C3650-Power denied to access points / PSU showing as bad or disabled

[James990]

Hello all.
Looking for some troubleshooting advice please,

We have several C3650 POE switches within our wifi infrastructure, on occasion and very intermittently we experience a wifi outage. On investigation the access points (Cisco 3802I-E-K9) connected to the switch have all lost power, a show power inline displays a "power denied" error to all interfaces with a POE device / Access point connected with a “show environment power all” displaying one power supply is bad with the other disabled, although the switch remains up and running. A reload of the switch or stack member affected clears this and restores POE to all access points.

We have seen this on at least 10 different switches over a year or so, all have been upgraded to the latest recommended IOS release – 16.12.07.

Power supplies have been replaced, power set to static (to eliminate the power negotiation process) switches have been replaced through warranty however we have seen this fault reoccur on a replacement switch. No log error other than power denied to the relevant interfaces.

As all affected switches have dual power supplies with an average of 15 AP’s on each switch there is plenty of remaining power so power consumption / remaining power isn’t really an issue here.

A Cisco TAC case through our supplier didn’t return any known issue / bug or root cause for this so hoping for advice or any recommended troubleshooting I can carry out to narrow this down with a view to a permanent resolution.

[balaji.bandi · Accepted solution]

Looks for me mosfet bug :

https://quickview.cloudapps.cisco.com/quickview/bug/CSCvj76259

https://quickview.cloudapps.cisco.com/quickview/bug/CSCvd46008

[Arshad Safrulla]

I wouldn't recommend configuring static power for the AP's. Rather I would let CDP/LLDP (make sure one of them is enabled) to negotiate it dynamically. 

Can you post a show environment all from one of these impacted switches.

[James990]

Thank you Arshad, static power was temporary on a few test POE interfaces, they also suffered the power deny issue so configured back to dynamic POE.

[Leo Laohoo]

---

@James990 wrote:
A reload of the switch or stack member affected clears this and restores POE to all access points.

---

1. How often does the switch needs to be rebooted? 
2. Check the logs of the switch.  Are there any port(s) that goes down/up continuously/non-stop? 

I think @balaji.bandi is correct and this is a MOSFET hardware bug.  I would strongly recommend RMA the switch before it is no longer possible (EoS).

[James990]

Thank you Leo, yes looks like the MOSFET hardware fault with the symptoms we are seeing.