[sdiy] (OT) Doepfer MIDI-CV oddity

[Olivier Gillet]

Hi,

I dug into this some time ago as I wondered what was the most
"standard" MIDI in circuit, and as I noticed some wide differences in
pull-up values for the same part. At the time I attributed this as
different choices of the ideal balance between slowness (obviously bad
for the MCU that will decode the signal) and sharpness (might generate
more noise on the board or cause weird sudden current surges?). I
wasn't aware that someone's 6N137 is not someone else's 6N137.

Here are the values I collected by looking at different schematics:

6N137
Roland U-series (1k)

6N138
Alesis Midiverb (470R)
Alesis Ion (220R)
Alesis HR-16 (47k !! then 10k in the latest revisions, they
acknowledge the error in the service manual)
Elby designs Monowave (1.5k)
Ensoniq ESQ-1 (220R)
MidiBox (1.2k)
Narrat1ve WTPA (220R)
Technics Digital Pianos (2.2k)
Yamaha drum pads (270R)

TLP552
Korg DW-8000 (620R)
Roland Juno-106 (1k)

PC900
Sequential Prophet VS (270R)
Oberheim Matrix1000 (270R)
PPG Wave (220R)

PC910
Korg Wavestation (1k)

When I did some tests what I observed was the following: the signal
with a 6N137 with a 10k pull-up has roughly the same shape as a 6N138
with a 220R pull-up - so I use these two combinations depending on
what I have at hand. With a 6N138 I can't go above 2k without noticing
dropped bytes (signal decoding courtesy of an ATMega series micro). I
have observed weird latch-ups (?) with a 6N137 and a pull-up < 1k, but
I can't remember if this was due to weird layout/breadboarding or
insufficient bypass cap. All those were with Vishay parts.

Olivier