Hello everyone ,
I am a new Mikrotik user and I have renewed my personal network equipment with a RB5009UG+S+ router and a CRS304-4XG-in. I use a SFP+ module between the two products and my problem is the excessive temperature with this SFP+ module. Winbox displays me average temperature alerts of 92 degrees Celsius and the journal constantly displays me warnings:
sfp-sfpplus1 high temperature warning! Please, improve module cooling/placement or use different type of modules for this solution
I specify that this module is a Mikrotik S+RJ10 revision 2.16.
I trying to find a issue to this problem to changing te rate but unsuccessfully.
Thanks for your help.
S±RJ10 general guidance
https://help.mikrotik.com/docs/spaces/ROS/pages/240156916/S-RJ10+general+guidance
…or modules are used in passive cooling devices, the network administrator has to ensure additional cooling. The airflow around the device should be increased or the overall ambient temperature should be lowered to keep the temperature of the transceivers within the recommended range.
RB5009 is a passive cooled device so you must somehow provide aditional cooling to the S±RJ10 module
i see that now 
i think to try to put one heatsink on the module with a thermal pad . Thanks for your reply
The RJ10 SFP produces a huge amount of heat. Best solution if you can do it is to swap to fiber modules instead.
Sonic Internet is installing 10G FTTH; their ONT is RJ45 only. sigh
I learned at 2.5G with CRS326 module shuts down above 87 F ambient; CRS326 CRS309 stacked on solid shelf in still air.
I have since come to love SolidRack 10 with 1U separation above and below all networking devices plus external fans.
Next S+RJ10 home is CRS309 or CRS305; both seem to have better heat pathways. If necessary, the FTC11XG!
As it is the last topic on S+RJ45, I put that info here:
Some proof-of-concept tests:
A. Old server radiator fastened to external parts of modules. No thermal pads … just “toch of copper”. No special change in temperatures 76C but “overheating” message disapperaed.
B. An old HDD fan cooler - “vintage” connectors from the unknow territories of drawers helped a lot - and full success 42 - 50C degrees when transmitting. Maybe 5V is enough to power these 2 fans or maybe a speed regulator could be used.
Last test to prove that “size matters”
Left small silent 40x40 fan keeps 55 degrees but the right one is enough to keep 46degs
OK, BartoszP - I LOVE the zip ties holding it together. I know it was a temp test, but I loved it.
At this rate, wouldn’t a 12cm, low speed, fan be better? I mean, this device IS 12cm across, isn’t it?
I have only 90x90 12V 4pin spare one.
Powered with 12V, no regulation & pedal to the metal, gives ~46C, 5V means silence&slow motion but ~60C.
Conclusions:
- the constant intensive air stream is needed to let it go through small sieve-like holes if you mount fan on the case
- smaller silent 40x40 fan has same efficiency as slowed down 90x90
- small fan could be sufficient without metal case but who knows?
@k6ccc
forget zip ties, watch that fancy folk tape used to temporary hold power cables in the plug, you should be really unable to sleep for love now 
I don’t know exactly, only judging from photographs … but device seems to have majority of holes on the top surface and only little elsewhere. With fan covering all top holes, air is pushed out through (scarce) holes at other parts of device, so air speed might be relatively low. With smaller fan air can escape through remaining holes on top of device, allowing for lower needed presure … and since lots if heat is generated by SFP modules this means that heat can escape via shortest path and doesn’t heat the rest of device components.
Plus: large fan pushes lots of air against solid surface wasting energy. In this aspect the solution of old HDD cooler seems better, both fans roughly cover holes on top of device.
From heat management perspective it would be better to suck hot air from top holes … so that cold air enters through the rest of holes and hence cools the rest of device.
My idea behind 12cm fan was less noise. 5V is a bit too low, I’d use about 7 - 9v.
I thought that the case was about the same up and down. Aren’t there any holes down? If there aren’t, the bigger fan would not work - as already pointed. My problem with 4cm fans is the noise - the bigger, the better in this case.
If air is pushed into a case then the stream reaches SFP cages directly as they are near, the stream is perpendicular to them and not too disturbed yet (except the case holes problem). When the fan sucks out the air then stream takes the easiest/fastest path so in effect it could not touch the space between SFP cages. On the other hand, when the jet is pushed inside, it cools the other elements too.
I would check 9V
Yes, this is as old as fans, is it better to blow air or suck it? 
And the answer - as often happens - is “it depends” (but it costs nothing or very little to try the fan flipped over).
If the fan is in contact or very near to the surface to be cooled, sucking should be in theory more efficient, but you never know until you try.
What I find surprising is that there are not yet common “industry standard” heatsinks/coolers for SFP’s, I would think that a 40 mm fan mounted at the front on a sort of sleeve around the receptacle/connector (or a pair of them) would be more efficient.
As well a design like this TP-link one, with an integrated, though smallish heatsink:
https://service-provider.tp-link.com/gpon/xm60a/
would help a lot.
Overheating SFPs are rather a novelty (it really began with 10Gbps RJ-45 modules) … and hopefully new generations will make the problem go away.
The big 12V fan, as on the last picture, powered @7,5V instead of 5V gives ~47C degs under traffic on both S+RJ10. DAC cable reports nice 4C degs