Title link takes you to the details.
Update: The original site appears to be gone. The following is from the Internet Archives WayBackMachine. Unfortunately the additional images referred to further below are gone, however you can make out the mod (albeit blurry) to the left.
"SCI MultiTrak (Model 615) battery drain fix
So you found this page. Maybe intentionally, as you don't like the lifetime of your backup battery of your MultiTrak, or just by browsing. Anyway - if you have problems with your backup battery only keeping up power for several months, you may want to modify your synth a bit. I found this fix today, when i searched the last problem of my MultiTrak after the monster repair i initially did on this synth. After the repair, i had a temporary battery in it, a button cell 3V lithium battery. It took about 3 months to drain the power from this battery. Even after the second one, after 3 months, end of lifetime. So i purchased the large cell this synth usually needs, but as i didn't feel very comfortable with this very short lifetime, i didn't put it into the synth. Until today, where i wanted to track down this problem and find a solution, if there is any. And i did.
For i impatient and the ones not interested in the technical description of this problem, just the following show modification is needed to fix the problem.
Modifying the MultiTrak
In fact, my MultiTrak had a design bug, which lead to the short lifetime of the battery. If you want to fix this, disassemble your MultiTrak (i don't describe the procedure here, you can find out yourself or somewhere else i think), remove the keyboard cable and put the electronics assembly in front of you, with the connectors away from you. Remove the voice board (the upper board of the stack of 2 large boards) to reveal the CPU board. Near the keyboard connector, you see a large 40 pin chip, which is the keyboard processor, and right of the keyboard connector, you find a CD4022 (maybe it has other letters, but it must have the number 4022 on it). Left of this chip, there are 2 resistors of 10KOhms. The left one (more away from the 4022) is the one we neeed to touch. Desolder the pin of the left resistor which is nearer to you (goes to a thick trace, which connects it with its neighbour resistor), you can also cut it, but we need to connect it elsewhere, so be careful when cutting it. Next, isolate the board and maybe the neighbour resistor, that your modification can't touch something it shouldn't. Now connect the resistors open end to pin 16 of the 4022. That's it. Maybe you want to fix everything with some hot melt glue or another piece of isolation tape. Of just use a piece of wire stiff enough to keep everything in place.
Have a look at the photo (sorry, it is very blurred, i seem not to be capable to do those near-shots in a better quality :) to see how it has to look.
If you are not interested on what you did in detail, you are done now and may reassemble your MultiTrak. Otherwise, read on.
What happens in this modification?
If you want to know, you may want to first measure the current flowing through the battery. On my synth, it was around 200µA. After the modification, it is around 1µA:
So why that? Let's have a look at the schematics:
(Strange. This near shot worked...) You see the 4022 here. As you can see, it gets power from Vnv, the non-volatile power. Besides this, what you can't see here, the RST pin is pulled high by the same signal turning off the RAM access when the main power goes away. Read the datasheet. Pulling RST of the 4022 high sets CARRY and Q0 high. For Q0 this doesn't matter if you don't hit any keys on the keyboard. And even if you would, it should make no trouble, as the CARRY signal also doesn't leak to the keaboard scanner (partially left in the photo). But there is a completely different problem. The 10KOhms resistor used as pullup for the CARRY signal is connected to +5Vd, which is the main power for all other electronics. So you have a high signal on CARRY. A signal of maybe 2,5V. This goes over the resistor, at the other end (the +5Vd) around 0,5V are left from this power. Use your calculator - 2,5V - 0,5V = 2,0V. Using Ohm's law, R = U / I, you get I = U / R, 2,0V / 10KOhm = 0,2mA = 200µA. Quite a lot. If your battery has 1000mAh, you need 5000 hours to drain the battery. 24 Hours per day (assuming you don't turn it on all the time) = 208,3 days and your battery is gone. After the modification, the NV circuits draw 1µA from the battery according to my multimeter. That's 1 million hours then, which again is 41,7 years. No, don't think "i never have to replace this battery again" - lithium batteries have a lifetime of about 10 years, so it's not the synth drawing the power, it's the battery itself, which leads to you replacing the battery after some years again. Maybe 5, maybe 10 - but not one year, as it might be the case until now."