15-year educational experience on autonomous electronic information devices by flipped classroom and try-by-yourself methods

: Since 2001, the departments of Electrical and Electronics and Information and Communication Electronics Engineering of Faculty of Engineering, the University of Tokyo (UTokyo) have jointly given a lecture on autonomous electronic information devices for undergraduate students. According to the on-line questionnaire, 80% of students in 2010–2012 replied that the lecture was useful for their future career. The task given to students was to design and realise an autonomous electronic information devices (so-called ‘the IoT-gadgets’) by themselves, and conduct a live demonstration in front of their colleagues. The device must have an ‘input’, ‘output’, and some ‘information processing’. Being aware of the speed of technology evolution as well as the short hours of the lecture, the professor tried not to give direct answers to students’ questions on how-to instantly build and program an information device. Instead, the students were told to beware of their ‘methods’. This refers to the deductive thinking introduced by René Descartes, and in the lecture's context how students should behave in order to realise the device. In this study, the backgrounds of various associated topics are discussed, such as the UTokyo's educational system, the world's rapid prototyping movement, open hardware, course design, students’ reactions, and


Location and challenges of the lecture in the educational system of UTokyo EE/ICE departments
The departments of Information and Communication Electronics (ICE) Engineering and Electrical and Electronic (EE) Engineering (abbreviated as 'EEIC' by students.) jointly give a lecture on an autonomous electronic information device entitled 'Introduction to Electronic Information Devices (the official title in Japanese is: 'Denshi-Joho-Kiki-Gaku'. The literal translation is 'Electronic Information Device Study'.)'. The lecture mainly targets undergraduate students of EE and ICE, who number approximately 125 every year. The lecture is an optional but a popular one. Fig. 1 summarises the academic calendar of the University of Tokyo (UTokyo) and shows where the lecture fits in the calendar. Since reformation in 2015, UTokyo runs two academic calendars simultaneously, consisting of four terms and two semesters. The EE and ICE are under the semesteral system. The lecture is held during the second (autumn) semester of the third year (it should be noted that school year starts in April in Japan.), which is the third semester of specialised courses for EE and ICE students. UTokyo is one of the last universities in Japan that emphasise the importance of 'Late Specialisation' after two years of 'Liberal Arts' courses under the faculty of Arts and Sciences (the authors find equivalence with 'classes préparatoires (preparatory course)' of elite engineering school system in France.). Students are expected to acquire a broad philosophical view, which we all believe is very important. However it means that all specialised lectures must be given in a very condensed way. Moreover, students start their research-oriented works as early as the fourth year. Hence, the greatest challenge is how to efficiently educate students within a very short time (i.e. within three semesters including the current lecture period). The EE and ICE try to achieve this goal through a discipline-based bottom-up approach. Students begin with basic subjects (e.g. linear circuit theory, computer programming and algorithm, electromagnetism, mathematics, and three-by-half-days of experimental courses) as obligatory courses, and later take on more specific subjects by their choice (the advantage of EE and ICE departments is that they are uniformly operated. There is almost no difference in the obligatory courses selection, and students can almost freely arrange their own lecture sets according to competences and interests.).
Such a bottom-up systematic learning curriculum is useful for lower grades in acquiring academic knowledge and basic skills. As the school year goes on, more emphasis is placed on top-down [1] 'design' aspects. Needless to say, the top-down approach is becoming more and more important to solve the 21st century's persistent problems as well as to encourage innovations in the realworld. Design-led companies such as Apple [2] owe their success to their unique culture that place a large emphasis on product design. Design is not just about modifying the appearance of a product but it is a purely holistic engineering process. Good design is born by contemplating the user's potential needs and providing optimal solutions to the problem of satisfying these needs. Because of the emergence of the Internet of Things (IoT), many electrical appliances are connected to the Internet, providing an integrated service across the network. This paradigm shift has influenced the design process of the devices. Good design can no longer be described based on the performance of the device itself, but is evaluated based on the 'experience' the service can provide to the user. The experience design of hardware devices cannot be easily described verbally. Instead, it is more effective to discuss the idea through the iterative design of actual prototype devices. Rapid prototyping tools such as 3D printers [3][4][5], LASER cutters [6], and computer numerical controlled (CNC) milling machines [7] play a very important role as enabling facilities of the design process of IoT devices. Because of these tools, even novice students can manufacture a simple prototype of the core design at a low cost, making it possible to obtain useful feedback from potential end users immediately.
Thus, the aim of the lecture is to make students work on 'application oriented projects'. The objective is to 'design, fabricate and conduct a demonstration in the classroom of your original electronic information device'. The device must include an (i) input, (ii) output, and some (iii) electronic signal treatment. The choice of application, device, and implementation are left open to students. Through the realisation process, students combine all the knowledge that they are supposed to have acquired from previous lectures and experimental courses, to produce their proof-ofconcept engineering model within a very limited time. Through this assignment, students experience the whole process of system design and development for the first time in their life, in most cases. Professors encourage students to work with one of their classmates. Having a partner when developing the product has positive effects. For example, the proposed ideas can be refined through discussion. By compensating for the skills that each students lacks, the quality of the product is improved. Of course, communication problems sometimes occur but this is also a part of real-world problems.
This paper is a completely revised version of the author's presentation at the EWME 2016 Conference [8]. Extracts of the lecture contents along with more detailed explanations and underlying principles, as well as some ideas on alternative demonstrations, statistical analyses, and feedback from students are presented the article, and future directions are discussed.

Contents of the lecture
The title of the original lecture, first held in the late 1990s, was 'Applied Microprocessor'. The objective was to provide students with the opportunity to use microprocessors. In 2001, the lecture was taken over by one of the authors and was subsequently renamed 'Electronic Information Devices' in 2003, owing to its broadening context. The lecture has been jointly handled by two professors from the EE and ICE departments. The ICE professor teaches the 'upper' (computational and application oriented) layer, while the EE professor takes part in the 'physical' (device materialbased) layers. In EE and ICE, undergraduate lectures are supposed to be open to all (∼55) faculty members' academic areas, hence any professor can, in principle, give any lecture. Three ICE professors subsequently took over, depending on their availability; in contrast, only one EE professor continuously took part, thus providing the lecture with coherency.

Lecture introduction with 'Discourse on the Method'
The problem with a hardware-oriented course is synchronisation with the rapid progress of microprocessor devices [9]. On the one hand, new microprocessor boards are released in the market every year and de facto standards change frequently. Therefore, if a lecture covers only the use of currently available microprocessors, even if students are interested in how to fabricate a demonstration device (to easily earn the credit), the acquired information quickly becomes obsolete, and hence not useful for students' careers. On the other hand, totally omitting the practical aspect is not (at all) appealing to students. Just like many experts in the field [10], the authors were aware of this dilemma from the beginning. One solution is to take more time. According to the syllabus [11], the lecture 'How to make (almost) anything' given at the Massachusetts Institute of Technology (MIT), USA, can be considered as an equivalent. MIT lectures on more practical subjects such as 'How to use PCB design software' and 'How to conduct 3D modelling and printing/CNC Machining' are given for longer hours (180 × 12 = 2160 min in MIT, as compared with 105 × 13 = 1365 min in UTokyo.). In fact, many equivalent lectures take significantly longer: 3600 min in DeVry Univ. [12] (45 h lecture + 15 h project.), Santa Clara University. [13] (6 h × 10 weeks for five credits.), and University de Sevilla (45 h lecture + 15 h lab sessions.) [14], and 3840 min in Beijin Institute of Technology [15] and University Tenaga Nasional, Malaysia [16]. However, the authors cannot take more time because this is 'just a one-slot, two-credit' lecture. Instead, the authors have tried to resolve the issue by teaching both (i) the 'theoretical part' to discuss generically applicable ways of thinking, and (ii) the 'practical part' by showing real working devices.
In the first day of the lecture, students are told to beware of their own 'methods'. As principal thread, René Descartes' Discourse on the method (Discours de la méthode [17]) is introduced. According to the questionnaire distributed during the lecture, very few students read the entire text. Thus, it is a very nice occasion for students to stop and think about philosophy, not because it looks 'chic'; however, because students who have graduated must always take his standpoint in philosophy. Descartes' 'methods' are known as deductive thinking. In the lecture, the four sentences of methods are translated into Japanese (classroom native language). The lecturer gives some interpretations in the context of electronic information devices, so that students can remember these 'golden rules' that are useful for device realisation, as well as their 'original meanings.'

Method #1:
Original meaning: Do not accept anything that has not been proven, even if it appears obvious. → In the lecture: Do not naïvely believe any easy-looking secondary information source such as blogs, and always verify the primary data source. Example (i): To learn about USB, go to www.usb.org and read the original specification document (at least once). Example (ii): To learn about a device driver, refer to the manual page or ultimately look into source code (for that purpose, open source software and operating systems (OSs) such as *BSD and GNU/Linux are useful.). Method #2: Original meaning: Divide a complicated and big problem into simple and small pieces. → In the lecture: Decompose the system into small subsystems. Method #3: Original meaning: Begin with the simplest and most familiar thing, deduce complicated things step by step. → In the lecture: Design and compose the system with smaller subsystems, and connect them with interfaces using the same protocols. Use all knowledge from preceding lectures about subsystems and protocols. Method #4: Original meaning: Once one thread of deductive thinking is established, enumerate again all the steps to verify if anything has been omitted. → In the lecture: Debug the system (evidently), and (more importantly) look at the system and see if the system was the best realisation of the objective and specifications, and more generally, the demands of users.
As shown above, the 'methods' explained in the lecture obviously refer to the method of deductive thinking proposed by René Descartes. However, more importantly, it is a way to assemble knowledge from other discipline-based studies that students have taken before. Interestingly, these classical methods are completely compatible with various modern software development methodologies. The most famous methodology called the waterfall model divides the development process into several distinct phases (e.g. requirements, design, implementation, etc.) and ensures that one should move to the next phase only after its preceding phase has been reviewed and verified [18]. This model is widely accepted because it is easy to understand but powerful enough to avoid putting too much effort into something that is not well thought out.

Considerations on design
Determining what to develop is sometimes more important than how to develop it, especially when available resource and time are tightly constrained. In this lecture, two small tips are given to turn out the ideas that are more powerful and appealing.
The first tip is to question how the proposed idea triggers innovation. Prior to this question, students are asked to pick and introduce a product that spurred innovation. Some students introduced Sony's 'Walkman [19]', others introduced mobile games. The answer can be anything but each student must state how users' lifestyles were changed by the appearance of the new product. Through this discussion, students can acquire a more comprehensive understanding of the difference between incremental improvement and innovation. One of the interpretations of innovation learned in the class was that an innovative technology is something that distracts users from using all the previous technologies.
The second tip is to question 'why the new idea is important to you (students)'. Simon O. Sinek developed what he calls the 'Golden Circle', consisting of three layers: Why (a core belief of why the business exists), How (how the business fulfils that core belief), and What (what the company does to fulfil that core belief) [20]. He points out that companies focusing on what to make and how to make products tend to lose their footing because they do not have emotional connection to their customers. On the other hand, charismatic leaders are very good at making emotional connections with customers by clearly stating why they are doing this. Such story emphasise the emotions that users feel when using the products.

Impact of digital fabrication tools
The innovation by Digital Fabrication Tools is discussed in the lecture. The emergence of various digital fabrication tools have changed the way hardware products are designed and manufactured. Various 3D printers, CNC, and LASER cutters at affordable prices are now available in the market. Many universities prepare their own 'Fab Lab' [21]-style machine shops with digital equipment inside the campus. Outside campus, there are now thousands of 'Makerspaces' such as TechShop [22]. Hobbyists and Makers have access to professional-grade machines without paying the full cost of the machines and maintenance. Such benefits are often described using the term 'sharing economy' [23]. Moreover, all design processes are in digital data format, and there are companies that can produce the device as soon as they receive the digital data through the Internet [24]. Consequently, the turnaround time (TAT) between prototyping and production, as well as costs have become almost zero. Owing to the short TAT and low cost, the Fab Lab is now used not only by university students but also by K-19 students [25].
In addition, the idea of 'open source' has changed the world. The principle of 'open source' was first introduced by the software community (such as *BSD, GNU, and GNU/Linux communities.). The idea, according to the authors' understanding, is to share a common environment to develop new things in an accelerated way. In the mid-2000s, the idea was extended to open source hardware: the specifications are open so that many vendors can develop compatible devices. One of the most successful hardware is Arduino [26]. Both models take full advantage of the 'network externality effect [27]': the more people use it, the more the development is accelerated, and the more popular it becomes. Students must experience the revolution by digital fabrication tools in two ways: one is through use them as a simple use, and the other is by trying to introduce new innovations in another field, possibly by using the same success model. OpenLab analyses are published in literature [28,29].

Hardware section with demonstrations
Four lectures (since 2015, the duration of each lecture is 105 min. Four lectures total 420 min.) are conducted as part of the hardware section. In the lectures, several methods to develop interfaces between subsystems are discussed with live demonstrations. Principal thread also comes from Descartes: Method #1: do not take any a priori knowledge. The lecturer discusses interfacing methods through demonstration devices that he fabricated himself. The hardware section, in contrast to the preceding concept section, is built in a bottom-up manner, and the skill level for each lecture is kept as basic as possible (i.e. slightly difficult for those who have no experience), to trigger voluntary motivation (such as: 'Hey, I may do the same thing or even better than my teacher'.) among students from both EE and ICE departments: i. Demonstration on LED on/off blinking. ii. Introduction of signal amplification by bipolar junction transistor (BJT) and relay. iii. Increasing the number of independent signal lines by latching and multiplexing devices. iv. Data transfer between circuit blocks #1: parallel communication. v. Data transfer between circuit blocks #2: serial communication. vi. Data transfer between circuit blocks #3: hierarchical and abstracted communication.

On-live demonstration of LED blinking:
The lecturer emphasises live demonstrations in front of students. By watching the lecturer's demonstration, students naturally think about their ability to do the same thing in front of an audience. The demonstration begins with the simplest example (cf. Redirecting value zero to a special file turns off the LED. To finish the demonstration, the GPIO port is closed through a special file.
# echo 0 > /sys/class/gpio/gpio16/value # echo 16 > /sys/class/gpio/unexport The purposes of the demonstrations are (i) to show students how easy it is (so that they are more motivated to start doing this with themselves); and (ii) to remind them that accessing the real device through a special file is a generic method for UNIX-like OSs, one that they have learned before. In the lecture, different hardware systems can also be demonstrated, such as Angström Linux on BeagleBone Black [8]. Thus, students can (iii) learn about the portability of the method between different environments.
The lecture is given to both EE and ICE students. In such cases, the lecturers must cope with diverse background knowledge and competency as also mentioned elsewhere [33]. For many ICE students handling GPIO is easy (but not necessarily easy for EE students.). However, they do not necessarily know how to calculate appropriate series resistance for LED (how many EE students know), which is a non-linear device. Therefore, the method to calculate resistance values is explained in the following lecture. The value seems to be easily calculated by Kirchhoff's laws. However, the equation for LED cannot be solved algebraically because it includes an exponential term. The lecturer pauses for a while when the students realise this difficulty, and then resumes his lecture by giving the solution using analytic geometry. The solution is given by the cross-section of two I-V curves of the LED and of a resistance in the same graph (Fig. 3). Analytic geometry is a classical method used to find the operating point in an electrical circuit. Revisiting the classical method in a real context, by explaining that the method can be applied to many problems, has a positive effect on students (especially because some ICE students do not take advanced electrical circuit courses).

Considerations in alternative interfaces for LED blinking -USB I/O:
The abovementioned procedure, 'output 1/0 data from computer (a.k.a. LED blinking)', is the first must-do step for a student to develop an information device. It is in fact a very large obstacle to overcome, especially for students who are enrolled in EE and ICE because they like programming, and not because they like soldering. Showing handy 'alternative methods' of LED blinking is therefore important. One attractive interface to use is USB [34]. USB is a universal device available for almost all computers since 1995, and large scale integration (LSI) companies provide interface integrated circuits (ICs) to convert USB into other communication protocols. The most popular IC in the market in 2016 was the FT232 series of Future Technology Device International (FTDI) Ltd [35] (due to the availability and wide variety of supported OSs). The popular application of such devices is USB to serial converters. Many component vendors including FTDI Ltd., offer a ready-to-plug interface devices using FT232. With these devices, students can learn three principles: (i) Students can try 'serial communication', as the IC is designed to be. The method is simple: just tie the transmit data (TXD) and receive data (RXD) nodes with a metal wire, and launch any communication terminal software (such as minicom software available on GNU/ Linux, *BSDs, and/or OS-provided commands such as cu in *BSD.). The students undergo a 'loop-back communication experience': they can see their typed characters on the terminal display as long as the TXD and RXD are tied together (Fig. 4), and these will not appear as soon as they disconnect the TXD-RXD (physical) line. The students can readily (ii) use the device as 'remote terminal control' of single-board computers such as RaspBerry Pi [31]. Serial connection of the students' PC to singleboard computers is useful for debugging and for demonstration.
The last yet important use of the FT232 interface is (iii) to use 'BitBang mode'. FTDI offers the functionality to turn FT2XX device pins into individual I/O nodes. An open source library libftdi [36] gives an OS-independent (libftdi supports Windows, MacOSX, GNU/Linux, and *BSDs because it uses libusb that is ported on many platforms) and uniform access to such devices. Fig. 4 shows an example of driving DATA3 (RTS in serial mode) line with a simple program employing libftdiprovided sample code. The three demonstrations can be integrated into the lecture separately. First, (iii) LED blinking with BitBang mode will be introduced to show students 'the first step', followed by (i) loop-back and (ii) remote board-PC control for serial communications class.

Interfacing a 'digital' signal to 'real world':
As soon as students try to create information devices, they encounter an interfacing problem with the digital signal (i.e. '1' and '0'). Some naïve students tend to presume that a board computer can generate as much voltage and current as they want, with no limitationsevidently this is incorrect. The lecturer warns them by showing AC100 V devices, such as light bulbs (Fig. 5). The importance of the interface circuit is thereby intuitively understood. A magnetic relay is introduced as an interfacing device candidate, and students can understand that even a magnetic relay a current that is requires too large for a board computer. A typical 5 V relay, widely available in the market, has a coil resistance of 100 Ω and turns on at around 35 mA. On the other hand, a typical GPIO terminal can only feed a maximum of 16 mA per pin. Therefore, a current amplifier is necessary to drive a relay. A BJT is introduced as the best current amplifier device in the lecture. The practical and educational benefits are as follows (i) small-signal BJT (i.e. controllable current around 150 mA, such as 2SC1815-Y of Toshiba semiconductor.) is still (although Toshiba closed fabrication of three-legs discrete devices.) the best available device in the market that students can access (Akihabara Electronic Shop Town in Tokyo, just a couple of kilometres from UTokyo Campus.). (ii) The base resistance value is obtained by the 'identical method' that they learned for LED resistance (cf. Method #3). (iii) As early as 2000, the EE and ICE had replaced BJT with field effect transistors (FETs), both in lectures and experimental courses. Thus, by saying that the FET has not been the unique three-terminal device, and by comparing common source and common emitter amplifiers, the lecturer expects that students to realise that the devices may evolve, but circuit design principles can stay the same, thereby prompting them to think about device evolution and keep their mind open to future devices (cf. Method #4). Mentioning another relay driving circuit deployed in the University of Alicante [33], which is an emitter follower, can further expand the view of students. In addition, (iv) by explaining the use of the fly-back diode to inversely shunt magnetic relays, students can recall their knowledge of electromagnetic induction that they were taught just one year before.

Following hardware section lectures:
The following lectures are conducted in the same way: guiding principles are explained and proven through live demonstrations (Figs. 6a-d).

Student lab work
The lecture is closed and the one and a half month sandwiching winter vacation (year-end and new year vacation, around 10 days.) is given to students for device realisation. A couple of office (or experimental lab) hours are open to students, to give them advice and/or physically solve specific problems that they encounter; for example, by fabricating printed circuit boards. Students can freely choose the fabrication environment according to their preference and skill level. Despite the caricature image of EE students, most of the students declare that they had zero years of experience in electronic hardware device making. The questionnaire in 2003 showed that 79% of students (22 out of 28) had zero years of experience, while 21% (6/28) had a maximum of two years of experience. In 2015, 92.5% (84/91), 4.3% (4/91), and 3.2% (3/91) of students had zero, one to two, and more than three years of experience, respectively (it shows that many novices became interested in starting the experience following the lecture.). The statistics show that this lecture provides the first occasion of electronics implementation work for most of the students. Indeed, every year several students tell professors that the lecture was a good reason to buy their personal electronics tools. However, the 'typical successful story' does not necessarily apply to all students; hence, the Fab Lab must exist somewhere. The authors' experiments with students indicate that mentioning external Fab Lab resources and encouraging them in the lecture (one class of 105 min was assigned for introduction of external fab-lab environments.) did not convince novice students to go there. Instead, a free access 'Fab Lab'-style experimental room inside the campus could encourage spontaneous use more naturally. The authors identified two reasons for this: (i) an in-campus Fab Lab is close to their accustomed place, and (ii) introduction by familiar staff, including professors, can be expected. Of course external Fab Labs remain important in cases when students consider deployment after successful prototyping in the in-campus Fab Lab. In UTokyo, there are three open experimental rooms for students, including that of authors, who maintain a 180 m 2 room. The room entrance is controlled by student ID card and continuous WEBcam surveillance. By following the overtime activity management rules of the Faculty of Engineering, the room is accessible for 24 h. The room is equipped with many digital fabrication tools, including printed circuit board engraving machines (LPKF Protomat S62 and S103), silk screen printing environment, a CO 2 LASER cutter (Universal Systems VLS 4.60 60W LASER), single-color 3D printer (AFINIA H800), 3D CNC machines (OriginalMind KitMill, Roland MDX-40A, and MDX-540S), and soldering and surface mounting environments. The perceptions of students about mechanical body fabrication machines are as follows: 3D printers are believed to be powerful tools for rapid prototyping, but requires the user to be experienced with 3D modelling. Moreover, the printing time is long.
Instead, a LASER cutter is a more intuitive and affordable tool for rapid prototyping because drawing a two-dimensional development diagram is easier [39]. The materials can be wood, paper, and acrylic boards, which are affordable and accessible in do-it-yourself stores. This suggests that main body parts would be better be fabricated by laser cutters, while specific small jigs would be better fabricated by 3D printers. According to Umapathi et al. [40] Mueller et al. [41], it is also possible to fabricate 3D structures using just a LASER cutter. As for electronics, printed circuit engraving machines are the most frequently used and powerful. Most of students are just fine with using double-layer no-throughhole (or even single-layer) PCBs because through-hole and silkscreen printing are time consuming; or more simply, breadboards are used although this is discouraged by lecturers. Another potential option for rapid prototyping is the use of conductive inkjet printing technology [42]. Home-grade inkjet printers can be used to print both single-layer and double-layer circuits for prototyping purposes [43].

Demonstrations by students
Because students presentations tend to go overtime (average time is 8 min, despite 5 min assignment) (Then students know by experience the difficulty professors have in their lectures.), and the number of groups increases, the total presentation time increases year-by-year. Therefore two lecture slots (105 min × 2) and an additional two hour (330 min in total) have been allocated for presentations. The presentations are done in the order of WEBbased self-evaluation sheets and are open to the public.
The demonstration devices can be categorised into three types: (i) input devices, (ii) output devices, and (iii) interactive (autonomous) devices. An example of a typical interactive device (automatic tap-back machine) proposed by Mr Ejiri and Sakuragi is shown in Figs. 7a and b. This machine automatically responds to loud tapping by neighbours through a very thin wall (which is often the case for students' tenement houses): an integrated microphone detects when the wall is tapped, and the machine taps the wall back by pushing a stick with solenoids activated through relay.
Fully custom-made input devices received the highest scores in the years 2014 and 2015. The best device of 2014 was 'Trickey' by Mr Shiro and Ogawa. This is a switchable and user customisable sets of keytops (Fig. 7c). The device was exhibited at the South by South West (SxSW) interactive tradeshow 2015 [44] in Austin, TX, as part of an outreach activity of the UTokyo: Todai To Texas (TTT) [45]. The device was also exhibited to the Kickstarter [46] cloud funding scheme. Despite significant attention from media, funding was unsuccessful. Because he could not prove the uniqueness of his idea (sic), Mr Shiro could not convince companies to invest in the device. Based on this post-lecture experience, he learned the fourth Method: look back at the system to determine if the system was the best one for its objectives and specifications. Nevertheless, in terms of teaching, showing that even students can create commercial-quality devices is encouraging for other students.

Statistical analyses and feedback
In this section, analyses of the technologies employed and students' behaviour during the lecture are discussed. The lecture style has remained the same; the behaviour of students have evolved with changes in technology, yet they have kept their high appreciation for the lecture. Fig. 8 summarises the evolution of processing devices. Owing to the policy of the lecture (Detailed 'how-to' style information should not be given.), professors do not recommend any specific device to students as an 'official technology'. It should be noted that this policy is rare for this type of lectures; in most of lectures found in literature, professors often take a particular architecture and prepare a ready-to-use hardware kit, such as Arduino [47][48][49][50][51], Arduino + FPGA [52,53], Arduino + DSP [54], Raspberry Pi [55,56], and BeagleBone Black [57]. Only one lecture has asked students to choose the processor architecture [58]. Some professors implement a meta-environment to simplify programming [59][60][61]. Evidently, the pros and cons of the policy are 'freedom or easiness'; students can choose any technology that best fits their existing skills; and students must create their own build-world by themselves (hereby the students follow the same method that the professor showed during the lecture. The second significant change in the 2010s is appearance of 'single-board PC' boards. Single-board PC in this article refers to a microprocessor board on which an OS such as GNU/Linux and FreeBSD runs, and users have a direct access to the OS. Other boards such as Arduino and its compatibles do not offer direct access to the internal OS. Typical brand names in this category include Raspberry Pi and BeagleBone Black. Such UNIX-running single-board PCs were quickly accepted by the EE and ICE students because they had already learned programming on UNIX. In contrast, no student used electronic gadget kits in 2015. The authors also noted that a significant change in the attitude of students in UTokyo: because the obstacles to implementation became easier to overcome, the students raised their targets. In 2001, the average student's main target was to develop 'electronic information circuits and devices themselves'. In 2015 the target became 'to do something with electronic information circuits'.

Feedbacks and publication through social media
As summarised in the previous paper [8], the lecture has been highly appreciated. For example, 80% of students replied that 'the lecture will be useful for future career'. Another interesting feedback from students comes from their spontaneous posts in social media such as Twitter and UStream. For a couple of years since 2011, tweets with the hashtag of the lecture have been analysed as the most active tweets in Japan. This kind of  spontaneous activity cannot be predicted nor designed by the professor. It can be said that the lecture has influenced the students, and the active participation of upper grade students motivates lower grade (next year's) students, resulting in continuously active participation of students in the lecture. However, the classical system cannot yet deal with such new media. For example, it becomes almost impossible to apply for patents once devices are presented during lecture. This was one of the reasons why 'Trickey' was not commercialised: no patent application. For the time being, the patent issue is left open to students. Some new scheme will be required for such a rapidly growing area.

Conclusion
Since 2001, an attempt to encourage the undergraduate students to actively work on their own electronic information device invention has been made by the EE and ICE departments of UTokyo. Aside from some modifications to synchronise with the latest hardware, the flipped classroom lecture style was unchanged and remains successful. After philosophical discourses in the lecture, freedom is given to students to realise their own devices. Moreover, with information technology -by putting their work on the database and making it accessible to students, as well as letting them express themselves through the social media during presentationsstudents have spontaneously begun thinking about 'how my class can become better than before'. The 'bottom-up' knowledge acquired from discipline-based lecture is thereby merged with 'topdown' design-based creative work, right before they proceed to their 'on-the-research' based course in their final year.