Imagine a scenario. A new CRS354-48P-4S+2Q+RM in staging environment - works perfectly. You configure the VLANs and whatnot, test the whole thing, everything works as it should. You bring it to customer and install it in his rack. Now suddenly all ports are dead and fault light is blinking when you power it on. No matter what you do the thing simply will not turn on it’s swtichports. Tough luck, it goes back for RMA.
You get a new unit which again is working perfectly in staging. Again you bring to customer, but now you test it on his bench before rack install. Great, it works. Try it again, it works. You mount it in his rack and power it on. Guess what? DEAD.
So now what? You remove the switch from the rack and do another test on bench. Still dead. Now you’re thinking what has changed since 5 minutes ago when it was working and now when it’s not. Only one thing has in fact changed: rackmount ears have been screwed on. Nothing to lose, lets remove them. VOILA, it’s alive and working smoothly.
Turns out one of the 8 screws must have been causing a short circuit of some sort, maybe even hitting the circuitry inside.
I’m posting this in case someone encounters a similar problem. In total around 15 hours of troubleshooting, replacing equipment and RMA process have been wasted in this whole endeavour, and as you can imagine the customer won’t hear any of it.
I was asking as the 6 mm screws are very short, so there is very little that the screw point can go beyond the thickness of the "ear" and of the device case, I was thinking that in factory they had no more M3x6 screws and packaged the thingy with (say) M3x8 or M3x10 mm screws they had handy.
If they are actually M3x6 it means that there is actually not a minimal tolerance/air gap between the case and that little board in the top left corner.
Whether the issue (probably an accidental grounding to the case) is due to the screw head actually touching a board track/wire or by the screw head somehow "pushing" that board deforming something it would be a serious design or manufacturing mistake.
Anyway, thanks for your heads up, but besides this forum thread you should open a ticket with official support, it is not given that Mikrotik personnel will read this.
Countersunk screws - unlike most other screws - are usually measured including the head, M3x5 and even M3x4 do exist, but are really d@mn short.
The countersunk head on a M3 screw should be around 1.7-2.2 mm, if the ears are 2.5-3 mm, and the case is 2-2.5 mm there is really very little margin to use shorter screws.
In your calculations, you're way more generous with the sheet thickness than MT's engineers...
@sebas1221 I can't really make it out, but to me it seems that the little floating board is not sitting horizontally. Does that have to do anything with this?
The ears should be however at least 2 mm (otherwise the countersunk screw at 1.7 mm wouldn't be "flush").
The case could be thinner, but at least 1.5-1.8 mm, otherwise the M3 thread (pitch 0.5) wouldn't have enough grip (not even 3 threads), and self tapping screws should be used instead.
This leaves us with 6-2-1.5=2.5 mm, that would be what the screw tip could go beyond the internal face of the case, when firmly tightened.
Which means that the gap between the board and the case is 2 mm or less.
I personally wouldn't consider anything below 5 mm (better 10 mm) between a grounded metal case and any PCB to be considered "appropriate".
It’s not my picture, I took it from “Servethehome”’s review of this model, so I can’t really say for sure. Our unit is now working with rackmount ears bolted on but without that particular screw.
