STM32F103 not communicating with computer

[adorno]

My multimeter must be broken, if it was a problem from one computer it wouldn't explain why it doesn't work with others.

I've checked with the scope and it displays the expected 5V and 3.3V   

[Rasz]

lets go further, check if FB1 common mode choke is not open, and track USB +- all the way to the microcontroller
can you make picture of the whole pcb ? both sides if its double load

[adorno]

I checked FB1 and both sides are not shorted. In the tracking red lines I mark each side of FB1.



Both data lines go where they should according to the stm datasheet:
http://www.olliw.eu/uploads/stm32f103rb-pinlayout-wp01.jpg
Pins 44 and 45 (CAN_TX and CAN RX), and the signal can be seen at the microcontroller pins.

Here are two photos of the board, the rest that is not seen is where all the buttons are placed, I took a pic of where the 'brain' is.

[Rasz]

I checked FB1 and both sides are not shorted.

not shorted in reference to ground? or between both sides?

In the tracking red lines I mark each side of FB1.

one of your lines goes thru a cap to the ground, while both usb signals go thru fb1, it has 4 pins, and should pass signals

and the signal can be seen at the microcontroller pins.

wait, what signal? your scope cant capture so there is no way of you knowing if there ever was an attempt at handshaking
if you cant get a scope able to sample at ~2MHz and capture on trigger and your desperate enough you could desolder U3, C18/19, FB1 and jumper over it.
I hope you checked with different usb cable already

[adorno]

not shorted in reference to ground? or between both sides?

Between both sides, FB1 should not have any ground connection. Both data lines goes through the same FB1, but each has its different path, they do not touch each other.

one of your lines goes thru a cap to the ground, while both usb signals go thru fb1, it has 4 pins, and should pass signals

Which cap? I see they have exactly the same path, one goes trough left side FB1, a filter capacitor (that passes high freq signals to ground), and through 220 resistor. The other signal is the same but going through the right side of FB1.

wait, what signal? your scope cant capture so there is no way of you knowing if there ever was an attempt at handshaking
if you cant get a scope able to sample at ~2MHz and capture on trigger and your desperate enough you could desolder U3, C18/19, FB1 and jumper over it.
I hope you checked with different usb cable already

I tried 4 different cables in 4 different computers. I showed what I see on the scope before, and this is what I see on the digital 'hi chinese quality' oscilloscope when I connect it:

https://youtu.be/kUXrbkRoPRY

[Rasz]

one of your lines goes thru a cap to the ground, while both usb signals go thru fb1, it has 4 pins, and should pass signals

Which cap? I see they have exactly the same path, one goes trough left side FB1, a filter capacitor (that passes high freq signals to ground), and through 220 resistor. The other signal is the same but going through the right side of FB1.

nvm, you painted red line between cap and fb1

https://youtu.be/kUXrbkRoPRY

not very differential, D- should be pulled high to signal low speed, but you have both sides pulled high(one of them so hard your "scope" cant even see it going down), wouldnt hurt to remove U3 and recheck

[adorno]

nvm, you painted red line between cap and fb1

Oh sorry, I think it's a weird effect due to the macro lense. fb1 has 4 pins, right below 'F' (FB1 written on the board) there are 2, and below '1' the other 2, so both data lines pass through the same ferrite bead.

not very differential, D- should be pulled high to signal low speed, but you have both sides pulled high(one of them so hard your "scope" cant even see it going down), wouldnt hurt to remove U3 and recheck

Ok, will do. What should I expect after doing it?

[Rasz]

Ok, will do. What should I expect after doing it?

there is a slight possibility diode array is clamping one of the lines high, or microcontroller itself is doing it, or transitions are too quick to show up on the 1MHz scope toy. You could try triggering on ~2V with analog scope and check it it catches anything at all one the faulty line.

btw low speed usb can be logged/captured/debugged using $4 "CY7C68013A-56 EZ-USB FX2LP USB2.0 Develop Board Module"
https://sigrok.org/wiki/Lcsoft_Mini_Board

[adorno]

I finally got time to continue this, removed U3 and the keyboard now works flawlessly so I'm going to order another diode array and replace it. Would it be bad to use the keyboard without U3 in the meantime?

Thanks a lot for all who helped, specially you Rasz

[coromonadalix]

the tvs is a protection against spikes / transient voltages,  serve as voltage limitting too,    just dont connect the usb cable when powered  ... you may have ground voltages potential / differences when unconnected  ...