Thanks for the detailed description.
We are prototyping a new revision (Rev D) of the SOM that improves the power all around. One improvement inparticular will help your scenario.
The power traces have been increased to reduce I*R drops. We also replaced Q13 with a lower Rdson part - footprint compatible.
For your lab development I suggest either:
a) replace Q13 with the part above if you have the ability in your lab
b) increasing the 5V input up to 5.50V to provide a little more head room for the dV across Q13
Sorry for the inconvenience. Let us know how it goes!
I replaced Q13 and the problem is solved. The PicoZed SOM no longer automatically shuts off power when the module temperature is elevated through self-heating and no fans. Yay!
One additional benefit is that due to the lower ON resistance, there is now lower power dissipation in Q13. Since Q13 is placed right next to the Zynq, reducing the heat produced by Q13 has resulted in the Zynq operating at a lower die temperature compared to using the former Q13 part.
Thanks for your help!
Today I found that the 1.2V drop across the Q13 MOSFET power switch causes the ADM1166 power sequencer to automatically shut off the PicoZed SDR power supplies.
The shutdown problem does not happen until the board self-heats to a Zynq die temperature of about 83 deg C. This happens at a lab temperature of 25 deg C if no fan is used.
When the Zynq die temperature reached 83C, I measured the case temperature of Q13 at 80C. The lab ambient temperature is 25C but the heat from the Zynq conducts to Q13 through the PC board and elevates the temperature. Also, Q13 dissipates about 1W. At this Q13 temperature, the ON Resistance increases high enough to drop the VIN voltage from 4.94V to 3.8V. The ADM1166 shuts down the PicoZed power supplies at that VIN voltage. Additionally, there are power converters on the board that also shut down at that VIN voltage if the ADM1166 firmware is changed to not shut down at VIN=3.8V.
We can temporarily prevent the auto power-off problem from happening by using the PicoZed SDR in a lab with fans. However, we will have to fix this problem to meet our long-term need goal of operating outdoors without a fan.
We would like to operate the PicoZed until the Zynq reaches it's maximum operating die temperature of 100C. To do this, we need to reduce the VIN voltage drop. I see 2 options for this: (1) replace Q13 with a different MOSFET that has a lower ON resistance over temperature and (2) replace Q13 with a wire and do not switch VIN.
For option 2 (that shorts out Q13), does this cause a PicoZed power sequencing problem if the VIN cannot be switched?
Thanks for your support!