13

I have a few switches that are in the same rack that must be connected via fiber OM2, 50/125 MM, LC/LC (2 additional examples added for reference are 10GbE and are OM3).

A Verizon tech told me that he had to add attenuators because the laser was too strong and would burn out the optics.

I've never heard of it, are there any best practices on these db levels?

Output on the Juniper

>show interfaces diagnostics optics ge-0/2/1
Physical interface: ge-0/2/1
    Laser bias current                        :  5.284 mA
    Laser output power                        :  0.3120 mW / -5.06 dBm
    Module temperature                        :  35 degrees C / 95 degrees F
    Module voltage                            :  3.2670 V
    Receiver signal average optical power     :  0.2986 mW / -5.25 dBm
    Laser bias current high alarm threshold   :  13.000 mA
    Laser bias current low alarm threshold    :  1.000 mA
    Laser bias current high warning threshold :  12.500 mA
    Laser bias current low warning threshold  :  2.000 mA
    Laser output power high alarm threshold   :  1.0000 mW / 0.00 dBm
    Laser output power low alarm threshold    :  0.0440 mW / -13.57 dBm
    Laser output power high warning threshold :  0.5010 mW / -3.00 dBm
    Laser output power low warning threshold  :  0.1120 mW / -9.51 dBm
    Module temperature high alarm threshold   :  110 degrees C / 230 degrees F
    Module temperature low alarm threshold    :  -40 degrees C / -40 degrees F
    Module temperature high warning threshold :  93 degrees C / 199 degrees F
    Module temperature low warning threshold  :  -30 degrees C / -22 degrees F
    Module voltage high alarm threshold       :  3.600 V
    Module voltage low alarm threshold        :  3.000 V
    Module voltage high warning threshold     :  3.500 V
    Module voltage low warning threshold      :  3.100 V
    Laser rx power high alarm threshold       :  1.1220 mW / 0.50 dBm
    Laser rx power low alarm threshold        :  0.0079 mW / -21.02 dBm
    Laser rx power high warning threshold     :  0.7943 mW / -1.00 dBm
    Laser rx power low warning threshold      :  0.0200 mW / -16.99 dBm

Output on my Core Switch

CoreSwitch#sh interfaces tengigabitethernet 0/46
TenGigabitEthernet 0/46 is up, line protocol is up
Port is part of Port-channel 127
Description: Juniper
Hardware is DellEth, address is 00:00:00:00:00:00
    Current address is 00:00:00:00:00:00
Pluggable media present, SFP type is 1000BASE-SX
    Wavelength is 850nm
    SFP receive power reading is -5.8704dBm

CoreSwitch#sh int te0/7
TenGigabitEthernet 0/7 is up, line protocol is up
Port is part of Port-channel 7
Description: Access Switch Stack 1
Hardware is DellEth, address is 00:00:00:00:00:00
    Current address is 00:00:00:00:00:00
Pluggable media present, SFP+ type is 10GBASE-SR
    Medium is MultiRate, Wavelength is 850nm
    SFP+ receive power reading is -8.9177dBm

CoreSwitch#show int te0/6
TenGigabitEthernet 0/6 is up, line protocol is up
Port is part of Port-channel 6
Description: Access Switch Stack 2
Hardware is DellEth, address is 00:00:00:00:00:00
    Current address is 00:00:00:00:00:00
Non-qualified pluggable media present, SFP+ type is 10GBASE-SR
    Medium rate is unknown, Wavelength is 850nm
SFP+ receive power reading is -3.0356dBm

Is -5 dBm acceptable? The default 'ALARMs' are between -1 and -16, but should I aim for -10 dBm and spend the money on 5db attenuators?

9
  • 1
    My question would be: why are you using laser optics for an inter-rack run? But yes, most optics will adjust their Tx power as needed, but in some cases even the lowest power will overdrive the Rx. If this is the case, an attenuators is needed.
    – EEAA
    Oct 23 '15 at 8:17
  • 1
    Because I have to, its not a choice I have. Oct 23 '15 at 11:27
  • 3
    I think he was bullshitting, almost all of the time "OM2, 50/125 MM, LC/LC" is used with LED-based optics not laser-based ones, which don't ablate - many people just assume they're lasers when often they're not - generally if you can see the signal (in red or whatever) it's LED-driven not laser-driven. Unfortunately LED's can't signal faster than about 7.5Ghz so any signal at 8Gbps or greater HAS to use lasers and so we do need to use '100% glass' based cables as the plastic ones often used with LED optics will spoil quickly.
    – Chopper3
    Oct 23 '15 at 12:29
  • 1
    @EEAA I did some work in a data center where the architect was ESD/Lightning-phobic. The SAN, network gear, and servers were all electrically isolated, with fiber between everything.
    – Chris S
    Oct 23 '15 at 13:44
  • 2
    @JacobEvans Check the model number of the SFP and then look up the datasheet for it. DO NOT look into the SFP while it's powered up. :)
    – EEAA
    Oct 23 '15 at 13:51
7

I found my answer from a link in the network engineering group,

Damage by Overpowered Transmitters?

  • Well, yes and no.
  • Actually, most optics transmit at roughly the same power.
    • The typical outputs of 10km and 80km optics are within 3dB.
  • Long reach optics achieve their distances by having more sensitive receivers, not by having stronger transmitters.
    • 80km optics may have a 10dB+ more sensitive receiver than 10km
    • These sensitive receivers are what are in danger of burning out.
  • There are two thresholds you need to be concerned with.
    • Saturation point (where the receiver is “blinded”, and takes errors).
    • Damage point (where the receiver is actually damaged).
    • The actual values depend on the specific optic.
    • But generally speaking, only 80km+ optics are at risk.

enter image description here

Source Page 77-78 from Everything You Always Wanted to Know About Optical Networking – But Were Afraid to Ask

1
  • How to read this diagram? I am confused.
    – alsadk
    Oct 11 '18 at 7:03
5

It depends on the fiber modules you are using. Use a module matched to your run length and you'll be fine. For an in-rack run, SX or SR modules will be fine without attenuators. If you use long distance modules (LH, LX, EX, ZX; LR, LRM, etc) with such a short run you could cause problems. Each module will have a minimum output, which could be above the safe floor (see your modules' specs for the max Receive Power, which is commonly -1 or higher).

I've never seen attenuators used on multi-mode cable. They seem quite common for ISPs that deliver fiber Internet with single-mode, where the customer is fairly close to the ISP's headend. This is probably why the Verizon tech thought you would need them in your rack, not understanding the differences between single and multi-mode cable, and short/long range fiber modules.

1
  • Thank you for your answer, it's clearly a difference in MM vs SM, I'm still trying to establish what is 'good operating range' for measured dBm Oct 23 '15 at 13:28
2

Jacob Evans Had a very good answer on here. One thing I would clarify though is that the transmitter on an LR is completely different than an ER or ZR.

Multimode transceivers usually use Vcsel or similar as it's light source, these are harmless to receivers and will never burn them out.

LR's use DFB or FB lasers which are not very powerful and also will not cause permanent harm to a receiver.

Now ER's and ZRs use an EML or similar laser (we use EML in ours) These are strong enough to cause permanent damage and overheating in close ranges without an attenuator.

We have gotten returns because people use ER or ZR optics in way too close of a range without an attenuator. Here is a good rule of thumb I pass along to people. Please Note, these are purely for the safety of the hardware. You may need to make adjustments based on your network setup.

10km Optic - No Attenuation needed

40km Optic - -4db attenuator at 20km, -8db at 10km

80km Optic - -10db attenuator at 40km, 15db at 20km, not recommended much lower than 20km distance.

0
0

-5dBm is perfectly acceptable, based on what you have posted. Generally, a "good operating range" in dBm is exactly "what is the receive spec for the SFP/SFP+" in dBm, though if possible being at least 3-6 dBm above the lowest listed rx sensitivity is good to limit issues, but there is no problem being right up against the maximum (ie, -2 dBm to -16 dBm looks fine to me for the specs listed, and given the low rx alarm is all the way down at -21 you have considerable built-in threshold between "warning" and alarm (where you might expect it to stop working.)

In general, you need detailed specs for the power limits, which you seem to have (or have posted) for only one device (presumably your juniper accesses more of the internal info, but you can always look up the manufacturer specs): Cutting down to the relevant numbers....

Laser output power high alarm threshold   :  1.0000 mW / 0.00 dBm

Laser output power high warning threshold :  0.5010 mW / -3.00 dBm

Laser rx power high alarm threshold       :  1.1220 mW / 0.50 dBm

Laser rx power high warning threshold     :  0.7943 mW / -1.00 dBm

You can see that for this module, the rx warning and alarm thresholds are ABOVE the tx (output) warning and alarm thresholds. You could connect two of these directly with a half-meter cable (or whatever the shortest thing you can get between them is) and ALWAYS be fine.

The LX single-mode SFPs I use are similar, in that the receive power threshold is above the maximum output power range, so they work just fine with a short fiber, or up to 4 kilometers. As such, it's not generically a multi-mode/single-mode issue; it just depends on device specs.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.