One day we realised that The Dreamcast Junkyard account on Twitter (or X, as it is now known...) was followed by none other than Kenji Tosaki, the peripheral development manager at Sega of Japan from the 1990s until 2001, when he retired. He led the design process for classic Sega Saturn peripherals like the 3D Control Pad and the Virtua Gun, but perhaps most importantly to us here at the Junkyard, he was heavily involved with the design of the Dreamcast console hardware, its controller, the Visual Memory Unit (VMU), and even the Samba de Amigo Maracas Controller, to name a few. To have such an important figure from Sega's heyday follow us—of all people—was truly an honour. We knew we had to reach out and try and arrange some sort of opportunity to speak with him.
What followed was a whole lot of machine-translated correspondence, and we must thank Tosaki-san for answering all our questions, as well as his time and patience throughout this process. Another thank you also goes out to Brian Vines and Laurence Goodchild for assisting me with the decision on which questions to ask, as well as Derek Pascarella for some technical explanations.
Of course, we couldn't just let the resulting interview you see below remain machine-translated, so this is where I must extend a huge thanks to translation extraordinaire (and my former Nakoruru fan translation comrade-in-arms) Duralumin, who graciously translated Tosaki-san's original Japanese answers into English.
What was revealed from our conversation with Tosaki-san is a whole lot of great insight into how the Dreamcast hardware and many beloved peripherals came to be. We also thought it was important to use this opportunity to try and dig deeper into some of the more esoteric Dreamcast-related lore, to try and get some light shone on topics that have fuelled debates within Sega fan communities for decades. Why was a second analog stick a no show on the Dreamcast controller? What was Sega's view on adopting the almighty DVD format that would've supposedly helped the Dreamcast go toe-to-toe with Sony's PlayStation 2? We also made sure to pick Tosaki-san's brain about some cancelled Dreamcast peripherals, such as the VMU MP3 player, with the hopes of unearthing new information about them.
While our interview does touch on the Sega Saturn somewhat, it is mainly referenced to assist the explanation of points relating to the Dreamcast's own design. If you'd like to hear Tosaki-san get more in-depth about the Saturn, I highly recommend you go check out his interview with our friends over at SEGA SATURN, SHIRO!
DCJY: Thank you for agreeing to talk with us about your career at Sega, Tosaki-san. We have read that you were involved with the design of the Dreamcast console. The design was definitely more compact than its predecessor, the Saturn, and very futuristic-looking for the time. When you and your team set out to design the console, what inspired the design?
Kenji Tosaki: The lead director and producer on the design of the Dreamcast was actually Mr. Oikawa [Akitoshi Oikawa], who also handled the design of the Sega Saturn. For the Dreamcast, the design of the console itself was contracted out, but Mr. Oikawa personally handled the controllers and other peripherals.
Now, when the Saturn was still in development, we ended up drawing up the design ahead of finalizing how much space all the internals would need. Stuff like the main board, the CD drive, power supply unit, all of that. The original design was quite a bit smaller than what the Saturn finally became. As development progressed, we added more components, and the footprint grew larger.
The preliminary designs for the console couldn't accommodate all the changes, so we had to go back to the drawing board, and it ended up quite a bit different from what we had originally envisioned. It had to be made larger, so we also had to redesign some of the finer details of its appearance to match.
With the Saturn, the technical development —the main board, the drive unit, etc.— and the product design had been sort of progressing on different fronts, and that disconnect caused us a lot of issues. Plus, the technical design turned out to be more complex than expected, so the production and assembly was more complicated. We learned from all of that when we were working on the Dreamcast.
Our top priorities in designing the Dreamcast were to optimize the internal layout and heat dissipation. I think the internal design turned out really nice and clean. Of course, keeping it simple also helped with assembly. We also waited to plan out the console design until the technical design had been locked in to a certain extent, so we didn’t have to go back and rework a lot of things.
Regarding the design itself, we shopped the job around to a wide variety of outside agencies for proposals; not just Japanese companies, but even American design houses. Mr. Oikawa reviewed a ton of ideas.
Ultimately, Mr. Oikawa went with a relatively minor Japanese company’s proposal. They did a very good job of taking Mr. Oikawa’s asks into account when building their approach.
On the conceptual approach, I’ll quote Mr. Oikawa himself:
"The Dreamcast design is made up of simple shapes; squares, triangles, and circles. The point of the triangle points toward the player, representing how the games and online content were to be player-focused. From the console, straight to the player. But then the start button on the controller points towards the Dreamcast. That represents how, in response, the players reach out and connect with the console."
When you and your team were designing the Dreamcast's controller, what were the most important features or aspects that you knew it had to include? It is well known that the Dreamcast controller shares some design similarities to the Saturn's 3D Control Pad, which you also worked on. How much did that controller inspire the Dreamcast's controller?
When we were developing the Saturn 3D Controller, that was the first time we had ever tried to design an analog-input controller. We looked at how we could best make an analog controller, and spent a lot of time working on the basic design, control methods, layouts, and all the associated design questions.
We also looked into all kinds of additional features; stuff like integrated displays, paddle wheel controls, mouse input, motion sensing, infrared, rumble, haptics, voice recognition... many different things. Most of these features proved too costly to have natively on the Saturn 3D pad, but we made sure to design the controller to include an expansion port so those features could be added later.
We had actually planned out a motion sensor and vibration pack, and even made a prototype. We made sure that the controller's cord ran out of the back, since anything to do with motion sensors that you would have to tilt or shake around would be harder to use with the cord coming out of the front of the controller.
For the analog inputs, we felt it was essential to use a sensor that wouldn't experience any change in the analog signal from drift or wear. To accomplish that, we used "Hall" elements. We knew we definitely didn’t want resistive sensors. We didn’t add them for the sake of having a selling point, we just saw it as the obvious choice. It wasn’t until 2020 that we saw other companies start making controllers with those same Hall elements. I guess the market finally caught up to the Sega 3D Control Pad, and it only took 25 years.
When it came to the Dreamcast controller, we applied a lot of what we had learned from the Saturn 3D pad.
As I mentioned, the idea behind the Dreamcast ecosystem was “play and communication.” The console could go online, but the concept extended beyond just that. There were also four controller ports, so you could have four people at once all playing together. Then, we decided we would try putting an individual display on each controller. We had originally conceived it as something like a personal monitor for each player for sports games, with individual displays. The VMU was our solution.
Connecting the VMUs directly to the console would have meant we would have to add four pretty big ports for four players’ VMUs, which wasn’t a great solution from either a cost or design standpoint. So, we decided to explore plugging the VMUs into the controllers instead. Every VMU would be like a memory pack, but with an LCD screen and control buttons; it could have its own independent apps, and when it was plugged into the controller, the screen would serve as an additional display.
Away from the controller, you could connect VMUs together to swap data with your friends, or play games against each other. That was another facet of the “communication” concept. For us on the peripherals team, we figured it wasn’t likely that every kid would have an internet connection they could use to play online, so we focused on ways we could build “play and communication” through controllers and accessories.
We also planned out a built-in gyro sensor, so you could control games by moving the controller, as well as built-in vibration, and a built-in light gun pointing device. It turned out to be too expensive to implement those features natively, so we decided to build two expansion slots into the controller. That would leave room for us to release expansion units after the fact. We expected one slot would usually be occupied by the VMU, so we needed to have one more for other expansions.
The most important decision made regarding the controller was to have memory units connect to it instead of the console. We designed the VMUs for that purpose.
The controllers themselves were developed to work with an interface device [MapleBus] that could also support VMUs and any expansion packs. MapleBus proved essential, as it had the kind of scalability we needed.
For the expansion slots, we had to think about what sort of functionality we would need in the future. For instance, if we put out a rumble pack, the vibrations from that accessory would travel down to the slot and the connector. That’s not necessarily healthy for the components, and we expected these to be used long-term. So, we had to come up with a design for the connectors that would be durable and reliable.
The analog input was also much more advanced than the 3D Control Pad. We improved the precision, and allowed wider strokes. At the same time, the units could be made smaller, and needed less space to install internally.
I really would have liked to have seen the gyros, vibration, and wireless pointing built-in on the controller, but it just would’ve been too expensive.
 |
| The original US patent for the VMU. Credit: Sega Retro |
%20(1).png)
