dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
316
share rss forum feed


TomS_
Git-r-done
Premium,MVM
join:2002-07-19
London, UK
kudos:5

2 edits
reply to raytaylor

Re: Fibre / GPON forums

Splitters usually live in a pedestal/cabinet somewhere around the estate. The splitter will take an incoming signal and break it up in 32 or 64 ways or some other number. These are then patched on to point to point fibres that are presented on little breakout boxes in a pit or on a pole somewhere along the street, and then a lead in cable run to the ONT.

PON splitter: »www.bitsonlight.com/ftth/images/···inet.JPG
"Breakout box" (with the angled connectors on it): »www.bitsonlight.com/ftth/images/···npit.JPG

(Actually the page I took these images from has some good info on it: »www.bitsonlight.com/ftth/)

"Breakout box" is probably not the right term, but its the best thing I can think of to describe it right now. The page above suggests "lead in module" but Im not experienced enough with PON to know if thats an industry standard term, or just what the operator of that PON network is calling it.

ONTs can usually be programmed to, e.g., turn on or off a service if there is no subscription. ONTs with multiple data ports can have their individual data ports provisioned for individual services, if you're using GEPON (which is ethernet based) then you can put each one in its own VLAN (IIRC) which would mean you could do something like what you were asking (Internet and IPTV services.)

Rate limiting might be achieved at a BRAS if you're using PPPoE for example. I guess it depends on the ONT and how smart it is, usually they are nothing more than a media converter with some basic smarts for maintenance and fault finding.

And of course it all depends on the type of PON you are using, and the vendor of your PON equipment. You'd have to contact some of them to get your specific questions answered, unless someone here has already done the work and can share their experience.

FWIW the network currently being rolled out in Australia is using Alcatel-Lucent PON gear. Huawei also manufacture PON gear, and its probably cheaper.


raytaylor

join:2009-07-28
kudos:1

said by TomS_:

Splitters usually live in a pedestal/cabinet somewhere around the estate. The splitter will take an incoming signal and break it up in 32 or 64 ways or some other number. These are then patched on to point to point fibres that are presented on little breakout boxes in a pit or on a pole somewhere along the street, and then a lead in cable run to the ONT.

I couldnt see a 32-way splitter on sun's website, all i could find are Y splitters/couplers

Clicking the splitters menu links on their website takes me to this page which they describe as couplers.
»www.suntelecommunication.cn/Spli···ers.html

so am trying to figure out if i can split it as the trunk fiber goes down the street by splitting it at each pedestal, or would i have to split it at a central location such as a cabinet you describe above using something i need to buy from another company?

they seem to sell everything else except a 32 way splitter.
Am i missing something?


TomS_
Git-r-done
Premium,MVM
join:2002-07-19
London, UK
kudos:5

IMO:

The problem with "split as you go" is that each time you split, the signal becomes weaker and weaker for the ONTs further along the line.

With a centralised splitter they are all equal, minus the attenuation of the fibre between the splitter and the ONT and any mechanical losses.

You also need to consider whether the ducting between pits is big enough to fit the bulky "breakout boxes" (or maybe its a tap?) through (if you are going underground.) Most times the taps comes with a length of fibre attached which you can order to suit, and you just pull those tails through back to your splitter cabinet and terminate them.

Its also a maintenance nightmare. If something becomes faulty somewhere in the middle you'll have to replace the whole thing, rather than just the one affected piece. And having a replacement manufactured to suit would take some time, whereas you can stock individual taps with various length tails for rapid response.

IMO you would be better off with a centralised splitter simply because its more modular.

And dont forget, the further your customer are from your head end, the lower number of ways you will want to split the signal. e.g. if the headend was right next to the splitter you might be able to split 64 ways, but if you have several km of fibre in between you might only want to split 16 or 32 ways to preserve signal strength.

You might also consider a lower split count for performance reasons. The more times you split, the more customers will be sharing the same bandwidth.


jcremin

join:2009-12-22
Siren, WI
kudos:2

said by TomS_:

The problem with "split as you go" is that each time you split, the signal becomes weaker and weaker for the ONTs further along the line.

With a centralised splitter they are all equal, minus the attenuation of the fibre between the splitter and the ONT and any mechanical losses.

There are a few architectures you can choose from when designing the layout of your PON which don't necessarily cause uneven attenuation:

1) Centralized split: One location where all the splits take place, with a dedicated fiber feeding each customer. The only difference in attenuation would be the length of the fiber feeding the customer.

2) distributed split: Rather than chaining splitters, you would put something like an 8 way splitter near the OLT which would run down 8 fibers and each fiber would branch off along the way to feed another 4 or 8 way splitter to feed clusters of homes.

3) distributed tap: This would be the "split as you go" approach. The taps can be ordered with uneven splitting ratio so that (for example) 10% of the optical power went to the drop to the customer, and 90% continued on to the next split point.

4) a combination of 2 and 3: This is actually the method I have chosen for my design, I'm doing a "distributed tap" architecture, but each tap is feeding an 8-way PLC splitter for a cluster of homes.

The reasons I am going with #4 above are 1) fiber count is GREATLY reduced. 1 strand of fiber can feed 8 blocks, each with it's own 8-way splitter. If I went with the centralized splitter, it would take a 64 count fiber to accomplish the same thing. 2) because fiber count is greatly reduced, the time to restore service in the event of a cut is also greatly reduced. I might be looking at re-splicing 12 strands of fiber rather than 768 strands to feed the same number of customers. And since I'm doing an aerial installation, there is bound to be a tree here and there that falls on the line.

said by TomS_:

IMO you would be better off with a centralised splitter simply because its more modular.

You might also consider a lower split count for performance reasons. The more times you split, the more customers will be sharing the same bandwidth.

And I think those are both very valid reasons to go with the centralized splitter. In the end, you have to weight out the pros and cons of each method and choose the one that is not only most economical up front, but also the best long-term decision you can make.

Hopefully I'll get a chance to post my plans on here when I get a bit more time.


TomS_
Git-r-done
Premium,MVM
join:2002-07-19
London, UK
kudos:5

But a 64 way split doesn't necessarily imply a 64 count fibre.

You'll have a single fibre core feeding the splitter cabinet from the headend. Split that up n ways at the cabinet.

Each of the "taps" Im talking about (in the picture I linked to earlier) has 6 or 8 ports on it, just like what you were talking about.

The tap has a 6 or 8 count fibre run back to the splitter cabinet, and each fibre plugs in to a port on the splitter. So you have a series of small fibre bundles. Granted, you'll have a bunch up of them somewhere, but you can progressively restore service in batches if something ever happens.

As you say, different methods of deployment, and you need to choose the one that makes the most economic sense, but which has the best long term outcome as well. The later is probably the hardest bit to determine, because you never quite know what the future holds.


jcremin

join:2009-12-22
Siren, WI
kudos:2

said by TomS_:

But a 64 way split doesn't necessarily imply a 64 count fibre.

I guess it depends on topology, in a true "centralized splitter" architecture with the splitter at the headend, a 64 way split would be 64 fibers.

said by TomS_:

You'll have a single fibre core feeding the splitter cabinet from the headend. Split that up n ways at the cabinet.

Each of the "taps" Im talking about (in the picture I linked to earlier) has 6 or 8 ports on it, just like what you were talking about.

The tap has a 6 or 8 count fibre run back to the splitter cabinet, and each fibre plugs in to a port on the splitter. So you have a series of small fibre bundles. Granted, you'll have a bunch up of them somewhere, but you can progressively restore service in batches if something ever happens.

So that would be more of a distributed split then, right? The taps are simply just a splitter fed by another splitter upstream from it. Like you say, it would be small bundles of cables, each feeding a downstream splitter. So rather than the 64 fibers for the single centralized splitter, or the 1 fiber for the distributed tap, the distributed splitter approach would be more between them, where 8 fibers could each be feeding a 8 way splitter.

said by TomS_:

As you say, different methods of deployment, and you need to choose the one that makes the most economic sense, but which has the best long term outcome as well. The later is probably the hardest bit to determine, because you never quite know what the future holds.

Exactly. Very hard to look 10 or 20 years down the road!


TomS_
Git-r-done
Premium,MVM
join:2002-07-19
London, UK
kudos:5

2 edits

said by jcremin:

So that would be more of a distributed split then, right? The taps are simply just a splitter fed by another splitter upstream from it. Like you say, it would be small bundles of cables, each feeding a downstream splitter. So rather than the 64 fibers for the single centralized splitter, or the 1 fiber for the distributed tap, the distributed splitter approach would be more between them, where 8 fibers could each be feeding a 8 way splitter.

Yeah its definitely a kind of distributed splitter model I guess. I see the confusion with the term centralised now, sorry.

The cabinet I am using as an example is remote of the head end, situated on the side of a street out in the 'hood. The article the image came from seems to indicate that there are several of these cabinets spread around the area, and matches my expectation of the particular model used for that deployment.

From the cabinet I believe that each house is fed by its own point to point fibre which is presented on one of the "taps" that are installed in each pit, rather than the tap being a splitter in itself.

I'll explain my reasoning

Looking at the picture here: »www.bitsonlight.com/ftth/images/···inet.JPG

Top right you have some aqua coloured patch leads. These are your feeds from the headend. Installed below, but hidden from view, are the splitters. Example product (since this is also a Corning installation): »catalog.corning.com/CableSystems···id=77146

The pigtails from the splitters are patched in to individual ports on the left, each of which represents a core in a feeder running to a tap installed in a pit (or on a pole.) From the tap is a lead in fibre running to a house.

So Im guessing behind the panel on the left, if you swing it out, would be a lot of splicing and patching, fibre spaghetti.

Further reading: »csmedia.corning.com/CableSystems···-735.pdf

Thats a datasheet for a similar style cabinet. It seems that two cables are pre-installed: one for the feeder from the headend, and one for the distribution to the clients. So I would guess there is a joint in a pit nearby where you either splice in higher count cables to fan out around the neighborhood, and/or the tails from the taps nearby.

Now of course none of this is suggesting that your way is any more or less the correct way to do it. There are several ways to skin a cat. This is just one.

jcremin

join:2009-12-22
Siren, WI
kudos:2

said by TomS_:

Yeah its definitely a kind of distributed splitter model I guess. I see the confusion with the term centralised now, sorry.

Haha, yeah, sometimes it's tough when I might use a certain term one way and you might use it differently. I think we're on the same page now!

said by TomS_:

Top right you have some aqua coloured patch leads. These are your feeds from the headend. Installed below, but hidden from view, are the splitters. Example product (since this is also a Corning installation): »catalog.corning.com/CableSystems···id=77146

The pigtails from the splitters are patched in to individual ports on the left, each of which represents a core in a feeder running to a tap installed in a pit (or on a pole.) From the tap is a lead in fibre running to a house.

I haven't spent any time looking up those products yet, but do you know if that splitter/tap box in the underground vault is actually a splitter fed by one fiber, or if the 8 (or whatever) ports each have a fiber going down the line coming in and it is more of just a breakout box?


TomS_
Git-r-done
Premium,MVM
join:2002-07-19
London, UK
kudos:5

said by jcremin:

I haven't spent any time looking up those products yet, but do you know if that splitter/tap box in the underground vault is actually a splitter fed by one fiber, or if the 8 (or whatever) ports each have a fiber going down the line coming in and it is more of just a breakout box?

Im actually trying to find some documentation to confirm this myself, to ensure Im not spreading rubbish and to confirm my own understanding of how it works.

I'll post back when I find something.

But my understanding is that each port on the tap has its own individual fibre.


TomS_
Git-r-done
Premium,MVM
join:2002-07-19
London, UK
kudos:5

Ok.

Im not spreading rubbish. Yay.

But it seems that Corning makes TAPs (yes, they call them TAP, but its an acronym for "tethered access point") in a couple of different varieties.

The first:
»catalog.corning.com/CableSystems···AFTA_web

Works the way I have been describing, where each port on the tap has its own individual fibre.

The second are split as you go type taps, and come in two varieties of their own:

1. With expansion port (for daisy chaining I suppose)
2. Without expansion port

Corning have numerous options, different configurations and tail lengths etc, you can find them all here:

»catalog.corning.com/CableSystems···tiSheath

Looks like they basically have an entire outside plant solution, one stop shop!



54067323

join:2012-09-25
Tuscaloosa, AL
reply to TomS_

said by TomS_:

Breakout box" is probably not the right term, but its the best thing I can think of to describe it right now. The page above suggests "lead in module" but Im not experienced enough with PON to know if thats an industry standard term, or just what the operator of that PON network is calling it.

The box in the handhole is a Corning OptiSheath multiport terminal, its purpose is to serve as a handoff where an SC to X drop cable can be connected to a fiber F2.

The lead-in is a fiber stub that is fusion spliced into the cable pairs and can be ordered in lengths as needed to meet the buried fiber run.


TomS_
Git-r-done
Premium,MVM
join:2002-07-19
London, UK
kudos:5

Yeah, worked most of that out in some earlier posts.

Not sure about terminology in the US, but in Australia where I come from a lead in goes in to the customer premises from the street.

What they call the cables in the street I am not sure.



54067323

join:2012-09-25
Tuscaloosa, AL
reply to TomS_

said by TomS_:

Ok.

Im not spreading rubbish. Yay.

But it seems that Corning makes TAPs (yes, they call them TAP, but its an acronym for "tethered access point") in a couple of different varieties.

The way the NAPS, TAPS, whatever, work is, they are run back to a crossbox which has a DWDM splitter, which lights up the distribution fiber based upon how the LGX cross jumpers are placed.


TomS_
Git-r-done
Premium,MVM
join:2002-07-19
London, UK
kudos:5

Im not sure DWDM is part of this kind of system, except maybe in a backhaul network.

In the distribution cabinet there are one or more splitters which duplicate the signal N ways to be distributed to the customers. But this is different to an WDM splitter, which separates different wavelengths for individual presentation.



54067323

join:2012-09-25
Tuscaloosa, AL

said by TomS_:

In the distribution cabinet there are one or more splitters which duplicate the signal N ways to be distributed to the customers.

Having done a lot of PON I seriously doubt that is happening.


TomS_
Git-r-done
Premium,MVM
join:2002-07-19
London, UK
kudos:5

All of the PON implementations I have seen work by sending the same downstream signal to all ONTs, and this is achieved using 1xN way splitters. The ONTs pick out the data that belongs to them, and ignore the rest.

In the upstream direction the ONTs use a TDM based protocol, having time slots in which they can transmit.

If you read some of my earlier posts I linked to datasheets and other information about specific products that achieve this method of operation.

Im not saying that its impossible to have a WDM based PON network, but Ive never seen or heard of one out in the wild. Im sure there are many different ways to achieve the same kind of outcome, we have just been discussing one particular method in this thread.



54067323

join:2012-09-25
Tuscaloosa, AL

said by TomS_:

All of the PON implementations I have seen work by sending the same downstream signal to all ONTs, and this is achieved using 1xN way splitters. The ONTs pick out the data that belongs to them, and ignore the rest.

In the upstream direction the ONTs use a TDM based protocol, having time slots in which they can transmit.

Interesting, that quite different from the PON's I have worked with and yes I can see how it would work.

Let me ask you a question, if an ONT locks up and just sends continuous light how does one go about finding it and how many ONT's can be on a strand before timing gets flakey?


zed173

join:2010-07-17
Mississauga, ON
reply to TomS_

You will have DWDM (or more likely CWDM) in a PON if for example you are doing an RF overlay (ie. like Verizon FIOS or similiar). You could also encounter DWDM from CO -> Remote on a main line before it's peeled off to different feeders to keep your physical fiber usage lower if need be.


lutful
... of ideas
Premium
join:2005-06-16
Ottawa, ON
kudos:1

1 edit
reply to TomS_

said by TomS_:

All of the PON implementations I have seen work by sending the same downstream signal to all ONTs, and this is achieved using 1xN way splitters. The ONTs pick out the data that belongs to them, and ignore the rest.

In the upstream direction the ONTs use a TDM based protocol, having time slots in which they can transmit.

Tom, you are absolutely correct. Both IEEE 802.3ah and ITU-T G.984 GPON standards dictate that method.

If some company wanted to use even simple WDM (let alone "dense" WDM) on top of GPON passive infrastructure, they have to use custom OLT/ONU designs. I am curious to see a link to such a commercial product - if they actually exist.

P.S. ITU uses ATM frames downstream while IEEE uses 802.3 frames. Most OLT/ONUs are "universal" and can support either ITU/IEEE method. However, they can't support WDM without a complete re-design.