At some point it became apparent to me that given the average time it takes to repair a broken (battery-corroded) KLM-367A patch manager board from the Korg Polysix, I would be better off just making new ones from scratch. Old boards would still serve a purpose: I can still salvage a number of parts from them to use on the new board and thus keep the price down. In any case, I spent a weekend with the schematic capture & PCB CAD program Eagle by CadSoft and the Polysix service manual, and came up with these (warning, images are fairly large pixel-wise):
Crow's KLM-367A schematic
Crow's KLM_367A PCB
Three-day quickturn board houses are great. I got these (and a big Mouser parts order) in time to spend the holiday weekend building boards:
Bare & assembled boards
Closeup of assembled board
A new board in a Polysix
The salvaged parts from an old, ruined KLM-367 that are visible in the closeup of the new board are: the 8048 (IC22), the seven white connectors, the 1K tempco resistor (looks like a red capacitor near the right side of the board), the dual matched PNP transitor pair (just to the irght of the tempco resistor), the NJM4560 (IC27), the LM13600s (IC20, IC21) and the four PCB mounting pillars (the white plastic parts where the mounting screws go).
Substituted parts are a 2N3819 for a 2SK30A (T1), 2N3904 for a 2SC945 (T5), 2N3906s for 2SA733s (T2, T4); the 0.01uF bypass capacitors are now 0.1uF as they should have been. The 6MHz crystal uses 22pF capacitors (instead of the original 10pF), again as it should have been. Signal diodes are all 1N4148, 3.9V Zener is a 1N5228. The test switch is now a pair of dipswitches (easiest solution as opposed to using that odd submini original switch). The prototype board shows multiturn trimmers, production boards will use single-turn trimmers as per the original.
The boards work fine. This is much easier and certainly less vexing to do than try and recover a badly-damaged board.
FYI: at the moment I am only offering a new board to those who have broken boards. The deal is I get $150 and the old board, and the customer gets an assembled, tested and calibrated new board. As mentioned before, providing me with the old boards keeps the cost down (an unprogrammed 8748 microcontroller alone is $18 these days--if I can use the 8048 from an old board thats $18 less I have to charge for a new board). In any case, enjoy.
I want to eventually phase out the old 8048-type microcontroller, so a future edition of the board will use two PIC-type microcontrollers. A PIC16F877 replaces the 8048 and additionally handles MIDI, including MIDI-IN to pitchbend and mod-wheel actions. A PIC16F628 is used to provide a MIDI-IN to keyboard interface; the 16F628 will obtain note data from the 16F877 with negligible delay as the serial data between the PICs is clocked at 1Mbps. Usual MIDI stuff like sysex dumps of patches can of course be done.
One thing to note is the battery is gone. I've placed a 4K serial flashrom on the board, which can hold 128 patches. The serial I2C buss is also available on a header to attach to things like my 2nd VCO board project.
And an important note: this board is not ready yet. It will be many months before I have it all ready to go.
Crow's KLM-367 replacement using PICs
