stevegs

Don't use a 6N138/9! They're linear devices. The PC900 isn't really up to much either. The HCPL2200 is much the best! See here: Best Steve G-S

This thread might be 8 years old - but here goes before anyone comes a cropper with propagation delays... The PC900 is a digital opto-coupler - ie. if the input current to the emitting diode increases slowly, the output will snap from one state to the other when a threshold is reached. As the input current decreases, the output will snap back at a lower level - so you won't get jittering at a critical level. This is called 'hysteresis'. Long cables (MIDI or otherwise) will slow up a signal, so a digital opto-coupler will square it up to maintain signal integrity. Or so it should... The problem with the PC900 is it's quite slow - and even worse it takes longer to react to a decreasing current than to an increasing one (typically 2us for increasing and 5us for decreasing). Result: a skew where an original 50/50 square wave could become 60/40. MIDI runs at 31.5kBd, ie. the time for one byte is typically 300us - so with the stop and start bits, each bit is only 30us. A 3us skew is 10% of this, so if you daisychain several MIDI devices together, data arriving at the last risks being corrupted. If the PC900's rise and fall times were the same, there wouldn't be a problem - nobody would notice a delay of milliseconds, let alone microseconds! The 6N138 isn't digital at all. It's a simple linear Darlington coupler, ie. the output current is (roughly) proportional to the input current. This is translated to a voltage by your pull-up resistor (R6). So you risk ringing, jitter and all sorts of signal corruption problems! You need a fast digital optocoupler, such as the HCPL2200. The rise and fall times are much shorter (typically 160ns and 180ns respectively), so there is negligible skew. Although the maximum cable length in the MIDI spec. is 15m, I've successfully sent a signal down a 100m drum of cable as a trial before using MIDI in our theatre as a lighting control signal - the distance involved there is about 40m. HTH Steve G-S