Ground Current MOD protects switch from poor grounding.

[Julian]

Power supply has short circuit protection - if a dead short exists, you won't get power out of it. Must have shorted something unintentionally with your jumper, is all i can tell you.

[Michael Day]

Both WS-12-250-AC Rev E units still run and operate. When a jumper is across C18, just the power LED goes out.

[lligetfa]

Both WS-12-250-AC Rev E units still run and operate. When a jumper is across C18, just the power LED goes out.

Maybe you're reading the rev in the wrong place. Did you read the following post?
viewtopic.php?f=17&t=2584#p18150

[Michael Day]

I put the jumper where the post shows. This is the only model that this happens with. And of course, without having the schematic, I can't find out why it happens.

[Julian]

so just board north (with ports facing you) of the C18 location is a tiny little guy labeled R201 - This is the current limiting resistor for the power LED - I could see bridging it to the left pad at C18 , which would create the symptom you're seeing. The power trace also runs directly next to the left pad, and the solder resist may have been scratched in the process of removing C18 or soldering your jumper in.

[JustJoe]

The MODs described below bonds DC negative to Earth Ground which will help protect the switch from Ground potential shifts between the tower and the AC Service Grounds.

We decided to do this modification to the original design because so many people do not do proper grounding and it will help protect your switch from ground current.

If the MFG date of your switch is after September 9th 2016 then this modification was done.

Since implementing this modification to the design we have not received a single unit that was modified returned for RMA repair damaged from ground current.

OK, this all sounds logical, so we have been doing it on all the older switches.

But now we inherited a site with WS-12-250-AC WISP Switch PCB=REV J switches. I even opened the case and checked between the PSU gnd pins and the Ethernet jack shields ... the meter reads infinite (open) ohms ???? Definitely no mod has been done to connect the two.

So is this "fix" being incorporated or not??

What's worse, Chris goofed in this thread and told a guy in viewtopic.php?f=17&t=2584#p18150 that the "F" in the config box is the board revision. No, if you go read the thread on current-capacity upgrade viewtopic.php?f=17&t=3934 :

"... Another way to tell is if you look at the board's "white stenciling", not the white rectangular box Labeled "CFG" which refers to the firmware configuration to be used, if the white Stenciling says Board Rev G it may already have the MOD done as we had the SMT shop start doing the modification mid production run. If it says Rev H or Rev I then your good. ..."

Well the guy in this thread showed a pic clearly showing Board Rev E , and should be using that instead of the "F" written in the white box.

I bring that up because Rev J I have in question still shows. "F" written in the white config box.

There is only one un-populated C11, which has pwr gnd on one end and shield gnd on the other.

There are two other un-populated C29 & C105 that have the shield gnd on one end, BUT their other ends definitely are NOT pwr gnd.

I looked several times and could find NO installed caps that have pwr gnd on one end and shield gnd, that could then be removed like in the true REV F boards either.

I don't like reviving old threads! But I find this one sorely needs an accurate update for current thinking.

Wouldn't you agree Chris ?

[JustJoe]

Bump ???

[JustJoe]

Bump

[Dave]

Joe

1) the "F" in the CFG box is more of a software configuration identifier...it means that basically the software will use its internal Rev F code for that board. It doesn't really correlate to the board hardware.

) after March 2019 all boards are configured so the CAPs you highlighted are not installed. We determined that we greatly reduced our failure rate, by at least a factor of 10, removing the possibility of an unwanted AC ground path on our board. Basically we removed the network xformer filter caps, and the filter caps connecting the shield to our DC ground. We made a couple of smaller batches where we tried replaced the shield filter caps with high current resistors, but this made no noticeable change in failure rates. Our solution in the end was to remove all xfomer filter caps (which a lot of other manufacturers do not also install), thus getting rid of a possible (unwanted) AC ground going across our current sensors. Our current sensors are rated for 65 bolts, and when you get unwanted AC ground ( a voltage potential difference), the current sensors would blow, sometimes taking the mosfets also.

[JustJoe]

Dave, I know you had other commitments, so I appreciate the informative reply. I wanted to think it through for a couple of days and have this followup.
1) Yes, that confirms the point I was making.

2) I completely understand that solutions can get complicated and often end in a compromise. As I understand it now, at least for the WS-xxx-AC switches, this 03/19 ground update replaces the idea of tying shield & power ground together, since that did not seem helpful at all. So on the REV J board, I should not try to tie them together, just leave it as is. But then this leads to 3 questions:

3) With the removal of the Ethernet transformer filter caps, will the ports be more susceptible to RF noise, for example FM radio transmitters on a shared tower, and can that make it more difficult to maintain full GigE copper links ?

4) What is the recommendation now for PCB REV A-E ? Should we still put in the shield-to-power ground jumpers? It dawns on me that with those in place, a PCB REV A-E would short the cable shield to brown (negative power) pair if it was cabled to a REV J board, basically defeating the new update. Doing nothing on the REV B would leave it exposed as before. But trying to REMOVE the 48 transformer filter caps from both sides of the board as on REV J would get tedious.

5) What would you think of this alternative mod for PCB REV A-E ?? ... Currently, on capacitor C290 (C29), one end seems to be the common tie point for all transformer filter caps. The other end is the connection to the "evil" shield ground. Unsoldering C290 would break that connection, but it would leave all of the xformer caps possibly cross-coupling their Ethernet lines together.

But what if after unsoldering C290, it was rotated 90°, one end soldered back to the common tie point, but the other attached with a short jumper to HDR6, which is instead power ground. (See attached pic) Do you think that would be a easy mod for those ?