History & acknowledgementsBenny Jørgensen designed and built a series of three headphone amplifiers which he called Kumisa I, II and III, respectively. The first two of these were IC opamp based, using the Analog Devices AD844. His third, the Kumisa III, was a big departure from the previous two, featuring an all-discrete, all-BJT fully complementary topology, no global negative feedback and a DC servo.The Jørgensen Kumisa III was loved by its creator, but was not realized by many others, partly because there was no pre-fabricated circuit board available; but more importantly, there were thermal instability and DC servo oscillation issues. This is not to say that no other Kumisa IIIs exist, and one review of the original design was posted by olesth, who built one and went on to say: "The ideas behind the Kumisa III design are compelling and very well worked out indeed, plus the designer is apparently a true music lover - something that is not easily said about some people who are into DIY electronics... I have built several amps in the past, tube, transistor and opamp based, but nothing really compares to this one. Sure, differences between amps are usually subtle, but in this case rather obvious." As with most good designs, the Kumisa III has been attractive to a number of members in the DIY community and thus, over time, this modified version of Kumisa III evolved. Alex Cavalli (a.k.a. "runeight"), well known for his numerous headphone amp projects on headwize.com and elsewhere, made several changes to the design of the Kumisa III while retaining the simplicity and integrity of the original concept. These changes evolved over the August to October 2006 timespan. Some of the ideas for the changes were discussed in a thread on headwize.com amongst Alex, awpagan, dBel84, Dave_M and others. An early prototype was built by dBel84, who liked the amp very much (and that is an understatement). It was originally decided that the modified amplifier be called "Cavalli Kumisa III". I have had several large and successful DIY community service projects under my belt, with a good collection of test equipment, simulation and other analytic software, and an efficient setup for laying out, producing and distributing high quality circuit boards and related parts. Thus, during the development phase of this amplifier, Alex approached me to become involved. With my help, Alex's wish was to make it possible for a large number of people to build and enjoy the enhanced amplifier. The Kumisa III had been an intriguing design to me, and this was a good opportunity for me to explore it in depth. I exchanged many private e-mails with Alex while testing the circuit under PSPICE simulation, several of the important modifications (such as the DC servo scheme and the change to JFETs at the input) were verified both on Alex's and my simulators. My collaboration with Alex on this project was announced on headwize.com in November 2006. Development continued in that thread until the we felt the amp was ready for prime time. The final design is described in the Tech highlights and Schemaic diagram sections. Alex sent me a PCB layout concept he drew using ExpressPCB, and I transcribed the design with some improvements to Eagle, so that I could export Gerber format files. This allows me to use more cost-effective PCB fabs. With feedback from forum members, I modified the board layout further and the outcome is the production board that is now available. At Alex's suggestion, the name of the modified amplifier was changed to "Cavalli-Kan Kumisa III". The new name is rather long, so the abbreviation "CK²III" came to be. Due to our confidence with the simulations, and the fact that a prototype was built by dBel84 (albeit the design has changed since that prototype), I decided to go from simulation directly to production boards without another prototype build. This was a bit risky, but having worked extensively with PSPICE on discrete circuits such as the β22 amplifier and the σ22 power supply, I had found that simulations are quite accurate. Utmost care in checking the PCB layout before sending the files out to the fab also helped ensure an error-free product. When the boards arrived, I built one and performed tests before releasing any for shipment. All was well and the amp performed as expected. Over one hundred and forty boards were shipped in the first week alone, so Alex should be happy and proud. dBel84 helped write some of the text that went into this website, his enthusiasm and effort are greatly appreciated. Thanks should also go to Benny Jørgensen for the original Kumisa III, without which this project would not have been possible. The CK²III belies its modest cost and and will surely delight anyone who wants to build a quality headphone amplifier. Main: CK²III Main | Prev: Specifications | Next: CK²III gallery |
