Circuit Overview:
A downscaled D-50 (losing chorus and other fx, and pwm, but with 32voice voices instead of 16 plus more samples). Review notes that these sound punchier which isn't surprising given trimming of features and probably increased processing speeds. D20 adds 8 track sequencer.
Service Tips:
Other than key cleanings and some dirty buttons I haven't seen much trouble with these machines. The one I recently got in the contacts were falling apart on. Most of them cracked around the contact and some had fallen out. I decided to turn it into a module :-). AS OF 11/17 though I decided to sacrifice the board to a nicer customer unit. HIS board had one of the most peculiar problems I've seen. The reset line was hanging at 1.4V. The black wires coming into the interconnect by the cpu (quad pack chip with a socket I'm told that is notoriously problematic. ) have the reset on the right two wires. Middle two are 5V and left 2 are grounds. Pressing hard to test on these type interconnects will unseat the wire btw so beware that. There is a 5 pin voltage regulator. Pin 2 charges a capacitor to threshhold. Pin 4 sends out the reset (low) signal to restart the CPU, reverb IC and a couple gate arrays when power is stabilized after 20msec or so I think. Now I tried swapping to the working power supply and still 1.4V instead of 5V that should be present at bootup. Swapping cpu did no good however installing whilE power is on would latch the patch name in the display instead of the 'blocks' it showed all the rest of the time. Interesting. Anyway when I moved the cpu side to side I'd see the reset go up to 5V so I bent all hte pins out to where they were seating a little harder carefully and cleaned. ROCK SOLID...at... 1.4V haha. Well that sucks. I tried cutting the reverb reset line since it's the only chip that didn't have some documentation showing that the reset line is input only. No variation. Yet I pull the cpu and the line goes to 5V! Yet the same cpu works in the other unit just fine. Absolutely bizarre. Defies all logic I can process on it since all the lines are clearly input only or have been eliminated and NONE of them are static leaking w/ cpu out. So when cpu is in obviously it's doing something that pulls that line lower. If anyone ever has a thought on this email me. It'd be swell to fix that board after the ridiculous time spent on it.
Parts:
Roland still stocks many parts for older instruments! You can always order them through your local Service Center and see what happens! Otherwise I have some parts also so don't hesitate to ask.
Contact Sound Doctorin'