Board power via USB UART click mezzanine - expected?

I can't picture the circuit arrangement and I don't have these boards to examine but the MIKROE-1985 USB-UART Click can generate full RS232 signals, which means it probably generates +-12 volts on the card. I have no idea if these voltages can make their way to the indicator but it is something to check.

Here are a few relevant links to various schematics:

MIKROE-1985 https://download.mikroe.com/documents/add-on-boards/click/usb-i2c/usb-i2c-click-schematic-v101.pdf

96Boards Click Mezzanine https://www.avnet.com/opasdata/d120001/medias/docus/197/HW_Click%20Mezzanine_v103_Output%20Files_PDF_Click%20Mezzanine_v103_Schematic.PDF

Ultra96-V2 /products/devtools/avnetboardscommunity/m/files/1271

With an active USB cable plugged into MIKROE-1985 while the Ultra96-V2 12V is NOT plugged in:

• MIKROE-1985 net VCC-5V is active, which enables LDO U2 so VCC-3.3V is also active. Both tie to the Click Site, which connects to the Mezzanine. Both VCC-5V and VCC-3.3V have protection diodes in place preventing backfeed of the USB connector or U2.
• The Click Mezzanine connects the MIKROE-1985 VCC-5V to VCC_5V, which goes directly to the mezzanine connector and over to the Ultra96-V2. MIKROE-1985 VCC-3.3V connects to mezzanine VCC_3V3, which is also generated from LDO U7 on the mezzanine, but VCC_3V3 does not connect back to Ultra96-V2. Depending on how mezzanine SW1 or SW2 or set, then both VDD and VREF will also be active with either 5V or 3.3V on the mezzanine. This will enable one side of several level shifters as well as the mezzanine ADC U5. However, none of these signals connect back to Ultra96-V2. Therefore, it would appear the VCC_5V on LS connector pin 37 is the culprit.
• Ultra96-V2 LS Mezzanine (J5) pin 37 connects to net +5.0V. The VIN_LED is D17 and is driven by net +VIN. +VIN does not go very many places
  • +VSYS_IN connects to the LS Mezzanine, but is not connected on the 96Boards Click Mezzanine, so no connection there
  • U19 source voltage for +3V3_PRI, which is not active, so no connection there
  • Possible Fan source voltage through R65, which should be DNP. If R65 and R63 were both populated, you could see the problem, but that would effectively short 5V and 12V which would cause a host of other problems, so I don't think that is it.
  • That leaves Infineon PMIC IR38060 U21. The output is +5.0V while the Vin and PVin inputs are tied to +VIN. This is the only possible connection that I can find, so I am now wondering if there is a feedback path from output to input voltage. I will do some more research with Infineon to find out if this is possible. I will admit that I have definitely seen this with other regulators before in my career, so definitely a possibility.

The intention was definitely not to supply Ultra96-V2 board power from a Click Site. There is a feature to allow powering the 12V / +VIN through +VSYS_IN on a Mezzanine, but NOT for the 5V or +VCC_PSAUX rails. Perhaps there is more to the Click specification that I'm not familiar with that allows driving 5V back through a Click board. I have asked MikroElektronika for clarification. I do not believe that the MIKROE-1985 should interconnect 5V and 3.3V between the Board and the Mezzanine, but that is what's happening.

Until we get clarification, I recommend that you do NOT leave the USB cable plugged into the MIKROE-1985 when the Ultra96-V2 is powered down. An alternative would be to clip both the +3.3V and +5V pins on your MIKROE-1985 as the Board is self-powered from the USB. The MIKROE-1985 does not need +5V or +3.3V from the mezzanine.

Thank you for the great feedback! We will work on getting some more answers.

Bryan

The short answer is yes, there is a leakage path from output to input on the 38060 (and most switching regulators). As you may know, supporting power from multiple sources becomes a game of "whack a mole" when more than one source is present, or in this case, when one provides power to a system that isn't expecting it. PVIN and VIN are tied together on the schematic, the reverse path through the high side switch provides at least one path -

If you were to take a meter and measure the voltage on VIN my assumption would be that it is roughly a diode drop below 5V. My bigger concern would be that 5V provides input power to most of the onboard regulators, which means the board will try to power and a standard USB source will most likely not be able to support the load. You then run into cases where some rails may drop, causing sequencing or voltage differential violations. Unfortunately there is no on-board protection against this particular case, which means unless you have the board powered up you need to disconnect any USB cables that can carry power.

My contact at MikroElektronika gave me the following information

Some of the early Click boards were developed without a diode. It was intentionally allowed to power up your board with the peripherals. All recent designs are developed without this feature. The diode is not mandatory, however if you want to protect main board from any unwanted powering from the peripheral, then a diode is the way to go.

We are working with MikroE so that the next version of the 96Boards Click Mezzanine will have this protection diode on the 5V to prevent backfeeding a 96Boards host with 5V from a Click board that may provide 5V to the Mezzanine.

If you are designing your own board with a MikroE Click site, then you would also want to consider protection diodes on any power rail that you don't want to backfeed into your board.

As for the Ultra96-V2, please do NOT leave the USB cable plugged into the MIKROE-1985 when the Ultra96-V2 is powered down unless you clip the power pins on the MIKROE-1985.

Bryan

Thanks for the detailed replies Bryan. I've gone ahead and clipped the power pins of the MIKROE-1985 click board as that seems like the most robust solution.