“Dad, the piano’s broken!” My daughter looked up, all knitted brows and tear-brimmed eyes. I held her little kid shoulders and gently slid her sideways on the bench in front of the old Korg electronic piano from Craigslist – I’d built a wooden stand to make it more furniture-like but it still looked unapologetically industrial. I hit several keys and the notes sounded right but were very faint. I adjusted the volume, wiggled a few wires, and turned things off and on. Nothing.
“Yup, it’s broken.”
I directed daughter-san to the living room, where the cat went belly-up and murgled as she approached. I changed the piano’s speakers, checked the power cube’s voltage, and poked at the output jack; nothing helped.
Crap. It actually was broken.
We couldn’t leave it in the shop for weeks because my daughter needed to practice, and I hated the prospect of getting it back with a large bill for shoddy work, only to break again.
That gave me an excuse to try to fix it myself.
The piano was old-school sturdy: 40 lbs of steel and absurdly heavy medium-density fiberboard. A web search found praise for its reliability but no service manual or schematic, and Korg didn’t list it as one of their products.
I flipped it belly-up, like the cat, onto the sofa. The bottom had two dozen benign screws around the edges, and ten wicked-looking ones in the middle that probably held the keyboard onto the base. I removed the less-scary screws, and the piano split into top and bottom shells connected by short bundles of orange wires.
Two circuit boards nestled under the top, one with buttons and lights and one with all the jacks. They could unplug from the orange bundles but were themselves inseparable, like a pair of flat green nunchuks held together by a stiff wire bridge.
Nothing was obviously burnt or broken inside, so I needed to see what the electronics were doing.
My diagnostic tools had dwindled to one cheap Radio Shack meter (I’d sold my boat-anchor of a Tektronix oscilloscope to an obsessive grad student who drove up five hours from NYC, with his girlfriend, to fetch it.) I measured what I could at some random and incomprehensible points, but after a few fruitless minutes I realized that without a wiring diagram, I’d never know where to look.
I had to draw my own schematic of its guts.
I removed the circuit boards and brought them to the dining room table, along with the biggest sheet of paper I could find.
Sketching a schematic is not unlike making an anatomical drawing while you dissect a fish, only it’s much less messy and you end up with arcane symbols instead of fish-bits. I turned the boards over and over, tracing the path of each lead from each component along its track underneath until it connected to another component. The parts were all comfortingly familiar from my boy-technician days, and I found that I could still read them; I remembered how to find part numbers and values and happily recalled exactly what each one did.
The outline of each part and its unique identifier within the circuit (like a little rectangle with R58 and a half-circle with Q101) were all beautifully silkscreened in white paint on the cards. I translated those dozens of physical components, and the connections between them, into their traditional symbols on the paper.
(I love how every business – baseball scoring, newspaper editing, marine shipping – has its own private symbols; they’re efficient shorthand for insiders, effectively mystify outsiders, and provide a very satisfying accomplishment when an apprentice, who began as an outsider, finally becomes an insider.)
With the symbolic parts and their interconnections on paper, I could see what each circuit was supposed to do. It was as though the designer, half a world and ten years away, were next to me to explaining what he did and why. It became obvious which parts formed the power supplies, or served in the amplifiers, or acted as muting circuits. It was merely a matter of following each path that the power or sound was intended to travel along, and seeing where it went awry.
In cases like this where the output is wrong, it’s best to start at the end and work one’s way backwards. I didn’t yet know why the volume was so low, but I knew that an audio signal always hovers around some middle voltage so that it can have the elbow room to move symmetrically higher or lower to represent the sound waves.
As soon as I started measuring, it was obvious what was amiss, though not the cause. The output amplifiers, which should float halfway between their supply voltages, were stuck high. Their reference voltage, which should direct them to the midpoint, was stuck low. I disconnected the circuits from each other to isolate the problem and found that the reference itself was still stuck, and as it consisted of only three parts, one of them had to be bad! Measuring each with my cheap meter was enough to find the culprit. But instead of it being the electrolytic capacitor (which, like that kid in school, is always your best first suspect) it turned out to be a 10,000 ohm resistor, worth about two cents and abundant in my junk bin, that had gone to infinity, just as though it wasn’t there at all.
I found a 10K resistor, soldered it in place, and screwed everything back together in five minutes. The piano played perfectly.
“Thank you, Daddy” was the best reward.
When nerds write. Thanks indeed, Jeff. And “the cat went belly-up and murgled…” 😉
Love how you describe the screws: benign and wicked…
It takes a certain sort of person, 1) to remove those covers in the first place, 2) to enjoy the process quite so obviously, and 3) to have such confidence that he’s got a 10K resistor handy. Thanks Jeff.