Powerline adaptors

Apart from nothing working, I cant see why that would be a problem...

The ISP's SFP * will measure (roughly) optical signal level returning, and have management communications by wire to a local monitoring system.

The ONT may well measure its incoming signal and pass the information back among other housekeeping information, at least while the return link is working.

• formatting link

There is more then one fibre connector in fairly common use. The most likely is the LC, which will be used at the ISP end but possibly not the subscriber end. The SC is another likely possibility. Patch cables for both should be easily available, as well as back-to-back joiners. Like Ethernet patch cables, fibre ones are all male-male.

The incoming single-mode fibre will be terminated with an SC/APC connector. This has an 8 degree angle at the polished end of the fibre to ensure that reflections are not sent back down the fibre to the far end. SC/APC connectors are almost always green. Similar shaped connectors with a flat end are normally blue. It is not a good idea to mix these types. John

GPON at the "home" end uses SC with APC (I've variously seen that described as angle polished contact, or angled physical contact)

It would surely depend on the type of interface provided by the ONT, on most (in the UK anyway), as you say, it wouldn't work. However I believe there are some fibre services where the ONT provides a bog standard TCP/IP connection, in which case the above warning would apply.

OK, it'll be physical, all fibre ends are polished. I've encountered E2000/APC connectors. I know where there are a pair of E2000/APC to LC patch cables...

It's pretty hard to get a straight answer. Something comes out of the ONT. Few broadband access points are known for using "vanilla anything", because this implies the ISP does not care to manage their infrastructure. Each option has pluses and minuses in terms of manageability. CGNAT is about "preserving IPV4 addresses, at a cost to the customer of the customers all appearing to have the *same* IP address".

If I knew you were using CGNAT, I would not rent your service.

Dialup, was using PPP for a while. There were options like header compression and the like. You could watch the log of the session setup, until the session went "on line" and you could no longer monitor it.

It was a surprise when the first ADSL came along and it was running PPPOE, still logging in with a username and password, but those were located somewhere and stored for you, so you did not type them. As a Mac user, I was given a piece of software to "terminate" the protocol (convert PPP encapsulation to regular packets), as the OS had no native method. After innumerable crashes of the software, I bought my first router for $300 and the router had a dialog for username and password. That's an example of a router "terminating" a protocol. What came out of the router, was plain/usable packets for Ethernet.

So it would go with the ONT. A modern router has some number of protocol handlers in it, for sniffing and switching to the correct protocol for whatever the "box" is sending. That's your ONT. The ONT could be relatively "transparent", allowing some reseller to determine the protocol.

While you could connect the OS directly to the ONT, you need some piece of software to terminate the protocol and make usable routable packets at OS level. The provider could give you the needed software, that's if they promoted a relatively insecure setup.

IPV6 uses security through obscurity (large address space for the home LAN), rather than a NAT cone. NAT is Network Address Translation, providing somewhat of a wall to unlimited (scanning) access. The port numbers are limited, but the packets can have sequence numbers to prevent spoofing on an arbitrary port.

Paul

Thanks for all the replies.

I have done a bit more investigating. Switching off the Humax PVR, and setting the input into the TV to 'Freeview Live TV', I then navigated to Settings. Under General Settings I selected Network & Internet, which shows several available networks, and displays the name of my new hub, saying it has a limited connection, along with other networks such as EE-Wifi (no internet access) etc. Clicking on my hub brings up its details such as 'connected', IP address, Randomised MAC address, signal strength 'Fair' etc.

From all that I take it the TV is at least seeing my hub and is connected to it.

But when I try to bring up BBC Iplayer for example, I get the message 'no internet connection', when the above tells me that the TV is at least seeing the Hub. Perhaps I need a password or something to activate it, but it doesn't ask for one, just 'no internet connection'.

But...

On my PC, which is in the same room as the TV and not far from it, I can access the Iplayer without difficulty and log in to it as I have always done in the past (it cleverly registered the fact that I have changed the hub, without my having to tell it). So the fact that the new hub is several rooms away, through several concrete walls, doesn't bother my computer connection.

Even the TV seems to be aware of the new hub, but is not able to make use of it. Does any of this explain to anyone why the TV can't apparently connect to the internet when the PC can? Is there something I should be switching on, or registering, to bring it to life?

Chris.

Two things see odd. What the F is a 'limited connection' and why does it have a randomised MAC address?

Look again at the TVs network setup screen and make sure its set for automatic address allocation and that it has a default route, DNS address and so on...and is connected to the correct wifi point.

I assume you keyed in the wifi password correctly on the TV?

It is a bog standard *Ethernet* connection, just one that goes nowhere until you run PPPoE over it to connect to your ISP. Who will require a name and password login.

If you are smart enough to know how to set up PPPoE on e.g a PC you wouldn't be worrying about internet hackery across an internet link that doesn't exist

... irrelevant bollocks snipped...

Sounds as though it knows it's connected to a network, but is unable to reach the internet from there?

Since the Panasonic is an android TV, it probably behaves the same as a phone/tablet would, which is to prevent tracking on public WiFi.

But that measurement won't be very accurate.

You can get an optical power meter, for reasonable measurement of optical signals. It can measure the invisible light on the ISP

Normally on equipment, a simple "LOS" LED is all you need, and that can indicate a loss of light, or a loss of modulation (the equipment is not able to "frame" to the stream of

To pick up power level, you can measure the AGC voltage on the receive amp, but that is not calibrated and would only be useful on a "bar indicator" type display. My signal is four bars, my signal is three bars, and so on. The amp for a PON system, might not be that fancy.

The first receiver we had in our lab, was a transimpedance amplifier. A university professor came for the summer, to design one for us. We did our own after that. While this makes fun of operational amplifiers, as the frequency climbs in circuit usage, the amplifier is built out of discrete transistors after a while. Later, it's a microwave amp. Still later, it shoots off to infinity and beyond (some DSP conditioning is applied early in the circuit, details not available).

Paul

Which isn't a lot of good is it?

Quelle bolleaux

That will be the sum of all the upstream traffic from everyone split from the same fibre ...

It will be split N-ways from a fused-together bundle of fibres. That's my understanding of how a PON works. We were not doing anything related to PON at the time, and someone in the optics groups was commenting on how amazing it was that a simple fusing technique could result in relatively even power split into each branch.

The fibre signal usually has some control over 1's density and DC balance, so an AC amplifier can amplify the signal. For the things we were doing in the lab, the actual packet content did not matter for a power measurement. There can also be a scrambler in the path, to improve the characteristics and prevent a user from sending a pathological pattern. If you send all zeros in your packets as a payload, they are not transmitted that way.

If the link was "idle" or if the link carries "100% data", the power level is the same, because you've gone to a lot of trouble to balance the number of zeros and ones. It's not like some little lab experiments we did in school which were NRZ or something.

In the time we had HeNe lasers in labs, only one of those was purchased with a modulation input, so of course we had to try that out. I could send characters from a keyboard across the room with that laser. And I think that was NRZ and you might have noticed the measured power level change, when a key was pressed. The data rate extremely low, the experiment pretty simple. Didn't need an amp. (Just some serial to parallel conversion on some breadboards.) But who ever purchased that laser, thank you for putting the modulation input on it.

Paul

Yes. BT fibre connections use PPPoE between the router and the ONT. Without PPPoE nothing useful will happen. There is no need to use the router supplied by BT. The username and password do not currently add to security, however, as they are not checked. Here is a snippet from a configuration file in a Firebrick 2900 router connected to the ONT of a 1Gbit/s BT fibre connection in the UK. Note the very generic username and password copied from the router originally supplied by BT.

In contrast, a similar setup in China which uses two fibre connections for redundancy DOES check the username and password of each.

The comment from Paul about IPv6 not using NAT is not really correct. It is perfectly possible to do so if there is a good reason. For example, in the China setup each line is allocated its own block of IPv6 addresses by China Telecom, but it is more convenient for the LAN only to have one address range. I therefore used a ULA subnet fd00:19:80::/64 for the LAN and NATed each of the incoming IPv6 ranges to it with appropriate traffic sharing and fallback rules.

John

Fibre combiners/splitters are remarkably cheap. I have some of these:

Most SFP modules have fairly accurate power monitoring. Here is the level of information available from an old SFP module that I happen tp have in service at the moment: Vendor FINISAR CORP. Part FTLF1424P2BCR-B1 Date code 2008/05/04 Wavelength 1310 nm Temperature 50.6 C Voltage 3.297 V TX Bias 31.09 mA TX Power 362 uW (-4.41 dBm) RX Power 303 uW (-5.19 dBm)

John

A long time ago you could connect a PC by ethernet to a Virgin Media cable modem, it did DHCP and it got a public IP. Nowadays their box includes a wifi router and you have to go out of your way to enable 'modem mode' to have the public connection come out of one of its ethernet ports, so this is not something do you by accident.

I believe some altnets operate in a similar fashion when they supply combined ONT and router in one box.

NAT is still useful with IPv6, but it's stateless NAT ie just rewriting one part of the address with another in a simple 1:1 mapping. What's problematic with IPv4 NAT is that it's stateful, ie it needs to keep track of all your connections in a giant table so that it can work out what mapping to apply. The first kind is basically free in performance terms, but the second has a performance penalty.

Theo