Since I like to give too much detail:
Range The TDM timing gives you a “maximum differential reach”. I’ve read that it’s 20km on GPON, 40km on XGS-PON, and 100km on NG-PON2. I haven’t read the actual ITU specs, but those numbers were touted to be the “standard”. Whether that’s true or not: In the real world, our Nokia platform lets us configure both GPON and XGS to 40km differential. The “differential reach” is the difference between the nearest and furthest ONT. So if our nearest ONT is 20km down the line, then the furthest could be 60km. Someone said XGS could go 100km. That might mean their vendor lets them configure a higher differential reach, or that might be assuming the first splitter is 60km away. It’s a true statement either way, but at 100km it’ll be at the razor’s edge of the link budget. The loss from distance is something like 0.2-0.5db/Km, it depends on the wavelength. It’s worse if the fiber is very old because manufacturing methods have improved the attenuation. If it’s old plant you’d have to get the part numbers off the cable and look up specs. I’m not sure how old it has to be to matter. I have not seen fiber old enough where that mattered, so that’s an academic topic for me. Fusion splices can be as low as 0.02dB attenuation, but I think the typical standard is 0.2dB or better is acceptable. Connectors vary, but you can say 0.5dB and it shouldn’t be worse than that. Most of your losses come from splitters, and that’ll be exactly what you’d expect. 3dB every time you cut it in half, plus maybe 0.5-1 for insertion loss. On C+ optics you start with a tx power of +6-+7dBm. Receiver sensitivity at the ONT could vary, but -28dBm is typical. The uplink wavelength has more attenuation, but the ONT has a little more Tx power and the OLT has a little more sensitivity. So if it’s good in one direction it’s probably good in the other one. We target -20 so we have plenty of margin. So we have some room to cut it a little closer when we have to, like adding a splitter for a duplex that was originally counted as a 1 family home during our planning, or losses from repairs, or whatever. The spec sheet for our everyday ONT says sensitivity is -28.5 at BER e10-3, which is an acceptable BER with FEC enabled. So in perfect conditions with a straight shot from the OLT to the ONT and no splices I could get 118km. I don’t think they sell 118km reels, so that’s not realistic, but 100km figure isn’t crazy. Realistically you won’t do that because you’ll have splitters. If you were going to dedicate a whole fiber at that distance to one customer then you’d probably do Ethernet with long range transceivers. With that much plant dedicated to them they’re hopefully paying a mint. So yeah. You could design with small nodes close to the customer, or you could design with one awesome POP reaching several towns. Either way is doable. Redundancy It’s possible. Google search for PON Type B and Type C protection. There are some fun little diagrams from Huawei here: https://forum.huawei.com/enterprise/en/hcip-access-01-gpon-type-b-protection-technology/thread/667281720950538240-667213871523442688 https://forum.huawei.com/enterprise/en/hcip-access-02-gpon-type-c-protection-technology/thread/667281974273916929-667213871523442688 I have never done it, and I don’t know who does do it. I’m not even sure if our Nokia equipment supports it. You’ll spend twice the money on OLT capacity, and you’d lose ~3.5dB from the link budget to split the path to two OLT’s. You’d consequently spend twice as much on power and all that goes with that (battery backup, generator capacity, air conditioning, space). The extra 1x2 split could potentially cut the number of subs on the circuit by as much as half. I haven’t done the math to confirm whether it’s worth it or not, but on the surface it sounds like it wouldn’t be. We’ve had bad transceivers and bad line cards, but not very often. In the market I’m responsible for I’ve had 3 transceivers fail in 3 years, and there are over a thousand deployed here. I can live with 0.3% failures on that time scale. I had one bad line card, but it had a dead port out of the box. I haven’t had one fail in service yet. Power failures happen, but when we’re building a POP to serve 12,000 households we’re not going to skimp on batteries. So redundancy would be nice to have, and someone must do it, but I’d have a hard time making the business case to my superiors. If we were charging top dollar for an enterprise/SMB service that might be a way to differentiate and justify the higher price. From: AF <af-boun...@af.afmug.com> On Behalf Of Ken Hohhof Sent: Saturday, March 16, 2024 5:40 PM To: 'AnimalFarm Microwave Users Group' <af@af.afmug.com> Subject: Re: [AFMUG] PON question Mike Hammett kind of touched on what I was asking and why. I was told that Metronet near me had a hut in Batavia that also served St. Charles, Geneva, West Chicago, etc. via PON. Also a company that built a middle mile / anchor institution fiber network with a BTOP grant 12+ years ago convinced the county to let them take it private, and they have run aerial fiber in most of Shabbona which is one of the towns Mike mentioned. With my misconception about how FTTH is typically deployed, I expected there to be at least one cabinet or hut in town. But I think they are just using strands from the BTOP project and feeding it passively from a distant town. I would prefer to see more redundancy, especially since both buried and aerial fiber definitely gets damaged around here, but I guess practical results matter more than what-ifs. At least local power outages shouldn’t take it down, and a central NOC or hut should be able to have serious battery and/or generator backup. From: AF <af-boun...@af.afmug.com <mailto:af-boun...@af.afmug.com> > On Behalf Of Josh Luthman Sent: Saturday, March 16, 2024 4:12 PM To: AnimalFarm Microwave Users Group <af@af.afmug.com <mailto:af@af.afmug.com> > Subject: Re: [AFMUG] PON question PON is one port at your end and then goes through splitters that reduce light and add ports to end up at customer ONTs. 1:128 is pretty short range and high customer count - we could never do that in a rural plant (5-15 miles). Maybe 1:64 but that's about the limit. There is NO redundancy in PON. Best you could do is 2x32 or whatever splitters which is where you feed the downstream fiber with two PON ports. An engineer from Metronet told me they did that but no one could ever answer why (technically or operationally). Think like you have an AP on a tower feeding 32 customers. What are the chances you have an AP right below it with the same SSID/PSK/frequency for the customers to connect to if the first AP goes down? Think of Active E like a bunch of dumb switches. You have a 48 port switch that goes to 48 customers using 48 fibers. If the fiber feeding the switch goes down, it can go to a different fiber/uplink port. On Fri, Mar 15, 2024 at 7:59 PM Ken Hohhof <khoh...@kwom.com <mailto:khoh...@kwom.com> > wrote: Since there are FTTH people here and I’m mostly ignorant of such things, maybe someone can clear something up for me. I always assumed a PON based FTTH system had a topology kind of like HFC. I expected fiber down the street with splitters, but fed by some sort of neighborhood node in a cabinet with power and electronics, fed by active EPL style fiber. Which could have redundant paths, rings, etc. so a fiber cut wouldn’t take down a whole town or multiple towns, the backbone traffic would reroute. I’ve been told this is not the case. And that instead, each PON could go back over a strand to a headend several towns and many miles away, all passive. Sorry for the poor description of my question, hopefully you can figure out what I’m asking. -- AF mailing list AF@af.afmug.com <mailto:AF@af.afmug.com> http://af.afmug.com/mailman/listinfo/af_af.afmug.com
-- AF mailing list AF@af.afmug.com http://af.afmug.com/mailman/listinfo/af_af.afmug.com