your valuable time to review the paper.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The authors present a study concerning different control schemes for a three-stage actuator system as proposed for future use in Hard Disk Drives (HDDs). For this purpose the authors use rather standard transfer functions for the three actuation devices and apply PID variations as well as Lag-Lead type of controllers for each stage. They finally conclude that the three stage system is superior to two and single stage systems.
Overall, the paper is potentially interesting, but exhibits some flaws:
1. The English language in the introduction needs extensive revision, even though it gets better in the latter sections.
2. There are some open questions about the three transfer functions for the respective plants (Gv, Gp, Gt). For one, what is the input and what is the output (e.g. is Gv = X(s)/V(s), where X is actuator position and V is input voltage)? The main reason that this is important is the transfer function for the thermal actuator, as I would doubt this form if the input is voltage; this looks more like a transfer function for the standard heat equation where the input is thermal heat flux (which can be negative). If it is deflection/voltage, there has to be some servo-type circuitry around it since electrical currents can only cause heating (i.e. positive thermal heat flux), hence expansion (if not only connected indirectly via circuitry). This has to be cleared up, as the whole study would be useless if this transfer function is not suitable.
3. Another question to the transfer functions is what values the authors have actually used. There are some graphs with resonance frequencies, but no damping ratios or similar are given.
4. The choices of PID etc. seem somewhat arbitrary. Why not e.g. include feedforward control. There are references given where these choices have been made, but (without me having checked the actual reference) one of the references is actually mentioning feedforward control (reference 40)
5. It needs to be more clear what the measure of a good controller is? The authors mention these parameters somewhat casually throughout (e.g. on page 14 they mention that bandwidth is the key parameter to determine the speed), but the goals are not clearly stated.
6. The fact that three stage actuators are capable of performing better than two stage actuators is not surprising. A quantization of how much better would be much more interesting.
7. There are some choices that seem arbitrary - for example in section 5.4, the three frequencies are plainly "chosen". Why have those been chosen?
As minor points, tables 2 to 4 are hard to comprehend, as they consists of many numbers. The reader would probably be helped significantly if the important numbers would be highlighted. Also, related to point 5 above, are all the values needed?
Author Response
Reviewer # 02
Thank you very much for your valuable comments and suggestions. We have tried our best to revise the paper as per your recommendations. Please see the following responses.
Comment 1: The English language in the introduction needs extensive revision, even though it gets better in the latter sections.
Response 1: Many thanks for your comment. We have proof-read the paper again and revise the sentences.
Comment 2: There are some open questions about the three transfer functions for the respective plants (Gv, Gp, Gt). For one, what is the input and what is the output (e.g. is Gv = X(s)/V(s), where X is actuator position and V is input voltage)? The main reason that this is important is the transfer function for the thermal actuator, as I would doubt this form if the input is voltage; this looks more like a transfer function for the standard heat equation where the input is thermal heat flux (which can be negative). If it is deflection/voltage, there has to be some servo-type circuitry around it since electrical currents can only cause heating (i.e. positive thermal heat flux), hence expansion (if not only connected indirectly via circuitry). This has to be cleared up, as the whole study would be useless if this transfer function is not suitable.
Response 2: Actually, we have used the transfer function of thermal actuator considering there was driver to drive this actuator with input voltage. We don’t have any experimental set up to verify this. However we have used this transfer function by considering the theoretical concept, where it has been found that thermal actuator does not exhibit any mechanical resonance. This characteristic seems to be significant for our work.
Comment 3: Another question to the transfer functions is what values the authors have actually used. There are some graphs with resonance frequencies, but no damping ratios or similar are given.
Response 3: We have included the structure of the actuator and actually through observations of the frequency response figures, the resonance frequency can be easily identified. But, yes it’s quite difficult for reader to determine the damping ratios from the figures, however we thought it’s quite not necessary to show the damping ratios since our focus was to emphasis the controller design in the paper.
Comment 4: The choices of PID etc. seem somewhat arbitrary. Why not e.g. include feedforward control. There are references given where these choices have been made, but (without me having checked the actual reference) one of the references is actually mentioning feedforward control (reference 40).
Response 4: Yes, feedforward control could be an option but in our work, we only focused on the comparative analysis of different PID controllers combinations and some other frequency responded based designed controllers.
Comment 5: It needs to be more clear what the measure of a good controller is? The authors mention these parameters somewhat casually throughout (e.g. on page 14 they mention that bandwidth is the key parameter to determine the speed), but the goals are not clearly stated.
Response 5: As your recommendation, the following sentences have been added in the controller section. As a result it will be clearer to the reader about the selection criteria of a controller.
“A controller is employed in a servo system to precise the desired output. A controller is said to be optimized controller including minimum rise time, settling time and percentage overshoot whereas, higher bandwidth, phase margin and gain margin. After observing the simulation result, the optimized proposed controller have been selecting with considering the above mentioned criteria.”
Comment 6: The fact that three stage actuators are capable of performing better than two stage actuators is not surprising. A quantization of how much better would be much more interesting.
Response 6: Yes, a quantization of how much better would be more interesting. Actually, it is compared in the summary Table 4. “It is observed that from the Table 4, triple-stage gives the minimum rise time (71.69 µs – 89.69 µs), minimum settling time (1281 µs – 1312 µs), minimum percentage overshoot (4.7% - 7.0), higher bandwidth (3.671 – 4.356 kHz), higher phase margin (75.2 – 79.9 Degree) and infinite gain margin.”
Comment 7: There are some choices that seem arbitrary - for example in section 5.4, the three frequencies are plainly "chosen". Why have those been chosen?
Response 7: This is the thumb rule for designing a lag-lead compensator widely used for VCM controller design.
As minor points, tables 2 to 4 are hard to comprehend, as they consists of many numbers. The reader would probably be helped significantly if the important numbers would be highlighted. Also, related to point 5 above, are all the values needed?
Actually we have analyzed more and more than mentioned in the paper. Here only optimized data have been included.
The authors would like to thank you once again for your important guidelines. Besides, thank you very much for your valuable time to review the paper.
Round 2
Reviewer 1 Report
Three stages actuator is good topic, but there is no sound work in current version. This is a typical MIMO system, some new development on controller design using modern theory is welcome. Of course, PID controller is a popular method in practice, but we do not find any validating experiments. Thus, if there is no practical experiment, please show some theoretical developments.
Author Response
Reviewer # 01
Thank you very much for your valuable comments and suggestions. We have tried our best to revise the paper as per your recommendations. Please see the following responses.
Comment: Three stages actuator is good topic, but there is no sound work in current version. This is a typical MIMO system, some new development on controller design using modern theory is welcome. Of course, PID controller is a popular method in practice, but we do not find any validating experiments. Thus, if there is no practical experiment, please show some theoretical developments.
Response: Thank you very much for your valuable comment and suggestion. In our lab, still don’t have the facilities to implement this project. We have analyzed dual-stage and triple-stage control architecture of a HDD system with MATLAB/Simulink considering different cross-over frequency. We have done a large number of simulation, only optimized results are presented in the Table 1, Table 2 and Table 3. Finally summary of the simulation, comparison and selected optimized data are shown in the Table 4.
The authors would like to thank you once again for your important guidelines. Besides, thank you very much for your valuable time to review the paper.
Reviewer 2 Report
The authors have addressed all my comments more or less satisfactorily. The language editing did help, especially on the introduction, so language wise it is fine.
Regarding the answers to my other comments, I can basically accept them, but would strongly encourage to include those answers into the paper, especially about the transfer functions (what is input/output and that the transfer function for the thermal actuator was selected by using a first order, avoiding any resonances). The same with the numbers - you have them, why not state them... that would make reproduction of the paper much easier for others.
Further, regarding section 5.4, I can see well known rules of thumb regarding omega_3=Fv/3 and omega_e=3fv, but I have a hard time identifying omega_1=20*pi, not being related to f_v at all. Again, if you state it is a rule of thumb, please state so in the paper as well.
Author Response
Reviewer # 02
Thank you very much for your valuable comments and suggestions. We have tried our best to revise the paper as per your recommendations. Please see the following responses.
Comment 1: The authors have addressed all my comments more or less satisfactorily. The language editing did help, especially on the introduction, so language wise it is fine.
Response 1: Thank you very much for your comment. I was try my best to revise the sentences of the whole paper.
Comment 2: Regarding the answers to my other comments, I can basically accept them, but would strongly encourage to include those answers into the paper, especially about the transfer functions (what is input/output and that the transfer function for the thermal actuator was selected by using a first order, avoiding any resonances). The same with the numbers - you have them, why not state them... that would make reproduction of the paper much easier for others.
Response 2: Thank you very much for your valuable suggestion and comment. I have added the following section according to your recommendation.
“The transfer function of thermal actuator was driver to drive with input voltage and output displacement. According to theoretical concept, it has been found that thermal actuator does not exhibit any mechanical resonance.”
Comment 3: Further, regarding section 5.4, I can see well known rules of thumb regarding omega_3=Fv/3 and omega_e=3fv, but I have a hard time identifying omega_1=20*pi, not being related to f_v at all. Again, if you state it is a rule of thumb, please state so in the paper as well.
Response 3: Thumb rule for choosing the frequency of a lag lead controller has included in section 5.4 according to your recommendation.
The authors would like to thank you once again for your important guidelines. Besides, thank you very much for your valuable time to review the paper.
Round 3
Reviewer 1 Report
please give a more detailed introduction about the actuator system for hard disk drive .
Author Response
Thank you very much for your valuable comment and suggestion. In section 2, different plant model of actuators used in a HDD servo system have been described. Besides this, the following sentences have added at the begging of the section.
“R/W head of the HDD servo system needs actuator to control its position on the disk that converts the electrical energy into mechanical energy. It requires a control input (voltage or current) to drive and the output is the movement of the R/W head.”
Author Response File: 
Author Response.pdf
