USB midi signal lost possibly due to electrical noise

[Firelord]

Does the issue go away without the analogue board connected?

Yes, it does:

I've done another test. This time I disconnected the analogue board and used MIDI-OX to monitor MIDI messages from MD in Fw Updater mode, pressing the left key. No noise was generated (obviously) and the USB connection was stable during this test (I was continuously turning the soldering station on/off).

With the analogue board connected in Fw Updater mode the USB connection was lost during similar testing.

[Firelord]

Yes, it does

I'm dreadfully sorry, it doesn't! I've done more tests and even with the analogue board disconnected the USB connection is still lost. In the previous test I had the digital board pulled out of the enclosure. More tests will follow.

[Firelord]

Ok, tested the digital board outside of the enclosure (LCD, controls and LED connected), same result -- turning off the soldering station instantly terminates the USB connection.

One more thing to mention, when I had the multiplexers removed, no false MIDI signals were caught.

Also, I've recorded a video capturing this issue during the previous tests. It has two parts -- in one I remove the IC's from the analogue board, and in the other I disconnect the analogue board completely yielding same results. Both are in a ZIP archive totalling ~45 MB.

[Firelord]

I *think* I have the USB connection problem sorted out. I've placed a jumper between ground and USB cable ground as shown in the attached image file. Now the USB signal is stable during soldering station interference test. I'm still getting false MIDI activity on random channels -- but hey, at least the connection is more stable.

UPDATE: strong noise still terminates the connection. If this becomes annoying (i.e. I will lose the connection while playing) I will continue testing the device. Otherwise, I guess I am more or less satisfied at the moment.

[Firelord]

I've come up with a solution on how to solve the multiple false triggering on electrical noise problem . Unfortunately, I cannot implement it myself, since it is a firmware solution. But I think it would make MD more solid without a schematic change.

Here's how it goes. Suppose several pads are connected to MD (it is unlikely that one would connect only a single pad) and on electrical noise, power fluctuations, electromagnetic disturbance etc. several channels simultaneously send false MIDI messages. Now, it is VERY unlikely that a human would hit all of these pads AT THE VERY SAME TIME in the order of several ns or more likely ms (should be an option). Thus it is possible for MD to mark those hits as false and not send any MIDI messages. Of course, there is always a possibility that a real hit (or multiple hits) occured at the same time, but the probability of this is not too high, is it?

What do you think, Dmitri? It should not be too difficult to implement (isn't crosstalk suppression based on the same idea?). And I also doubt that it would seriously impact performance, especially if you make this an optional feature.

[dmitri]

I've come up with a solution on how to solve the multiple false triggering on electrical noise problem . Unfortunately, I cannot implement it myself, since it is a firmware solution. But I think it would make MD more solid without a schematic change.

Here's how it goes. Suppose several pads are connected to MD (it is unlikely that one would connect only a single pad) and on electrical noise, power fluctuations, electromagnetic disturbance etc. several channels simultaneously send false MIDI messages. Now, it is VERY unlikely that a human would hit all of these pads AT THE VERY SAME TIME in the order of several ns or more likely ms (should be an option). Thus it is possible for MD to mark those hits as false and not send any MIDI messages. Of course, there is always a possibility that a real hit (or multiple hits) occured at the same time, but the probability of this is not too high, is it?

What do you think, Dmitri? It should not be too difficult to implement (isn't crosstalk suppression based on the same idea?). And I also doubt that it would seriously impact performance, especially if you make this an optional feature.

I will think about it and it may be doable but right now I have no idea how I would test it in real time.

[Firelord]

I will think about it and it may be doable but right now I have no idea how I would test it in real time.

Yeah, I can understand that.

It's weird though that only few people from this forum have experienced problems with noise. I wonder why.