The Millett Hybrid MAXed
Millett MAX History
In early October, 2006, Colin Toole (forum user "cetoole") posted a new board layout for the Millett concurrently in the Head-Fi,
Headwize, and DIYForums.org forums. The board was unique in its integration of a number of headphone amp options that up until that
time, had been separate or custom offerings. The board included the basic Millett circuit, a fully-integrated DB output stage with
options for MOSFETs, extruded heat sink footprints, an integrated STEPS-like power supply, and relay-delay circuit for headphone
protection on power on/off. By this time, your MAX webmaster had built several Milletts and Diamond Buffer boards, and
enthusiastically started commenting on Colin's new design. Others chimed in, and Nate - the ever-present uber-Millett-Guru - lent
an official approval/assistance to the development. Having experienced the growing pains of developing a board from design to
production Group Buy, Nate's experience and encouragement was invaluable. After a short while, Colin (Mad MAX) ordered the first
group of prototype boards for several forum users. The result is below:
The first prototype had some teething problems, not unexpected for such a complicated prototype. Forum users tomb (MAX webmaster), ottopig, tcpoint,
and ferrari were instrumental in quickly solving these problems. Consequently, a 2nd prototype group of boards were ordered with corrections.
The 2nd prototype was completely successful using BJT's in the Diamond Buffer.
The Millett Hybrid MAXed design was then fully vetted for that output option. The major Group Buy that followed sold over 400 boards to 175 users in over 20 different countries.
Here's Colin Toole's prototype:
Many successful versions of the MAX were built. Early on, though, Head-Fi user Bperboy started constructing a MOSFET version and ran into a bit of trouble. NeilR took up the challenge and constructed the first MOSFET version. His story can be found on the original MAX website at Tweaks -> MOSFET Configuration. After buiding a breadboard version and with a few parts corrections, Neil succeeded in coming up with a successful parts BOM for the MOSFET version. The MAX was now fully vetted for the MOSFET version, as well.
The standard configuration of the MAX was specifically designed for the Hammond 1455-series enclosures. Hammond offers a choice of black or silver in those enclosures. Here's a representative example of built MAXes in both black and silver:
As the months passed, many builders built Millett Hybrid MAXes - following the standard configuration, some with variations, and some whose cases were completely different:
As time went on, dozens of the BJT MAXes were built - its lineage traceable all the way back to the revMH Millett Hybrid with Diamond Buffers. A few more besides Bperboy and NeilR built MOSFET versions, too. Some reports started to trickle in that maybe the MOSFET version sounded a bit bassy with some of the highs slightly rolled off. Unlike the long-vetted BJT Diamond Buffer, the MOSFET Diamond Buffer was based on a schematic that AMB (Ti Kan of AMB Laboratories) had published long ago on the forum at DIYAudio.com:
However, the design of that buffer probably did not anticipate input from a tube. The BJT diamond buffer has a successful history with tube pairing. However, the MOSFETs present enough capacitance, that along with the input impedance of the circuit at left, present a low pass filter at the output of the tubes. This results in a slight drop off in the highs and a possilbe slight reduction in clarity. Note that this was very slight and not reported by many users who built the MOSFET version - only a few.
Head-Fi user Negatron (Michael Gresh) was one of those who was searching for the absolute best combination with the MOSFETs. Around January of 2008, he contacted AMB about the design of his buffer at left and its implementation in the MAX. AMB's reply explained the situation above, and suggested a pair of JFETs for signal input instead of the accepted BJT's.
His detailed response follows:
"Yes, Negatron is right, the MOSFETs have a significant amount of gate capacitance which, with Q2B and Q3B jumpered, would be driven from the Millett's tube stage directly. The tube has a high output Z, which forms a low-pass filter with the MOSFET gate capacitance and causes a rolloff of the high frequency response.
With QB2/QB3 installed, there is now a complementary emitter follower between the tube and the MOSFETs. The emitter follower has low output Z and moves the rolloff frequency well into the ultrasonics.
At Negatron's request I did some simulations of the Millett Max diamond buffer section (alone by itself, sans the tube stage) with MOSFETs. The new twist is that QB2/QB3 are replaced with 2SJ74BL and 2SK170BL JFETs, respectively, making them a complementary source follower. This dramatically increases the buffer's input Z as compared to BJTs. As "seen" from the tube, the buffer becomes a negligible load, which is a good thing.
To make this work, the RB4/RB6 and RB5/RB7 resistance ratios must go back to 1:1 (use 100 ohms for all of these), and RB2/RB3 should be increased to around 1K-1.2K ohms. QB1 should be 2N5486 and RB12 should be 1K ohms per the Millett Max BOM (MOSFET option). This change prevents the operating current through QB2/QB2 from coming too close to the 2SJ74BL/2SK170BL's Idss low range limit (6mA), yet still allows the output MOSFETs to be properly biased within RB12's adjustment range.
The result is quite good. Here is the frequency response of the buffer section from the simulator. Note that the -3dB point is up near 1MHz, which is significantly higher than the tube's frequency response."
"For people who contemplate doing the MOSFET option on the Millett Max, I recommend this configuration."
Several users made the JFET-MOSFET mod to their MOSFET-MAX and reported excellent results. The MOSFET-MAX website was a result of that development history and documents some of the other changes that have took place since that significant modification. Needless to say, the flexibility in the printed circuit board design that Colin Toole created allowed this modification with a very minimum of changes - no changes to the board, in fact: only a few BOM changes. The MAX board proved truly versatile.