Aruba rode the lightning

[kmm]

Howdy folks,
I picked up this dead Aruba AP-175 on the 'bay, it's a weatherized wireless access point. Given that it's designed to be mounted on a building/tower, I was half expecting the insides to be completely melted and carbonized from a lightning strike, but was hoping it'd be something silly like a blown fuse or dead cap.

Looks like it did get hit with lighting. 

The obvious damage on the power supply board isn't too bad though. One MOV [RZ3] (I assume , not much left...) and the cap next door [C1] immolated, cylindrical fuse blown, and a large electrolytic [C7] further down the line shot its goo all over.

Not sure if the electrolytic failure is in any way related to whatever surge event took the input protection. It doesn't show any signs of burning, just looks like any other failed cap. I think that's a Nippon Chemicon logo, but given this thing could have been living in Phoenix or Vegas in the summer a "typical" heat-related failure seems just as likely as a surge-related failure.

I haven't repaired many fried power supplies so I'm looking for some tips:

• Should I replace all the MOVs, even the ones that aren't crispy?
• I don't see any markings on the surviving MOVs. I assume any MOV suitable for 120-240V and of similar dimensions would be acceptable, but if anyone recognizes the survivors and can point me to a part # or likely spec it'd be great.
• The PCB under the toasty MOV is pretty well carbonized. I probed around the area and didn't find any measurable surface conductivity - is this likely safe? There's a creepage slot that could easily be extended to the edge of the board with the Dremel if not, but I'd prefer to leave it alone.
• There's a heat-shrunk axial cylindrical component next to RZ3, looks outwardly like a ferrite bead choke or feed thru cap, but is labeled on the silk as "U1"; any idea what this thing is? Gas discharge suppressor maybe?
• Finally, is there anything else that's likely to be dead that I should throw in my shopping cart when I buy parts?

Included some disassembly pics in case anyone is curious as to how an outdoor-rated AP is constructed.

[SeanB]

RZ3 likely was a 275VAC MOV. The shrunk component is a gas discharge tube, most likely made by EPCOS. It fails in a lightning strike by blowing to pieces, otherwise it simply works till you get 1000V across it when it will arc over. That the sleeve is still intact means it did clamp and the current was below the maximum, and was of short enough duration that the device did not heat up. Way more than the VDR could handle though, so it went short circuit and cooked till a breaker tripped. Just grind the slot through to the board end, otherwise the carbon will track with mains applied. Make it deeper as well into the board till you get to undamaged GRP.

Capacitor is likely just old age and hot living, it is NCC, and is after a power factor corrector so likely spent it's entire life running at 400V. I would also be worried that the data interfaces will have had the same high energy applied, and this is, from experience, very likely able to arc over the transformers and fry the Ethernet chipset.

[kmm]

RZ3 likely was a 275VAC MOV.

Excellent, thanks!

That the sleeve is still intact means it did clamp and the current was below the maximum, and was of short enough duration that the device did not heat up.

Is it best practice to replace this as well?

I would also be worried that the data interfaces will have had the same high energy applied, and this is, from experience, very likely able to arc over the transformers and fry the Ethernet chipset.

Yeah, I'm not really getting my hopes up, pretty much bought it with the expectation I'd be getting a nice die cast enclosure and some N connectors, didn't pay much.
That said, the design of this thing looks rather robust. The failed PSU feeds a PoE take off/interface board (pic 3, upper right) that has another set of MOVs and (un-shrunk this time) GDTs, as well as its own ethernet isolation transformer, and I'm not seeing any evidence of damage on it or on the AP logic board itself.

[SeanB]

Would not replace the spark gap, they are pretty robust devices.

Simplest is to grab 4 12V SLA batteries and power it up using the POE board and see if it still works. That way no need to fix the PSU if the interface side is fried as well or the onboard controllers.

[kmm]

Read the manual, it turns out there are 3 models of this AP, -DC, -AC, and -P. Only the -P version can run from PoE, the -AC and -DC versions need dedicated power and can supply PoE. As you might expect from the AC PSU, mine is the -AC model, so no love for PoE in.

Did a little hackery to see if I could get the PSU working before I ordered parts to fix it proper, figuring if the PSU is still viable the rest probably is too.
(Yep, it's a scary sight. Make like Alanis Morissette and keep one hand in your pocket!  O0)

Jumped the fuse, bodged in the closest (in terms of voltage, at least) cap I had on hand, and after putting on my face shield, plugged it in.



~54V out just like on the sticker!



After checking the PSU output for ripple and other funk with the scope, I temporarily reassembled it. Looks like it's still alive, at least to some extent! 


Status LED comes on solid green for about 10 sec, then switches to alternating red/green, which according to the manual indicates "booting, not ready". I don't know a whole lot about the operation of these devices, so at the moment I'm not sure if this is indicating a dead AP board, or just the AP board indicating it doesn't have a config or can't find a management/tftp/whatever server.
 
Another positive indicator is that when I plug an eth cable in the link light comes on and flickers with activity, so it appears some eth phy somewhere is at least partially intact.

I apologize in advance for referencing Alanis Morissette.