I feel a long post coming on...
I am just wondering how long on average does it take a BT Fibre Splicer Engineer to splice/join a strand of fibre?
Another video, courtesy B4RN.
A mix of stills and video, but useful part is around 5 mins in.
Also I recall a little while back we was discussing actual speeds down a single strand of fibre, I don't recall getting a definite answer.
How long is a piece of string?
Minimum speed probably around 100Mbps. Maximum speed achieved in a trial between Huawei and BT in 2014 was 3Tbps. But achieving that required a "grid" of wavelengths on the same fibre: 15 wavelengths of 200Gbps each. That is backed by serious transmission hardware.
For the access network, they need cheaper hardware at each end of the fibre.
The GPON system BT are trying desperately to put in your street is 2.5Gbps downstream, 1Gbps upstream, shared. With a distance limitation of around 20km.
The fibre network being deployed is quite capable of carry 10G-PON signals too, which is 10Gbps downstream, all to be shared.
Those speeds are more limited by the hardware they put at each end of the fibre than on the fibre itself.
Meanwhile, the fibres used for FTTC carry a point-to-point connection running at 1Gbps. They too could be upgraded to higher speed PtP connections in the future, depending on the hardware in the OLT and the DSLAM.
Standardised speeds, cheap-to-achieve, suitable for 1 wavelength on a fibre in the access network seem to be 100Mbps, 1Gbps, 10Gbps. With perhaps 25Gbps coming.
Also not really knowing how many fibres get spliced from a strand of fibre.
So say a fibre cable leaving an exchange consists of 48 strands, I assume that will get spliced as it runs down the road, but to what amount of fibres get spliced from one of the 48 fibre strands?
One fibre will be spliced to one fibre. No question about that.
When you see B4RN's video, watch the screen showing the two fibres being spliced ... there is no room to branch the connection there.
My understanding was a fibre strand that is to be used for FTTP has a bandwidth of 10Gbps and each strand gets spliced off into 30 smaller ones.
Where as a fibre strand for FTTC has a bandwidth of 1Gbps that gets split up into 15 smaller strands.
The fibre strand that exits the exchange, identical for FTTP and FTTC, will be one strand within a cable of up to 288 - but usually in groups of 12 strands known as an element.
Where one fibre (at the exchange end) is used to serve multiple homes in an FTTP GPON, the division isn't done by "splicing off 30 smaller ones". It is done by splicing that single fibre to one end of a "splitter device". That device will have 4, 8, 16 or 32 fibres at the other end - all ready to be spliced onto by the fibres strands that (eventually) make it to your house.
The splitter devices appear to be passive silicon wave-guide devices.
At first, BT seemed to be set on using a single level of splitter: one 32-way splitter device. This would be housed within a "splitter node" which had room for 4 devices, so could serve 128 properties, all from 4 fibres back to the exchange. Towards the homes, there would be a final splice at a fibre DP - where the fibre strands would change from being within a cable alongside other subscribers, to a fibre blown down a BFT that is exclusively for your house.
Nowadays, BT seem to be planning for two levels of split: The first level would be an 8-way primary splitter, and the second level would have 8 four-way secondary splitters. It still amounts to one fibre in, 32 homes served.
The 8-way splitter looks likely to be housed alongside 3 others in a smaller splitter node, while the 4-way splitter might be housed underground, in a pole-mounted housing or even at the top of the pole (in the new FoD2-style of connectorisation). In this 2-level setup, the secondary splitter is likely to act as a DP too.
The fibre for FTTC doesn't go through a splitter device at all. It is spliced 1:1 through the aggregation node to a port on the FTTC cabinet.
And where exactly does the fibres get spliced for FTTP, is it as it goes down the road from the exchange to the first daisy chained FibreDP hardware.
Or does it only get spliced in the actual FibreDP hardware where 1 ro 2 strands are spliced into 15 strands on each resulting into 30 lines for 30 addresses on the same phone pole.
There are lots of places that splices happen, for a multitude of purposes.
First, take a look at some example architectures for FTTP.
This PDF includes BT's thoughts for the whole access network circa 2010:
IWCS presentation 2010
This image is of a considerably more modern look
Two-level of split FTTP
Finally, you can also see some details of the architectural changes for FoD2:
TBB Blog on FoD2
Here's how the architecture boils down to a set of splices...
Stage 1 - Out of the building
1. The FTTP fibre starts as a connector plugged into a GPON port on the OLT.
2. The first splice is within the OCR (Optical consolidation rack), where it starts to joining up with all the other strands heading to the same area of town.
3. The yellow cables from the OCR are probably internal grade (low smoke). They will be spliced to external-grade cables down in the "cable chamber" in the basement - A "CCJ" (cable chamber joint?)
4. The cable leaving the exchange, via the chamber, will be up to 288 strands of fibre, and will form one of the fibre spines snaking out of the exchange to an area of the town.
Stage 2 - Fibre Spine
5. When the cable gets to the first area to service, it will go into an aggregation node. Here , perhaps some 48-60 of the fibres will be extracted from the spine, and "laid up" in splice trays for future use. The remainder will pass straight through unspliced to the next aggregation node in the daisy-chain.
6. Each aggregation node is a "flexibility point" like today's PCP, in that this is one of the main locations for swapping connections in the case of failures or breakages. It will support up to 1500 premises, so is perhaps equivalent to 2-4 PCP's in scale.
7. A fibre spine that starts as 288 fibres might drop 48-60 of them at each aggregation node. That might make for 5 large aggregation nodes along one spine, or more smaller ones.
8. Each fibre cable is of a limited length, so might not be able to run the complete distance between exchange and aggregation node. If so, extra "TJ"s will be inserted, complete with splices, so a new section of cable can be run. Track Joints? or Transit Joints?
9. The fibre spine is the equivalent to today's E-side cables. The fibre from the aggregation nodes towards the premises is the distribution fibre, equivalent to the D-side.
FTTC areas will only have the architecture above. FTTP areas will continue with the next level.
Stage3 - Distribution Fibre
10. The first step of the distribution fibre is to be routed to a tray in the aggregation node, and spliced to one of the spine fibres. The cable will then be fed through ducts into a splitter node. I guess that these cables are likely to have 36, 48 or 96 fibres.
11. In the example 2-level architecture, the distribution fibre will then, for example, be sent to one of the 8-way splitter nodes - where it will be located in a splice tray, and spliced to the input to the splitter device.
12. The output from the splitter device will then be located in another tray, and spliced to the fibre strand that is heading further in your direction. This fibre will be in another cable, fed through ducts towards the house.
13. This fibre will enter the 2nd-level splitter, and again put into a tray where it will be spliced to the input to the splitter device.
14. The output from the splitter will be routed into a separate tray, where it will be spliced onto another strand of fibre heading in your direction. If this secondary splitter forms the job of a DP too, then this strand will be the final strand of fibre destined for your house.
15. The final strand is likely to be fibre for blowing in BFT over to your house. Once it arrives, it will be put into the splice tray located on the outside of your property. Here it will be spliced onto cable suitable for running inside your house.
That's a lot of splices, but they're always 1 fibre strand to one fibre strand.
There are some pictures and descriptions of the exchange-side of this here:
There are some details of the insides of aggregation nodes, splitter nodes etc in here:
These guides show how fibre cable needs to be handled to form all the connection work within the nodes, but without detailing the fusion splicing itself.