oneTesla MIDI interrupter issue

[futurist]

Hi everyone

I'm using opensource oneTesla interrupter for my DRSSTC and I noticed few issues in MIDI mode which were also confirmed by Mads and Hydron
Basically the interrupter in MIDI mode makes random "pop" sounds, sometimes misses a note and at note start doesn't always get the first pulse right

Here are the scope shots, yellow signal is MIDI data

First pulse problem


One pulse at note start, then pauses, then continues playing the note


Doesn't play a note + one pulse at note start


The problem affects both their old and new LCD interrupters, since many people use them I'm curious how no one got around fixing this. Loneoceans sells MIDI interrupter based on (old?) oneTesla code and today I verified it has the same issue.

oneTesla new code, LCD option removed

I tried playing same MIDI on Phil Slawinski's mini MIDI controller and it sounds ok. Also I verified that note start and note stop commands are ok
On some MIDIs those glitches almost can't be heard (if you don't know what to listen for), on some like final countdown and pink panther they sound really bad

Pink panther and final countdown MIDIs played on my coil

oneTesla coil mortal kombat

oneTesla final countdown

Problem doesn't seem to be affected by max pulse-width, sometimes sounds very bad and sometimes it sounds ok. I don't know why
I tried debugging it this weekend with a friend but we failed so I hope someone could help a bit. Thanks!

oneTesla old interrupter
http://onetesla.com/oneTesla_interrupter_firmware.zip

oneTesla new LCD interrupter
http://onetesla.com/media/SDinterrupter.zip

[Garyf]

Ive been using this interrupter for a few months now, in fact i have two of them. To be clear, are you playing local omd files, or using the midi input?
I've noticed some anomolies on certain song files, but i assumed they were caused by the omdconvert process unhappy with too many notes at the same time, or some other pulse width issue.
The big problem i had was with the solder connections on the microSD reader.

[futurist]

I'm using midi input only. Could you try scoping output of the interrupter when playing those omd files?

[Garyf]

Sure.  Whats in the blue trace?

[futurist]

It's fiber optic transmitter output. Signal is distorted because I connected a small speaker directly to 74HC14 output on the receiver side

[Garyf]

Here is a capture during "yakety sax" played from an omd file.  I guess the fastest way to measure fidelity is to listen on headphones for a pop.  I don't see any wide pulses.
It occurs to me that the requirement for no more than 2 simultaneous notes can be enforced in the omd conversion, but converting on the fly with a midi input is more difficult.  Do you get pops when you just tap out your own notes, one at a time?

[Garyf]

Listened to some of these on earbuds, didn't hear any popping.  Attached is capture during Mortal Combat.

[oneKone]

Would you be able to post some pictures of your interrupter futurist?

[futurist]

It's just atmega328p board bodged together with lipo battery and two pots, looks messy but that isn't causing the problem

Listened to some of these on earbuds, didn't hear any popping.  Attached is capture during Mortal Combat.

Could you trigger at note start?



This is mortal combat midi playout I captured

[Mads Barnkob]

Great catch finding this abnormal behaviour futurist!

This is a very interesting issue, because I did notice differences between the 3 MIDI modulators I have.

I got a dual channel from Eric Goodchild, dual channel from Loneoceans and the single channel from oneTesla.

Are you just scoping the MIDI dataline and the opto output in the above shot?

[futurist]

Yes, I'm scoping serial data and optic transmitter output there. At first I thought those single audible "pop" pulses were random, but at closer look you can see they are not
Sometimes at the note start command the interrupter makes one pulse, then pauses, then continues playing the note

Currently in my setup I'm unable to test .omd (SD card) input, but I would be surprised if I wouldn't find the same problem. I tried disabling pitch bend and polyphony and the problem still remains

[Garyf]

I can confirm that I hear it too, now that I know what I'm looking for.  I'd call it more of a skip than a pop. It happens with both midi input and off the SD card in the OneTesla interrupter.  In the first 10 seconds of mortal combat, single notes are played. If you turn the OneTesla interrupter up to full width and put the fiber output an an earbud, you can hear it frequently. But not always on the same note.
I also identified it with midi coming from my PC using Anvil Studio.  I slowed down the tempo and for awhile it happened on the same note each time I restarted. But then it became less regular and predictable.  It's not random, but its not purely deterministic either.
My cheap Hantek doesn't have enough memory depth to capture it very well.  I see the Rigol has 12M pts.