USB midi signal lost possibly due to electrical noise

[Firelord]

So you don't loose the USB connection any more but still get random signals due to electromagnetic noises, right?

I believe he had used a MIDI connection (via MIDI cables), not USB, during this test.

I would investigate into this problem more thoroughly, but in order to do this I need an oscillograph which I do not have. However, I am quite sure all these problems arise due to noise (USB noise, in particular, terminates the connection). Now, noise can be suppressed by a correctly chosen filter. But which filter design to use? What type of filter is required? Unfortunately, I lack experience in this field and am not able to suggest anything at the moment.

[dmitri]

Then I would suggest to try it over USB in another environment, e.g. another home. If it still looses USB I'd suggest to try:

1. another capacitor between pin 14 and the ground.
2. another resistor between pin 1 and Vcc, and maybe with lower value, say 1k.

[Firelord]

2. another resistor between pin 1 and Vcc, and maybe with lower value, say 1k.

That's the _MCLR pin. Why then choose a lower value for the additional resistor? Maybe a 10k would suffice?

[dmitri]

2. another resistor between pin 1 and Vcc, and maybe with lower value, say 1k.

That's the _MCLR pin. Why then choose a lower value for the additional resistor? Maybe a 10k would suffice?

To pull it to Vcc stronger on an assumption that an an electromagnetic noise forces it to go down causing resets.

[Firelord]

To pull it to Vcc stronger on an assumption that an an electromagnetic noise forces it to go down causing resets.

All right, I had a blurry understanding of the pull-up resistor in this case. It's an inverted input, thus VCC of +5V (logical 1) would mean 0 on reset. Is that correct?

[dmitri]

To pull it to Vcc stronger on an assumption that an an electromagnetic noise forces it to go down causing resets.

All right, I had a blurry understanding of the pull-up resistor in this case. It's an inverted input, thus VCC of +5V (logical 1) would mean 0 on reset. Is that correct?

It is not configured as in input pin, so no logical 1 or. By default it serves as MCLR. It should be pulled to Vcc. Pulling it bellow a some voltage level and then back to Vcc will cause a PIC to reset.

[Firelord]

Pulling it bellow a some voltage level and then back to Vcc will cause a PIC to reset.

In other words, an electrical impulse (~ voltage fluctuation) could generate a reset. And pulling MCLR up to Vcc via 1k res should probably increase stability here. This makes sense.

[der Papst]

Even my Toshiba digital TV looses picture/sound when a fridge turns on/off and becomes totally unwatchable when a kitchen mixer is being used.

...and an ignition-coil from a car is able to disturb and even distroy all kind of electronic stuff.
The reason is a spark in engines (except induction-engines) on the conductor, which emites an electromagnetic wave with a very broad band of frequencies.

Buy a "high speed usb-cable", like delievered for example with externel harddiscs.
They have a mesh of metal around the cable which is connected to earth and are very good protected.

Look for something like this:

[Firelord]

They have a mesh of metal around the cable which is connected to earth and are very good protected.

That's great! Nonetheless at this point we are discussing primarily the design of the USB part of MegaDrum -- it is possible that it needs some updating.

I have a MIDI keyboard (commercial though), and it does not lose MIDI connectivity whether I turn the soldering station off or a rocket is being launched into space next door. Nor does it send false MIDI messages. It is obvious that I would prefer MegaDrum functioning much the same way Hopefully we will solve this problem soon. At least the USB connection issue. I am going to get some 1-1.5k resistors and temper with the board a bit.

[Firelord]

Replacing R22 of 3k3 with a 1k res does not make a difference.