Hi all, I've been working on the Mini³ redesign and I thought that time is ripe to show what I have so far. I started this new thread because in some ways this is a new amp and the reader shouldn't need to search through many pages of the old thread to find this.
Background
The original Mini³ project was a collaboration between Morsel and myself, and was announced in February of 2006. It was based on two AD8397 opamps, with twin 9V battery power, and was designed to be housed in a Serpac H65 plastic case. We did extensive development and testing, and built a couple of prototypes.
We found that the circuit was sometimes unstable when powered by two 9V batteries or with a 24VDC wallwart, when driving low impedance loads. The result was oscillation, rapid heating and blown opamps. Tangent's PINT project which was virtually identical also experienced a high problem rate. He pulled the project as a result.
Due to other priorities, the Mini³ project stalled, and nothing was released. However, it was decided that a switch to single 9V battery power was a key solution, which also gives an opportunity to use a smaller, more pocket-friendly case. The heat dissipation is also halved by this change.
Changing the ground channel opamp was another important solution. We determined that the AD8397 did not serve that role well, and two of these opamps is not good for battery run-time due to the high quiescent current. The AD8397 with its rail-to-rail output and high current capability is still attractive as the choice for the left and right channels.
Fast forward to April 2007. After the San Jose headfest meet, I finally found time to restart the project. So, here it is, I present Mini³ v2.
A brief summary of Mini³ v2:
- As before, the opamps are SMD (SOIC-8) but all other parts are thru-hole to be easy to build.
- The target case has been changed from the Serpac H65 to the Hammond 1455C80x. This represents a significant reduction in size (credit card form factor, and less than 1" thick), and is also more robust. The Serpac's screw bosses wear out quickly, and the plastic shell did not provide RFI shielding.
- The board is a simple rectangle shape. When inserted into the bottom Hammond slot, there is room to fit a 9V battery snugly, while the jacks and pots are aligned very close to the horzontal center-line of the front panel.
- As mentioned, this case choice was due to the change to single 9V battery power. For low impedance headphones this is a non-issue, and with rail-to-rail opamps (like the AD8397), the output voltage swing will still be sufficient for most high impedance headphones (e.g., Sennheiser HD580/600/650) to reach very high volumes without clipping (e.g., Ray Samuels Hornet has single 9V power).
- I will test the AD8010 and OPA690 to see how they perform as the ground channel opamp. (EDIT: The OPA690 has been selected)
- The rail splitter is no longer handled by one side of a AD8397. The well-proven TLE2426 will be used, which has low supply current draw, low output impedance, and excellent stability.
- (EDIT: The AD8397 + OPA690 configuration is now the "high performance" variant of Mini³ v2. An "extended-runtime" variant using LMH6643 + LMH6642 opamps has also been added, with ~2.5x the battery life between charges)
- As before, ferrite beads are employed at the output of all three channels to isolate cable and load capacitance.
- The battery charging will be done with a LM317L configured as a constant current source. The previous resistor charging method provides slight advantage in charging times, but was fiddly because it requires the charging voltage and resistor value to be very finely tuned. It is not possible to tune it to work well with both a 8.4V 7-cell battery and a 9.6V 8-cell battery simultaneouly.
- There is an additional onboard 7812 voltage regulator, Along with the LM317L CCS battery charger, they make it possible to use a DC wallwart anywhere from 15V to 24V output as the charger/PSU without fear of damaging the battery or the opamps. Even unregulated wallwarts could be used as long as the voltage does not exceed 24V.
- There is a top-layer V- plane under the power section of the board, and a square area with no soldermask where the back of the 7812 regulator contacts the board surface. Simply apply a little heatsink compound there and then bolt the regulator down. The V- plane then acts as a heatsink for the regulator.
- As before, there are separate power-on and charging indicator LEDs.
- The input/output jacks, volume pot/power switch, DC power jack, LEDs and battery contacts are all board-mounted. There are no wires in the case.
- There is an exposed copper rectangle area beneath each opamp. When using opamps with a E-pad (such as AD8397ARDZ), a small dab of heatsink compound there adds a little bit more thermal dissipation capability. No need to solder the E-pad to the board like the old Mini³ design. With one 9V battery, the heat disspation is only half of what it was anyway.
- The bottom side ground plane has strips of exposed copper (no soldermask) at each edge (where the board would contact the Hammond case slots). This allows the case to be grounded without requiring any wiring. The slots needs to be scraped a little to get through the anodizing on the black versions.
- While the power section is minimally more complex than before, I tried very hard to avoid feature-creep. The goal is still to design an outstanding portable amplifier while keeping things simple (thus, no bass boost, no gain switch, no crossfeed, etc.).
The design is not cast in stone yet, and as always I welcome any feedback/comment. There is no ETA on boards, but I hope to make this development process fairly quick. All the knowledge learned in the old Mini³ attempt have been incorporated into this design, and I have high confidence level in it.
(EDIT: The development and prototyping phase is now complete. The following schematic and layout are out of date. For current information please use the links in the "News" heading above.)
Schematic digram
Board layout
