dslreports logo
 
    All Forums Hot Topics Gallery
spc
Search similar:


uniqs
5620
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

1 edit

lutful to switchman

Premium Member

to switchman

Re: SFP module for 100km

Finisar is an excellent brand.

I also agree that OP needs to get official specs for the chosen 160km SFP to make sure (minimum TX power - RX sensitivity) comes out a few dB higher than the 30-32dB loss on this fiber.

OK, I was looking at marketing spec for Transition TN-SFP-LX modules.

TN-SFP-LX5 [50 km/31.1 mi.] Link Budget: 19.0 dB
TN-SFP-LX8 [80 km/49.7 mi.] Link Budget: 24.0 dB
TN-SFP-LX12 [120 km/74.6 mi.] Link Budget: 32.0 dB
TN-SFP-LX16 [160 km/99.4 mi.] Link Budget: 37.0 dB
TN-SFP-LX20 [200 km/124.3 mi.] Link Budget: 41.0 dB

My hunch is that the RX sensitivity is the same for all modules and they are only increasing TX power for the increased range and link budget guarantee. If RX sensitivity was -30dBm across the modules, minimum TX power would have to be +7dBm for LX16 and +11dBm for LX20 ... which boggle my mind. Definitely worth asking for their complete technical specs.
switchman
join:1999-11-06

switchman to spectrumhead

Member

to spectrumhead
Agreed, the lowest I have ever seen is -32dBm. But I don't usually deal with anything other than LX at the Ge rate. Ge rate is a client port on my designs so I don't need much reach.
spectrumhead
join:2009-05-03

spectrumhead

Member

So I received TN-SFP-LX16 yesterday and plugged it into only the one end of the fiber (other end still HP SFP mentioned in my original post) and bingo ! I see a link on Transition Netowrk SFP side.

1 side is HP other is Transition Network SFP.

Dont really understand how this is happening though. Any ideas ? Better RX sensitivity for TN-LX-16 maybe ?
raytaylor
join:2009-07-28

raytaylor to spectrumhead

Member

to spectrumhead
Sorry to interrupt - can someone tell me what an OTDR is and how you can use it to find bad parts of the fibre?
From a couple of posts here it sounds like it can tell you how far down a cable a break or bad joint is?

John Galt6
Forward, March
Premium Member
join:2004-09-30
Happy Camp

John Galt6

Premium Member

Optical Time Domain Reflectometer...it sends a light pulse down the fiber and times how long it takes for the pulse to return.

You're correct in your presumption of what it is used for...
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

lutful to raytaylor

Premium Member

to raytaylor
said by raytaylor:

OTDR ... sounds like it can tell you how far down a cable a break or bad joint is?

Yes, even the most basic OTDR can tell you where a break or bad splice is located. Just like radar which uses a short microwave pulse, OTDRs send a short laser pulse and measure the reflection.

My master's thesis from way back in 1992 has both theory and application of a more sophisticated "photon counting" OTDR which can tell you the quality of each length of fiber, each splice and each connector along the whole link. If something was stressing the fiber, you will also detect the tiny change in attenuation.

»era.library.ualberta.ca/ ··· de11/DS1
lutful

lutful to spectrumhead

Premium Member

to spectrumhead
said by spectrumhead:

Dont really understand how this is happening though. Any ideas ? Better RX sensitivity for TN-LX-16 maybe ?

Probably they are almost equal in RX sensitivity. But the TN-LX-16 definitely transmits at much higher power level. See my post with TX power estimates for all their LX models.
spectrumhead
join:2009-05-03

spectrumhead

Member

But I see a link on TN-LX 16 side.
switchman
join:1999-11-06
ARRIS SB6183
(Software) OPNsense

switchman to lutful

Member

to lutful
said by lutful:

Probably they are almost equal in RX sensitivity. But the TN-LX-16 definitely transmits at much higher power level. See my post with TX power estimates for all their LX models.

And either the SFP you removed was bad which is doubtful or the one still installed is running at the high end of its power range. A cheap way to get an estimate what you remaining budget is is to use optical attenuators to pad the rx level down until it will not establish the link. This is assuming your system will not report the RX power level or that you do not have the correct test equipment.
54067323 (banned)
join:2012-09-25
Tuscaloosa, AL

54067323 (banned) to spectrumhead

Member

to spectrumhead
said by spectrumhead:

Dont really understand how this is happening though. Any ideas ? Better RX sensitivity for TN-LX-16 maybe ?

Is the HP side still not linking?
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

lutful to spectrumhead

Premium Member

to spectrumhead
said by spectrumhead:

But I see a link on TN-LX 16 side.

Seems you have that particular HP model.

7dBm TX power from TN side - 32dB path loss = -25dBm RSSI at the HP which is below its RX sensitivity of -23dBm.

5dBm TX power from HP side - 32dB path loss = -27dBm at the TN which is above its RX sensitivity of -30dBm (my guess).
switchman
join:1999-11-06
ARRIS SB6183
(Software) OPNsense

switchman to spectrumhead

Member

to spectrumhead
Another thought is that when they gave you the loss values, did they give you the loss for both fibers? It is possible that they gave you worst case of the two and that the fiber loss is a lot worse on one fiber. A simple test is to roll the fiber, swap the TX and RX at both locations and see if the link comes up both ways.

But to be honest, I am a lot more confident that it is what lutful is showing. Plus there is probably a 1db of connector loss (.5db per connector) that is not in his numbers. I think you are going to have to purchase another TN-LX 16 to get the link working.
54067323 (banned)
join:2012-09-25
Tuscaloosa, AL

54067323 (banned)

Member

said by switchman:

A simple test is to roll the fiber, swap the TX and RX at both locations and see if the link comes up both ways.

That is why I asked if the HP linked up as well, could very well be one of the fibers is long optically, frogging them would prove that out.
spectrumhead
join:2009-05-03

spectrumhead

Member

changed both sides to LX-16 and it works.

Thank you very much guys, this forum still rocks
spectrumhead

spectrumhead

Member

Now I know this might sound a bit stupid but since I had this 1gbit running, I have already filled it.

I may want 10G out of this line.

@TomS
You mentioned about fiber preamps. Could you give me a bit more info on how these are used and which brands models we need to consider ?

Is it possible to have 10G connectivity out of this fiber line ?

lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

lutful

Premium Member

said by spectrumhead:

Is it possible to have 10G connectivity out of this fiber line?

»www.amazon.com/optical-t ··· 06CCO22C

Yes, theoretically, via 10G transceiver modules ... but this fiber simply does not have sufficient margin.

Amplifiers may allow nominal 80km range 10G modules to work on that fiber by bumping up TX but amplifiers always increase noise which work against the RX side of the much more complex 10G receiver. Anyway, you should consult with a Finisar technical guru.

Rhaas
Premium Member
join:2005-12-19
Bernie, MO

Rhaas to spectrumhead

Premium Member

to spectrumhead
Look into CWDM. Just shove multiple wavelengths on the fiber. You will have to aggregate them somehow.

Inssomniak
The Glitch
Premium Member
join:2005-04-06
Cayuga, ON

Inssomniak to lutful

Premium Member

to lutful
Heh 2,700 smackers for that and usually bought in pairs. doesn't the cage you stick that thing into also have to support 10gige?
Inssomniak

Inssomniak to Rhaas

Premium Member

to Rhaas
said by Rhaas:

Look into CWDM. Just shove multiple wavelengths on the fiber. You will have to aggregate them somehow.

My upstream delivers me fibre via CWDM. So 1550 one way and 1310 the other down a single strand. Then a media converter to 100bt Ethernet. Not that I've looked hard but I was gonna try to remove the media converter and find a gbic that did it so I can inject the fibre right into my switch and have one less thing to power and or break.
switchman
join:1999-11-06
ARRIS SB6183
(Software) OPNsense

2 edits

switchman to spectrumhead

Member

to spectrumhead
CWDM XFPs/SFPs do not have the reach needed by the OP nor can the be amplified. OP is going to need to use a DWDM XFP with an amp on the TX side most likely. If used on the RX side, the RX signal will probably be too low for the amp to detect so it will shut down.

By the way does your switch have an XFP port or only SFP?

Rhaas
Premium Member
join:2005-12-19
Bernie, MO

Rhaas to Inssomniak

Premium Member

to Inssomniak
said by Inssomniak:

said by Rhaas:

Look into CWDM. Just shove multiple wavelengths on the fiber. You will have to aggregate them somehow.

My upstream delivers me fibre via CWDM. So 1550 one way and 1310 the other down a single strand. Then a media converter to 100bt Ethernet. Not that I've looked hard but I was gonna try to remove the media converter and find a gbic that did it so I can inject the fibre right into my switch and have one less thing to power and or break.

That is a standard BiDi (Bi-Directional) Optic.
Just have to match the TX and RX pairs (IE if they are TX on 1550 then your optic is RX on 1550)
»www.prosfp.com/sfp-trans ··· bidi-sfp
And of course make sure it works with your switch.
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

lutful to Inssomniak

Premium Member

to Inssomniak
said by Inssomniak:

Heh 2,700 smackers for that and usually bought in pairs. doesn't the cage you stick that thing into also have to support 10gige?

Setup using 10G long haul transceivers will actually be "low cost" compared to any viable WDM alternative.

Rhaas
Premium Member
join:2005-12-19
Bernie, MO

Rhaas to switchman

Premium Member

to switchman
I disagree, On one of our projects we were looking at a CWDM sfp with 41 or 42 db of operating budget. As far as the amp, that I have no idea. Never had a need for one so I've never checked in on it.

On the CWDM, even with a 3db insertion loss for the mux/demux he is above the total link loss.

* The prefered choice would be to just shoot a 10g link IMO. I'll have to defer 10g questions as I have not messed with it and have no first hand experience.
switchman
join:1999-11-06
ARRIS SB6183
(Software) OPNsense

switchman

Member

said by Rhaas:

I disagree, On one of our projects we were looking at a CWDM sfp with 41 or 42 db of operating budget.

I should have limited my response to 10G XFP.
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

lutful to Rhaas

Premium Member

to Rhaas
said by Rhaas:

I disagree, On one of our projects we were looking at a CWDM sfp with 41 or 42 db of operating budget.

I am curious what brand/model? OP was originally asking for 1G SFP with (32dB+a few dB) budget and obviously that 42dB capable SFP would have been his first choice if you mentioned it earlier and the price was reasonable.

FYI, the TN-SFP-LX16-Cxx CWDM versions from Transition Network have exactly the same 37dB budget as the TN-SFP-LX16 he purchased. The CWDM muxes and extra 2 splices will take away some of the margin which could make the link marginal.

That was my concern.
lutful

lutful to spectrumhead

Premium Member

to spectrumhead
said by spectrumhead:

You mentioned about fiber preamps. Could you give me a bit more info on how these are used and which brands models we need to consider ?

Ultraspan power booster from Finisar (for example) will add 6dB to the TX dBm level ... but it has "noise figure" spec of 8dB.

»www.finisar.com/products ··· -Booster

dmburgess
join:2006-09-12
House Springs, MO

dmburgess to spectrumhead

Member

to spectrumhead
can recommend »www.wlan1.com/product_p/ ··· s120.htm
switchman
join:1999-11-06
ARRIS SB6183
(Software) OPNsense

2 edits

switchman to lutful

Member

to lutful
said by lutful:

Ultraspan power booster from Finisar (for example) will add 6dB to the TX dBm level ... but it has "noise figure" spec of 8dB.

»www.finisar.com/products ··· -Booster

This was the amp I was thinking of. Its gain rage is 6-20db with up to 26dBm launch power. The OP will need a DWDM optic as the input range is 1528-1567nm.

I would no worry about the noise figure as the 8db value is based on max gain at min input power. If it is being used as a booster, then this will not be an issue. If it is being used as a Rx amp then you probably have an issue.

Note this is a C-Band edfa. I believe this is a Red-C edfa. Red-C makes good amps.

Also dispersion is now a potential issue. At 92km, you are looking at ~1500ps of dispersion assuming the fiber is SMF-28. Depending on the optic chosen, he may need to add DCMs. The OP needs to get more details on the fiber type to calculate the actual dispersion.

Rhaas
Premium Member
join:2005-12-19
Bernie, MO

Rhaas to lutful

Premium Member

to lutful
said by lutful:

I am curious what brand/model? OP was originally asking for 1G SFP with (32dB+a few dB) budget and obviously that 42dB capable SFP would have been his first choice if you mentioned it earlier and the price was reasonable.

I can't recall the part number on it, only that it was a Tiawan manufacturer.
It was part of a coop project that never materialized, so the shredder ate all those documents.

It looks like I will be turning up a 93km route. Hoping to start getting the fiber in the ground soon.

warwick
join:2009-06-05
Hollywood, FL

warwick

Member

said by Rhaas:

said by lutful:

I am curious what brand/model? OP was originally asking for 1G SFP with (32dB+a few dB) budget and obviously that 42dB capable SFP would have been his first choice if you mentioned it earlier and the price was reasonable.

I can't recall the part number on it, only that it was a Tiawan manufacturer.
It was part of a coop project that never materialized, so the shredder ate all those documents.

It looks like I will be turning up a 93km route. Hoping to start getting the fiber in the ground soon.

I'll take the leftover fiber if you have any . But in all seriousness, I've been trying to find about 1.2km (OSP) on the cheap. but no real luck.