Arduino R3 Uno Clone - When barrel plug connected there is 5V on USB. Why?

Question

I have been using Arudino Megas for some projects and never had this problem.

Simply enough - I am powering arduino with 12V in via barrel and take power out via VIN to 12V low very low current small fans. I want operating model, where when USB is disconnected it shuts itself down.

I have made some plan with R3 Uno. Build it up, it behaved well. However when I disconnected USB I noticed there is still 5V on USB.

Why? I understand there might be some small voltage but full 5V as if connected? Something is fishy to me and not sure what. Mega did not behave like this.

• st2000: Is this a real Arduino R3 Uno or perhaps a clone? It does not sound like it is behaving correctly (see my answer below for details as to how I believe it should work). Maybe the transistor T1 is shorted?
• st2000: Hang on, can you try removing the fans and testing the USB for voltage again? Also, if you have a volt meter, can you check to make sure the supply voltage is actually 12 volts with and with out the fans? (If the fans draw enough current and the supply can not handle it, the voltage may drop. If it drops below about 6.6 volts there might be some voltage seen at the USB port.)
• Petr: st2000: Yes, I have tested all with my trusty midrange-multimeter. I have sw logic to slowly ramp up PWM to avoid overcurrent and fans in cage to avoid human factor. But I unsealed another unit to verify it was not me -> And yep, this batch works like this. Good on me to verify everything before shipping this as release candidate. Tested one D1 W2 (R2 clone) and one R4 original, plus couple of Mega 2560 boards and non of them bring up 5V on USB. One Mega2560 brings some 300mV on USB rail, but then this has been study Mega... it went through hell.
• timemage: "there is still 5V on USB" I'm just guessing you're measuring this VBUS and GND on a disconnected A plug. For your own sanity and for anyone helping, you should probably take that measurement with some load between the meter positive and negative, e.g. 10k. In other words, take a "Low Z" measurement. Some meters have a setting for this sort of thing.
• KIIV: Another posibility for 5V backfeeding USBVCC is for example combination of 7-9V VIN and shorting 3V3 to 5V - threshold jumps to 10V instead of 6.6V

Answer 1

Looking at this part of the Arduino UNO Rev3 schematic, it appears that if half the voltage of the unregulated power (12 volts (shown in green) according to your post) is above 3.3 volts, then transistor T1 (circled in red) will disconnect the net "+5V" from the net "USBVCC".

enter image description here

If this is the case, there should be no voltage seen at the USB port when powering the Arduino UNO Rev3 with a supply greater than 6.6 volts.

• Petr: Thanks, this makes sense to me and what I thought. Thanks for ref to schematics, will make more use of that. I am not primary from low-voltage world, I come from world of "if are honest about making software, you make your hardware", so was trying to verify. Nope these are clones have 4x of this batch at home and all behave the same (incl ones I have not tried to push.)