MIDI Thru box

[G. Hoffman]

More of a theory question than anything.

A while back, I made a couple of MIDI thru boxes for my pedal board.  It's a pretty straight forward design - a solid 7805 based power supply, and MIDI spec optoisolator for the input circuit, and I used a 74HC14 Schmitt trigger hex inverter to clean up the transients and drive the outputs.  Very straight forward, very inline with the MIDI spec, and every single test I could try showed it working perfectly.  But my two RJM Mini Effects Gizmos (MEG) never responded well to MIDI signals.  Ever.  I thought they were broken.

So, at some point I bought a Kenton Thru-12, just because I didn't want to reroll a new PCB to make a bigger splitter, and I have never had great luck with daisy-chaining MIDI.  And while I was rebuilding my board last week, I tried it on my RJM MEG's for the first time.  Perfect.  Every single time, perfect.  I tried pluging the MEG's directly into my MIDI interface from my computer - awful.  But plug that same signal into the Kenton - Perfect.  MIDI clock signal running?  Perfect.  MIDI clock AND a continuously sweeping CC controller on another channel?  The MEG caught the one message meant for it perfectly. 

So I open up the Kenton, because I want to see if I can figure out their wizardry.  It's a compact switching power supply, a MIDI spec optoisolator input, and 3 74HC14 Schmitt trigger hex inverters.  It ALMOST looks like a functionally identical circuit.  Same size resistors driving everything (it is, after all, dictated by the MIDI spec).  The only thing I see is a 555 timer and a few passive components attached to it. 

WTF is that 555 doing to create what must be a form of necromancy?  It's been a while, so I don't remember the spec sheet well, but does the 74HC14 have a clock input or something?

[ElectricDruid]

The 555 might be actiing as a pulse stretcher so you get a nice big flash on the indicator LED when a byte arrives. They can be a bit faint if you leave them as-is.

Let's be clear about one thing: If your circuit is as-per the MIDI spec (it certainly sounds like it*) then there's nothing wrong with your box, and there's nothing especially *right* with the Kenton one. There *is* something wrong with the MEG. You've provided a standard MIDI Thru circuit that has been used since time immemorial (yes, MIDI has been around that long now..) and if the MEG doesn't like it, the fault lies there. If it happens to work with the Kenton box for some reason, then that's a happy accident. Rather than look inside the Kenton to find out what it's doing, have a look inside the MEG to find out why it's not working. Perhaps you can fix it so it's not so damn fussy?

* Sounds like your circuit is virtually identical to the Thru part of my own MIDI PCB Schematic

[G. Hoffman]

The 555 might be actiing as a pulse stretcher so you get a nice big flash on the indicator LED when a byte arrives. They can be a bit faint if you leave them as-is.

Let's be clear about one thing: If your circuit is as-per the MIDI spec (it certainly sounds like it*) then there's nothing wrong with your box, and there's nothing especially *right* with the Kenton one. There *is* something wrong with the MEG. You've provided a standard MIDI Thru circuit that has been used since time immemorial (yes, MIDI has been around that long now..) and if the MEG doesn't like it, the fault lies there. If it happens to work with the Kenton box for some reason, then that's a happy accident. Rather than look inside the Kenton to find out what it's doing, have a look inside the MEG to find out why it's not working. Perhaps you can fix it so it's not so damn fussy?

* Sounds like your circuit is virtually identical to the Thru part of my own MIDI PCB Schematic

Yes, pretty much identical.  Not too surprising - it's not a difficult circuit to develop, and is almost completely spelled out in the MIDI spec.

The indicator LED idea is excellent; except it doesn't have an indicator LED.  And yeah, the MEG is definitely where the problem lies.  I'd still like to know how the Kenton does it. 

[FiveseveN]

Well can you trace it out or provide some pictures? Otherwise we're kind of speculating in a vacuum.

[G. Hoffman]

UPDATE:

Once I got the board together, the MEG stopped working again, so I sent a email to RJM.  Surprisingly, they got back to me almost immediately on a Friday night, which kinda shocked me, and absolutely impressed me. 

Unfortunately, he then told me that, with their new version (the MEGX, which I was thinking of getting as it is stereo, and I would only need one!) if it receives CC messages and PC messages at the same time (which doesn't make sense, as MIDI is sent sequentially, but I assume he means too close), it can sometimes get stuck.  I'm guessing the old one is going to be much the same.  Which, unless I'm very mistaken, means they are selling a product which does not conform to the MIDI spec. 

It's frustrating, as I need to send parallel sends, and none of the other small format loop switchers I can find are setup for it.  I could just make something, of course, but it takes time I really don't have right now.

[ElectricDruid]

Unfortunately, he then told me that, with their new version (the MEGX, which I was thinking of getting as it is stereo, and I would only need one!) if it receives CC messages and PC messages at the same time (which doesn't make sense, as MIDI is sent sequentially, but I assume he means too close), it can sometimes get stuck.  I'm guessing the old one is going to be much the same.  Which, unless I'm very mistaken, means they are selling a product which does not conform to the MIDI spec.

Not only doesn't conform to the spec, but doesn't work that well if it can't cope with CC and PC messages together. What on earth have they done? MIDI coding isn't *that* hard. They should have asked on here, there's lots of people who could have helped out!

[G. Hoffman]

Unfortunately, he then told me that, with their new version (the MEGX, which I was thinking of getting as it is stereo, and I would only need one!) if it receives CC messages and PC messages at the same time (which doesn't make sense, as MIDI is sent sequentially, but I assume he means too close), it can sometimes get stuck.  I'm guessing the old one is going to be much the same.  Which, unless I'm very mistaken, means they are selling a product which does not conform to the MIDI spec.

Not only doesn't conform to the spec, but doesn't work that well if it can't cope with CC and PC messages together. What on earth have they done? MIDI coding isn't *that* hard. They should have asked on here, there's lots of people who could have helped out!

Yes, well, it has put me off my plans to buy the updated unit (which IS a better feature set for what I'm doing).  RJM has a pretty good reputation for reliability, so I'm assuming their other gear must be better.  I think I'm going to be looking at the Morningstar ML10, which would also let me get rid of my line mixer because it can run loops in parallel.

[pruttelherrie]

Are you using both CC's and PC's to command the MEGs around? If only one of them, you could build something like a MIDI-filter with an Arduino so that they will receive only one type of message...

[G. Hoffman]

Are you using both CC's and PC's to command the MEGs around? If only one of them, you could build something like a MIDI-filter with an Arduino so that they will receive only one type of message...

Only sending CC's to it.  Being VERY careful about that, since they told me it won't work otherwise.

But I THINK I just got it working.  I'd still like to find something a bit more robust, but if I send a completely meaningless CC after the CC for the MEG, it works fine.  So, I just added a CC on a unused channel to the controller. 

Given my experience with the MEG, though, I'm sure I'll something else will mess up before too long.  I thought I could make do with a Morning Star ML-10, but it won't do what I need either, so for now I'm stuck with two MEG's.  I'd make something (it's not hard), but I just don't have the time.