Joined: May 26, 2010
|Posted: Sat May 29, 2010 7:55 pm Post subject:
Roland JX-3P DCO out of tune due faulty 8253?
glad to be on board! I got a JX-3P last winter and realized few hours later that some patches sound weird on random notes and especially when playing chords. After founding out about the diagnostic mode on Florian Anwander's site, it turned out that DCO 2 on voices 4-6 is pitched way higher than it should be. Furthermore, the keyboard did not track pitch on these voices correctly.
Not giving much attention to the schematic first and frankly I didn't really understood it either, I thought it's some tuning problem that could be managed with trimpots inside, and buried the synth in the loft for months.
Getting back to it the last week, I opened it and cleaned the PCBs for cleanness' sake and to make it sure the malfunction is not caused by corrosion. The components looked fine to me apart from some dry yellowish residue near cap C301 or C201. I got a scope then and gone through the calibration procedure. It all did not help the problem.
Next, I began to tap various points on the main board. Supply voltages seemed to be relatively clean, no ripple. I poked around the S/H area (multiplexers) first, just to find out there's some more fine manual on the net for me to digest.
Then, checking the clock of the CPU and the frequency of the master oscillators for, they seemed to be stable. Next came the four 8253 timers which are providing 4x3 square wave outputs what makes up for 2x6 DCOs. There, I think I found it. The fourth timer chip (IC47) is consequently outputting bogus frequencies. This chip is feeding DCO 2 on voices 4-6. All three of its outputs are always at least about one or two octave higher than of the other three chips' (|C48-50), which are looking consistent.
Since I have no tool to read what the CPU is sending, I just swapped it with IC48, which is for voices 1-3, to see if it's really the chip that is misbehaving. The problem has shifted to voices 1-3 accordingly, and 4-6 are now fine. Hopefully that's all. Now, I need to get a suitable replacement for a faulty 8253.
Besides to make a record of the problem, I wanted to ask which chip should I get.
The original is a Mitsubishi M5L8253P-5. I assume, almost every 8253s are drop-in compatible, but I'm a noob and don't know how much small differences in the electrical specs and chip technology could matter.
NOS Intel P8253-5 and NEC D8253C-5 are looking cheap and easy to source. Looking at the datasheets, the NEC D8253C-5 has a max. clock frequency of 5 MHz vs. the M5L8253P-5's 2.6 MHz, and a lower min. high level input voltage of 2 V versus 2.2 V. There's no mention of the process of the NEC chip, anyway it's operating temperature range differ a bit from the Mitsubishi's. The Intel's operating range differs too, but the clock freq. and other electric specs are the same. As for the chip type, N-channel ED-MOS is given for the Mitsubishi and NMOS for the Intel.
I wonder how much these differences can matter. Maybe different chips having different rise and fall times of their square wave outputs. Most likely unimportant soundwise, I guess, but anyway It's always good to hear opinions.