MikroTik CRS309-1G-8S+INL -- 10G RJ45 Transceiver?

Uneasily3723 · Tue Aug 29, 2023 12:58 pm

Hi there

I have a router that has a 10G Wan Fibre but on the lan side there's only 1x 10G RJ45 port. So I bought a MikroTik CRS305-1G-4S+IN switch becuase on my home network, I have another two CRS310 switches. I thought I can just get a 10G RJ45 transceiver and connect the CRS305 with the router. Well, it works but it also overheats. I tried attaching heatsinks and fans but it still overheats. So I'm looking now for something else.

I saw the CRS309-1G8S+-INL. To me it seems that it has much better (still passive) cooling than the CRS305. However I'm not sure if that cooling is sufficent because the 10G RJ45 transceivers produce a lot of heat. Will I also have overheating problems with the CRS309? If so, what else can you recommend, so I can you the 10G RJ45 LAN port on my router/modem?

mkx · Tue Aug 29, 2023 2:09 pm

10Gbps RJ45 modules from any vendor dissipate quite a lot of heat and are thus problematic when used in passively cooled appliances. Mikrotik published a general guidance for S+RJ10 modules with some suggestions. But these suggestions only improve the problem so far.

So: either use actively cooled switches or start using DACs or optical SFPs for speeds higher than 2.5Gbps.

Uneasily3723 · Tue Aug 29, 2023 2:15 pm

As I said. My modem/router has on the LAN side only 1G RJ45 ports and 1x 10G RJ45. So I need to use that RJ45 on the router/modem.

What solutions are there? I need 2x SFP+ for my home network and then somehow connect it to the router/modem.

mkx · Tue Aug 29, 2023 3:53 pm

I'd say that any of passively-cooled solutions will somehow bear with single 10Gbps RJ45 module. But not with more of them. If you have to use UTP cables to connect your gear to 10Gbps switches, then your only option is to go with actively cooled device, such as [rl=https://mikrotik.com/product/crs312_4c_8xg_rm]CRS312-4C+8XG-RM[/url] or any of actively-cooled *XS* models.

Uneasily3723 · Tue Aug 29, 2023 4:04 pm

I'd say that any of passively-cooled solutions will somehow bear with single 10Gbps RJ45 module. But not with more of them.

I just need one to go to the router/modem.

If you have to use UTP cables to connect your gear to 10Gbps switches, then your only option is to go with actively cooled device, such as CRS312-4C+8XG-RM or any of actively-cooled *XS* models.

Thanks for the recommendation - the price is in a different league. You mention UTP cable. Does it matter whether it's shielded or not? I'm using currently a shielded cat 7 network cable.

The CRS309 has 8x SFP+ ports and on the back heatsinks. If I juse use S+RJ10 module in slot 1 and use the 7 and 8 for the 10G fiber transceivers, could that be run (as suggested in S+RJ10?

The CRS309 is here 1/4 the price of the CRS312.

ConradPino · Tue Aug 29, 2023 8:07 pm

SFP+ module make and model:

MikroTik S+RJ10 in CRS326 port 25
Arista Networks SFP-10G-SR in CRS326 port 26
Arista Networks SFP-10G-SR in CRS309 port 5,6,7,8

Note the module temperature readings in ambient 75F room. The S+RJ10 is very warm, almost hot to the touch.

CRS326 port 25,26

/interface ethernet monitor sfp-sfpplus1 once
                               name: sfp-sfpplus1
                             status: link-ok
                   auto-negotiation: done
                               rate: 2.5Gbps
                        full-duplex: yes
                    tx-flow-control: no
                    rx-flow-control: no
                        advertising: 10M-half,10M-full,100M-half,100M-full,1000M-half,1000M-full,10000M-full,2500M-full,5000M-full
           link-partner-advertising: 10M-half,10M-full,100M-half,100M-full,1000M-half,1000M-full,2500M-full
                 sfp-module-present: yes
                        sfp-rx-loss: no
                       sfp-tx-fault: no
                           sfp-type: SFP/SFP+/SFP28
                 sfp-connector-type: RJ45
  sfp-link-length-copper-active-om4: 1m
                    sfp-vendor-name: MikroTik
             sfp-vendor-part-number: S+RJ10
                sfp-vendor-revision: 2.16
                  sfp-vendor-serial: F0600396D91C
             sfp-manufacturing-date: 22-02-15
                    sfp-temperature: 82C
                 sfp-supply-voltage: 3.316V
                sfp-tx-bias-current: 0mA
                    eeprom-checksum: good

/interface ethernet monitor sfp-sfpplus2 once
                      name: sfp-sfpplus2
                    status: link-ok
          auto-negotiation: done
                      rate: 10Gbps
               full-duplex: yes
           tx-flow-control: no
           rx-flow-control: no
               advertising:
  link-partner-advertising:
        sfp-module-present: yes
               sfp-rx-loss: no
              sfp-tx-fault: no
                  sfp-type: SFP/SFP+/SFP28
        sfp-connector-type: LC
       sfp-link-length-om1: 30m
       sfp-link-length-om2: 80m
       sfp-link-length-om3: 300m
           sfp-vendor-name: Arista Networks
    sfp-vendor-part-number: SFP-10G-SR
       sfp-vendor-revision: 0002
         sfp-vendor-serial: XCW1351FD05U
    sfp-manufacturing-date: 13-12-23
            sfp-wavelength: 850nm
           sfp-temperature: 58C
        sfp-supply-voltage: 3.272V
       sfp-tx-bias-current: 7mA
              sfp-tx-power: -2.861dBm
              sfp-rx-power: -2.881dBm
           eeprom-checksum: good

CRS309 port 5,6,7,8

/interface ethernet monitor sfp-sfpplus5 once
                      name: sfp-sfpplus5
                    status: link-ok
          auto-negotiation: done
                      rate: 10Gbps
               full-duplex: yes
           tx-flow-control: no
           rx-flow-control: no
               advertising:
  link-partner-advertising:
        sfp-module-present: yes
               sfp-rx-loss: no
              sfp-tx-fault: no
                  sfp-type: SFP/SFP+/SFP28
        sfp-connector-type: LC
       sfp-link-length-om1: 30m
       sfp-link-length-om2: 80m
       sfp-link-length-om3: 300m
           sfp-vendor-name: Arista Networks
    sfp-vendor-part-number: SFP-10G-SR
       sfp-vendor-revision: 0002
         sfp-vendor-serial: XCW1351FD1L5
    sfp-manufacturing-date: 13-12-27
            sfp-wavelength: 850nm
           sfp-temperature: 48C
        sfp-supply-voltage: 3.295V
       sfp-tx-bias-current: 6mA
              sfp-tx-power: -2.409dBm
              sfp-rx-power: -2.468dBm
           eeprom-checksum: good

/interface ethernet monitor sfp-sfpplus6 once
                      name: sfp-sfpplus6
                    status: link-ok
          auto-negotiation: done
                      rate: 10Gbps
               full-duplex: yes
           tx-flow-control: no
           rx-flow-control: no
               advertising:
  link-partner-advertising:
        sfp-module-present: yes
               sfp-rx-loss: no
              sfp-tx-fault: no
                  sfp-type: SFP/SFP+/SFP28
        sfp-connector-type: LC
       sfp-link-length-om1: 30m
       sfp-link-length-om2: 80m
       sfp-link-length-om3: 300m
           sfp-vendor-name: Arista Networks
    sfp-vendor-part-number: SFP-10G-SR
       sfp-vendor-revision: 0002
         sfp-vendor-serial: XCW1406FD1MR
    sfp-manufacturing-date: 14-02-06
            sfp-wavelength: 850nm
           sfp-temperature: 50C
        sfp-supply-voltage: 3.291V
       sfp-tx-bias-current: 7mA
              sfp-tx-power: -2.605dBm
              sfp-rx-power: -3.382dBm
           eeprom-checksum: good

/interface ethernet monitor sfp-sfpplus7 once
                      name: sfp-sfpplus7
                    status: link-ok
          auto-negotiation: done
                      rate: 10Gbps
               full-duplex: yes
           tx-flow-control: no
           rx-flow-control: no
               advertising:
  link-partner-advertising:
        sfp-module-present: yes
               sfp-rx-loss: no
              sfp-tx-fault: no
                  sfp-type: SFP/SFP+/SFP28
        sfp-connector-type: LC
       sfp-link-length-om1: 30m
       sfp-link-length-om2: 80m
       sfp-link-length-om3: 300m
           sfp-vendor-name: Arista Networks
    sfp-vendor-part-number: SFP-10G-SR
       sfp-vendor-revision: 0002
         sfp-vendor-serial: XCW1351FD0YA
    sfp-manufacturing-date: 13-12-23
            sfp-wavelength: 850nm
           sfp-temperature: 49C
        sfp-supply-voltage: 3.312V
       sfp-tx-bias-current: 7mA
              sfp-tx-power: -2.508dBm
              sfp-rx-power: -2.609dBm
           eeprom-checksum: good

/interface ethernet monitor sfp-sfpplus8 once
                      name: sfp-sfpplus8
                    status: link-ok
          auto-negotiation: done
                      rate: 10Gbps
               full-duplex: yes
           tx-flow-control: no
           rx-flow-control: no
               advertising:
  link-partner-advertising:
        sfp-module-present: yes
               sfp-rx-loss: no
              sfp-tx-fault: no
                  sfp-type: SFP/SFP+/SFP28
        sfp-connector-type: LC
       sfp-link-length-om1: 30m
       sfp-link-length-om2: 80m
       sfp-link-length-om3: 300m
           sfp-vendor-name: Arista Networks
    sfp-vendor-part-number: SFP-10G-SR
       sfp-vendor-revision: 0002
         sfp-vendor-serial: XCW1406FD1QX
    sfp-manufacturing-date: 14-02-06
            sfp-wavelength: 850nm
           sfp-temperature: 48C
        sfp-supply-voltage: 3.295V
       sfp-tx-bias-current: 8mA
              sfp-tx-power: -2.657dBm
              sfp-rx-power: -3.074dBm
           eeprom-checksum: good

All modules perform trouble free up to 88F ambient seen recently but I wonder what happens when S+RJ10 steps up from 2.5G to 10G.

mkx · Tue Aug 29, 2023 9:35 pm

You mention UTP cable. Does it matter whether it's shielded or not?

No, it doesn't matter (much). The problem with 10Gbps over UTP cables is that Tx amplifiers for 10Gbps are very energy ineffective, UTP cable category doesn't change this much. And the inefficiency means that excess ebergy is converted into heat, but SFP modules aren't designed to dissipate lots of heat (no heat dissipation surfaces, only contact with SFP port case which then has to be cooled this way or another).

Optical modules are much nicer in this regard, so your idea about use of CRS309 (one RJ-45 module, the rest optical modules) would probably work.

Uneasily3723 · Tue Aug 29, 2023 10:20 pm

No, it doesn't matter (much). The problem with 10Gbps over UTP cables is that Tx amplifiers for 10Gbps are very energy ineffective, UTP cable category doesn't change this much. And the inefficiency means that excess ebergy is converted into heat, but SFP modules aren't designed to dissipate lots of heat (no heat dissipation surfaces, only contact with SFP port case which then has to be cooled this way or another).

Optical modules are much nicer in this regard, so your idea about use of CRS309 (one RJ-45 module, the rest optical modules) would probably work.

So, a UTP cable would be better... and then preferrably a long one? I just use a 50cm one for bridging between the Mikrotik and the Modem.

Tue Aug 29, 2023 10:29 pm

This matter of 10GBase-T Modules will be increasing with the advent of XGS-PON into mainstream

Uneasily3723 · Tue Aug 29, 2023 10:31 pm

This matter of 10GBase-T Modules will be increasing with the advent of XGS-PON into mainstream

If only the router had a SPF+ slot for the LAN side or the MikroTik switches a 10G RJ45 slot then there would be no problems. I could have 25Gbps internet here for the same price as I have the 10G one. Only difference is the setup fee. Monthly rate is the same. But having already trouble getting 10G to work (properly), I'd even have more troubles for 25G.

Tue Aug 29, 2023 11:00 pm

......And the inefficiency means that excess ebergy is converted into heat, but SFP modules aren't designed to dissipate lots of heat (no heat dissipation surfaces, only contact with SFP port case which then has to be cooled this way or another)....

My way ... the World without zip ties will stop

cooling1.PNG

mkx · Tue Aug 29, 2023 11:22 pm

Zip ties in electrical engineering and telecommunications ... and PU foam in civil engineering

Tue Aug 29, 2023 11:24 pm

Do not forget silver tape which holds both worlds together

mkx · Tue Aug 29, 2023 11:25 pm

So, a UTP cable would be better... and then preferrably a long one? I just use a 50cm one for bridging between the Mikrotik and the Modem.

Why a long one? I'd expect that 0.5m long cat5e UTP would do just fine.

ConradPino · Tue Aug 29, 2023 11:36 pm

So, a UTP cable would be better... and then preferrably a long one? I just use a 50cm one for bridging between the Mikrotik and the Modem.

NO, not longer, length matters. Total capacitance and resistance rise with length.

UTP shielding has no effect on energy efficiency is the prior point.
High capacitance degrades frequency bandwidth.
High resistance makes energy efficiency worse.

Good things:
UTP higher category (more twists per meter) improves noise immunity.
UTP shielding improves noise immunity, valuable for very long lengths, pointless at short lengths.
At very short lengths the highest category levels are money poorly spent.

Since Category 5e & 6 are mainstream, choose a shorter length of either.

Uneasily3723 · Wed Aug 30, 2023 7:33 am

NO, not longer, length matters. Total capacitance and resistance rise with length.
[...]
Since Category 5e & 6 are mainstream, choose a shorter length of either.

Thanks for the clarification. I already have cat 7, 50cm one.

olivier2831 · Wed Aug 30, 2023 9:29 am

In the light of common practices in networking, it seems to me that jumping from 1G/s to to 10Gb/s is a too long jump as it introduces thermal or compatibility issues we didn't have to deal with in the past.

Maybe going from 1Gb/s to 2.5Gb/s would be a smarter move allowing bandwidth increase without 10Gb/s hassles.

I hope Mikrotik will soon update its product portfolio to include 2.5Gb/s where currently 1Gb/s or 10Gb/s exist.

mbovenka · Wed Aug 30, 2023 9:49 am

In the light of common practices in networking, it seems to me that jumping from 1G/s to to 10Gb/s is a too long jump as it introduces thermal or compatibility issues we didn't have to deal with in the past.

Oh yes we did. I was there for the rise of Fast & Gigabit Ethernet. Thermal issues, not so much (mostly because SFPs didn't exist yet), but compatibility issues? Up the wazoo.

woland · Wed Aug 30, 2023 10:26 am

My main switch at home is a CRS309, but I also use a CRS305 my startegy is to always try to avoid 10G RJ45 ports. First choice is a DAC cable (cheap, low power, but short runs only), second choice is using MM SFPs with fibre. (bit more expensive, but not much, less power than RJ 45 SFP+) . I only use RJ45 if I must. In this case I use only every second slot and I put some small heat sinks on the SFP+ modules themselves (don´t forget the thermal pads between the SFP and heat sink).
Also non MT 10G RJ45 Modules from China tend to stay cooler than MT modules (if they work).

For the copper cable: theoretically at least Cat6 is needed, but I have some 10G Links over Cat5e, which do work well.

Shielding for the cables is only needed for two reasons: if you have lots of cables tightly packed, you will have issues if those cables are not shielded.
If you have long runs it is better to have shielded cables for better lightning protection. (shielded cables help to lessen the potential difference between the two ends of the cable if a lightning strikes).

ConradPino · Wed Aug 30, 2023 11:14 am

I already have cat 7, 50cm one.

50 cm!

Very short, very sweet!!

mkx · Wed Aug 30, 2023 12:27 pm

If you have long runs it is better to have shielded cables for better lightning protection. (shielded cables help to lessen the potential difference between the two ends of the cable if a lightning strikes).

Not really. Shielding takes the induced voltage (so that data pairs don't get it as much). And shield has to be grounded on one end and one end only (specially with longer runs), otherwise some ground loops can happen and those show as high interference to data lines. In case of direct lightning strike, only perfect surge suppressors might save the equipment ... optical cable is indispensable in these scenarios.
If one installs longer cable runs, then it's recommended to terminate that cable in patch panel ... which should be grounded anyway (so remember to only connect shield to patch panel ground only at one end) and use short patch cords to connect equipment (on both ends).

woland · Wed Aug 30, 2023 2:04 pm

Not really!

For a direct lightning hit, forget it (also your Cat8 cable will just explode and your AP will burn nicely as well) . In case you have proper lightning protection and potential equalization in your home and the lightning did not hit your AP directly, shielding helps to _some_ extent especially if you connect both ends of the shield to your equipment.
Then if this is of concern, you should take more measures, like install more grounding (potential equalization), surge protectors.

Ground loops are mainly not of concern to ethernet (contrary to unbalanced audio, which was used for your HiFi) , as ethernet uses wire pairs, which carry the signal balanced (only the amplitude difference contains information, not the amplitude measured from ground). Again, this doesn´t mean that you should connect 2 different skyscrapers via Cat7 SFTP, you would be much better off with fiber.

Here is a nice presentation about this:
http://www.pennington.net/archives/cabl ... abling.pdf

Uneasily3723 · Wed Aug 30, 2023 2:07 pm

I always thought shielding was to minimize interference sind those copper wires are basically just long antennas.

woland · Wed Aug 30, 2023 2:12 pm

I always thought shielding was to minimize interference sind those copper wires are basically just long antennas.

The real world is mostly more complicated, than we would think at first...
Still shielding is used mainly to minimize interference.

mkx · Wed Aug 30, 2023 2:19 pm

Not really!

I wrote: might

Several on-line sources indicate that use of "might" indicates low probability of something taking place while "may" indicates high probability of something taking place.
But what do I know, English is foreign language to me ...

Re ground loops: while theoretically ground loop (i.e. unwarranted current flowing through cable shield) should not affect quality of signal, carried differentially by a pair of wires, in reality it does happen. So the best thing to do is to connect shield (to a good ground) only on one end. The only place where it's relatively safe to ground shield on both ends is when used between two points with same ground potential (in practice that means that both are connected to same grounding system and that resistance between the two points over grounding system is next to zero). Which often means inside same building where using shielded cables is almost not necessary.
Another bad side effect of ground loops is corrosion of grounding points (due to constant low current).
(I'm not saying this is always, connecting cable shield in more points when it runs down from metallic tower is a pretty good idea, specially as high metallic towers tend to attract lightning strikes).

mkx · Wed Aug 30, 2023 2:44 pm

I always thought shielding was to minimize interference sind those copper wires are basically just long antennas.

They key in UTP is the "TP" part ... twisted pair. The idea is that if there are two wires running close one to another, external EM will induce similar amount of interference in both wires. However, if they run in parallel all the time, one wire might be systematically closer to source of interference and would thus receive slightly higher amount of interference ... receiver, connected to wire end and measuring voltage difference between these two wires, will still see some noise. The idea behind twisting the pair of wires into same bundle is that none of wires will be systematically closer to source of interference (due to twisting they will alter between being closer and being farther) and ideally the same receiver will measure noise at negligible level.
UTP cables with higher category ratings have more twits per unit of distance making those pairs less susceptible to interference even if interference is close to the pair (e.g. power cable crossing path of the cable).

Shield in S(F)TP cables certainly reduces interference because it acts as a Faraday cage ... and it even doesn't have to be grounded to suppress the interference. Grounding is important for other reasons ... however, if it's not grounded, it might induce interference (or some other nasty electrical effects) at the end of the shield (usually that's right before RJ45 connector).

woland · Wed Aug 30, 2023 2:58 pm

I totally agree in one point: using heavily shielded cable makes your life hard. I did not realize it for quite long, but it´s mostly unnecessary at home and you can pull and terminate unshielded cable so much more easily.

For corrosion caused by currents: yes that´s a common problem. Epecially for below earth stray currents caused by trams running on 600V DC and sipping around 100Amps.

Wed Aug 30, 2023 3:13 pm

In the light of common practices in networking, it seems to me that jumping from 1G/s to to 10Gb/s is a too long jump as it introduces thermal or compatibility issues we didn't have to deal with in the past.

Maybe going from 1Gb/s to 2.5Gb/s would be a smarter move allowing bandwidth increase without 10Gb/s hassles.

i agree

Took almost a decade for 10G Base-T to become mainstream, even so the power consumption is still high, and expensive
so much so that it turned out to be a good idea to use 10g technology at lower frequency to give birth to 2.5Gbit 5.0Gbit speeds, those can be very much more feasible in most conditions, but still expensive

Now the problem is XGS-PON making 10g speeds (and beyond) a reality at mainstream Home so suddenly we need 10g at the home to stay at pace with these speeds

I think 10G Base-T is far from practical, i think a real solution will be some sort of single mode optical fiber implementation friendly for end user manipulation

5nik · Thu Aug 31, 2023 12:45 am

I hope Mikrotik will soon update its product portfolio to include 2.5Gb/s where currently 1Gb/s or 10Gb/s exist.

It is here: CRS310-8G+2S+IN

Thu Aug 31, 2023 1:44 am

It is here: CRS310-8G+2S+IN

Beautiful !!!

onnoossendrijver · Thu Aug 31, 2023 10:52 am

Shut up and take my money!!

olivier2831 · Thu Aug 31, 2023 3:08 pm

It is here: CRS310-8G+2S+IN

At last !

leopardus2 · Thu Nov 09, 2023 5:50 pm

The S+RJ10 I have fitted in my CRS309, when connected at 10G speed, runs around 90 degrees Celsius, this is just insane.
From time to time it reaches 95C and the OS shuts it down. Another S+RJ10 in same switch running at 2.5G stays around 70 degrees.

Now, I have read a lot about the topic on this forum.

The question I have is: why other 10G SFPs from other vendors run way cooler??
I replaced the S+RJ10 with a 10Gtek SFP and it runs way cooler at 10G speed, around 65 degrees...

Rick

Uneasily3723 · Thu Nov 09, 2023 5:59 pm

I replaced the S+RJ10 with a 10Gtek SFP and it runs way cooler at 10G speed, around 65 degrees..

Got a link?

leopardus2 · Thu Nov 09, 2023 6:03 pm

I replaced the S+RJ10 with a 10Gtek SFP and it runs way cooler at 10G speed, around 65 degrees..
Got a link?

I got it from Amazon. Search for "10gtek asf-10g-t".

Uneasily3723 · Thu Nov 09, 2023 6:07 pm

I got it from Amazon. Search for "10gtek asf-10g-t".

Thx

Uneasily3723 · Thu Nov 16, 2023 11:51 am

I got it from Amazon. Search for "10gtek asf-10g-t".

Got it now since monday and it's been running without issue. Although the outside temperatures have come down also. However in my CRS305 it doesn't show the current temperature of the module. Is it the same for you?

[...]
Vendor Name		Intel Corp
Vendor Part Number		FTLX8571D3BCV-IT
Vendor Revision		A
Vendor Serial		IN101N80571
Manufacturing Date		23-08-07
Wavelength		850.00 nm
Temperature		 
Supply Voltage		 
Tx Bias Current

leopardus2 · Thu Nov 16, 2023 12:07 pm

It looks like you got a different model than mine.
You sure you bought the 10Gtek?

          sfp-vendor-name: OEM
    sfp-vendor-part-number: SFP-10G-SR
       sfp-vendor-revision: 02
         sfp-vendor-serial: CSF101N25484
    sfp-manufacturing-date: 23-02-14
            sfp-wavelength: 850nm
           sfp-temperature: 54C
        sfp-supply-voltage: 3.221V
       sfp-tx-bias-current: 6mA
              sfp-tx-power: -3.001dBm
              sfp-rx-power: -3.001dBm

leopardus2 · Thu Nov 16, 2023 12:10 pm

An additional, unrelated but amusing, piece of information.
My Imac PRO has a 10Gbit ethernet interface, which suffers same problem -- when running at 10G, it goes over 90 degrees C and causes
the computer to termal throttle!!!! Had to buy a separate thunderbolt to 10G adapter....

Uneasily3723 · Thu Nov 16, 2023 12:27 pm

It looks like you got a different model than mine.
You sure you bought the 10Gtek?

          sfp-vendor-name: OEM
    sfp-vendor-part-number: SFP-10G-SR
       sfp-vendor-revision: 02
         sfp-vendor-serial: CSF101N25484
    sfp-manufacturing-date: 23-02-14
            sfp-wavelength: 850nm
           sfp-temperature: 54C
        sfp-supply-voltage: 3.221V
       sfp-tx-bias-current: 6mA
              sfp-tx-power: -3.001dBm
              sfp-rx-power: -3.001dBm

Yes, I got it in a black / yellow metal case on the back:

Mfg: Shenzhen 10Gtek Tranceivers Co., Ltd.
Web: www.10gtek.com
Mail: ....
[E010140001].

leopardus2 · Thu Nov 16, 2023 1:27 pm

I got it from Amazon. Search for "10gtek asf-10g-t".
Got it now since monday and it's been running without issue. Although the outside temperatures have come down also. However in my CRS305 it doesn't show the current temperature of the module. Is it the same for you?
Code: Select all
[...]
Vendor Name		Intel Corp
Vendor Part Number		FTLX8571D3BCV-IT
Vendor Revision		A
Vendor Serial		IN101N80571
Manufacturing Date		23-08-07
Wavelength		850.00 nm
Temperature		 
Supply Voltage		 
Tx Bias Current		 

You using RouterOS 6? Your output is different than mine.
Also, funny that your 10Gtek product is actually an Intel. I have several 10Gtek SFPs and they all show up as "OEM" and similar output to the one I posted.

Uneasily3723 · Thu Nov 16, 2023 1:47 pm

Enabled		
Name		routeros
Version		7.10.2
Build Time		2023-07-12 09:45:11
Scheduled

The other two SPF+ that are inserted do show the temperature.

And here's the one I ordered

https://www.amazon.de/dp/B071RSC54C?psc ... ct_details

Just noticed that the title says "for Intel".... according to manufacture date it's only 3 months old

Manufacturing Date 23-08-07

Maybe you have an older one and they changed their hardware inbetween? Did you also get a black metal box with yellow writing as case for the sfp+?

leopardus2 · Thu Nov 16, 2023 1:58 pm

Mine is not old... sfp-manufacturing-date: 23-02-14
You have a different chip it seems. But if it works for you without thermal shutdown, then fine

Uneasily3723 · Thu Nov 16, 2023 2:01 pm

Mine is not old... sfp-manufacturing-date: 23-02-14
You have a different chip it seems. But if it works for you without thermal shutdown, then fine

Yours is still half a year older. Could be that they changed hardware inbetween. It works fine so far. Have to wait for summer to see how it behaves then because the apartment will be hotter then. Probably still fine. Too bad, that I can't see the temperature though.

leopardus2 · Thu Nov 16, 2023 2:02 pm

I don't think they have changed hardware. Yours must be a special version, like you said.
Why is your output different than mine? I am running latest RouterOS 7. Maybe you could try upgrading.

Uneasily3723 · Thu Nov 16, 2023 2:19 pm

I don't think they have changed hardware. Yours must be a special version, like you said.
Why is your output different than mine? I am running latest RouterOS 7. Maybe you could try upgrading.

I'm running Version 7.10.2

leopardus2 · Thu Nov 16, 2023 2:32 pm

that’s old.

Uneasily3723 · Thu Nov 16, 2023 4:34 pm

that’s old.

Running now 7.12 - still looks the same.

leopardus2 · Thu Nov 16, 2023 4:42 pm

Likely , the SFP is not reporting the info, although I find it strange.
I assume the S+RJ10 does if you plug it in?
You can also try and get another 10Gtek SFP from Amazon, that has not the "for intel" note, and return your current one.

Uneasily3723 · Thu Nov 16, 2023 4:45 pm

Likely , the SFP is not reporting the info, although I find it strange.
I assume the S+RJ10 does if you plug it in?
You can also try and get another 10Gtek SFP from Amazon, that has not the "for intel" note, and return your current one.

yeah, the S+RJ10 did report temperature - hence it got deactivated. As long is the current works fine, I have no need to replace ist.

leopardus2 · Thu Nov 16, 2023 4:47 pm

BTW I see you are pasting info from the webfig interface, that's why your output is different from mine (I am using CLI).
Amazon has free returns... I would try and get another SFP from Amazon and see if you get the same version I have...

Uneasily3723 · Thu Nov 16, 2023 5:47 pm

BTW I see you are pasting info from the webfig interface, that's why your output is different from mine (I am using CLI).
Amazon has free returns... I would try and get another SFP from Amazon and see if you get the same version I have...

Why returning if it's working?

leopardus2 · Thu Nov 16, 2023 8:51 pm

BTW I see you are pasting info from the webfig interface, that's why your output is different from mine (I am using CLI).
Amazon has free returns... I would try and get another SFP from Amazon and see if you get the same version I have...
Why returning if it's working?

To get the model which reports temperature correctly

kehrlein · Sat Jan 13, 2024 8:48 pm

For those interested in adding fans to the CRS309, take a look at this video for some inspiration: https://www.youtube.com/watch?v=ZDTv6NGBhYk
Of course there are more elegant ways, but this guys' solution is more efficient than just using heat sinks.

leopardus2 · Mon Jan 15, 2024 12:37 pm

Why buying a fanless unit if then you add a fan to it?
Just get one of the ones that have built in fan.
My reason to go fanless is desktop use. Otherwise I would have chosen a different model.