From Michael E Caloroso Sent Thu, Dec 27th 2018, 06:33
> However, there=E2=80=99s no evidence that any synth manufacturer has prob= lematic MIDI > Thru circuitry that would actually cause bit errors, no matter how far do= wn the chain > they are placed. There are actually a pair of inverter gates ahead of eve= ry MIDI > Thru circuit, and if those are Schottky inverters, such as common 74LS04/= 14/06 hex > gates, then they=E2=80=99ll actually clean up any slew rate problems on t= he input such that > the output is just as good as the original. The '04/06 are not Schottky types, but you are correct that Schottky '14 gates can clean up dirty logic signals. If you buffer the input to the Schottky gate and add a cap of the right value, it can turn a narrow input pulse into a nice square one. Interestingly enough, the MIDI spec - my 1983 or 1988 specs - do not specify a Schottky inverter, they even suggest a discrete inverter. > If someone has a reproducible test case to > the contrary, I=E2=80=99d be very interested to read the details. I documented the routing on how NOT to do it in the future. I was trying to minimize the live rig cartage and avoid using a MIDI switching matrix. But that was before I uncovered the grounding problem, so may have been a comedy of errors. Interesting... Sent from my iLogicMagnifyingGlass, MC On 12/27/18, Brian Willoughby <xxxxxx@xxxxxxxxxxxx.xxx> wrote: > > On Dec 26, 2018, at 6:54 PM, Michael E Caloroso <xxx.xxxxxxxxxxx@xxxxx.xx= m> > wrote: >>> 3) Although 99% of MIDI devices have MIDI Thru wired properly with 0 >>> latency from MIDI >>> In, I seem to recall hearing about a few MIDI devices that have CPU >>> processing on the >>> MIDI stream for MIDI Thru, and that would introduce latency. That would >>> be in violation of >>> the MIDI standards, so I hope you don=E2=80=99t have any devices like t= hat. >> >> The telltale is a device with a port labeled "MIDI OUT/THRU" or only >> IN/OUT ports but no THRU. Buried in the menu system may be a >> configuration to change the OUT port to a THRU. Under this condition >> the "THRU" signal isn't the circuit spelled out by the MMA and is >> actually coming from the CPU on the device, which WILL have some delay >> to it. > > Theoretically, a MIDI device could use a digital multiplexer to switch th= e > MIDI jack between CPU output and a textbook MIDI Thru circuit. I=E2=80=99= m not aware > of any synths actually doing this, but it would be possible to have a > switchable port that doesn=E2=80=99t introduce latency when set to Thru. > > >> I'm not a fan of chaining THRU ports as one too many can distort MIDI >> signals to the point where they are not recognized at the receiving >> device or can appear as another erroneous MIDI message. I'm not >> talking clock messages, I'm talking performance messages like note >> on/off and CC messages. > > Perhaps I=E2=80=99m missing your point, but there is nothing electrically= different > between MIDI clock messages and MIDI Note On/Off and CC messages. They=E2= =80=99re > all 8-bit serial words with 1 start bit and 1 stop bit. If distortion wer= e > to occur, it could affect any bit, and there=E2=80=99s no reason that MID= I clock > messages would be immune to such bit errors. > > However, there=E2=80=99s no evidence that any synth manufacturer has prob= lematic > MIDI Thru circuitry that would actually cause bit errors, no matter how f= ar > down the chain they are placed. There are actually a pair of inverter gat= es > ahead of every MIDI Thru circuit, and if those are Schottky inverters, su= ch > as common 74LS04/14/06 hex gates, then they=E2=80=99ll actually clean up = any slew > rate problems on the input such that the output is just as good as the > original. If someone has a reproducible test case to the contrary, I=E2= =80=99d be > very interested to read the details. > > Brian Willoughby > >