DIY ADSL connection?

[The Natural Philosopher]

No. Somes sort of FSK IIRC.

[The Natural Philosopher]

there is in the limit no distinction.

Digital signals are approximated by analogue electronics which have at their heart digital atomic electron transitions as the core element.

If you like an analogue signal is in fact a digital approximation done by counting electrons.

The sound of each electron hitting, is the noise.

[The Natural Philosopher]

No, it's not - its amplitude and phase modulated carriers.

[The Natural Philosopher]

That is exactly correct for TCP/IP

An interface can have many many connections, some even to the same port.

They just aren't 'physical'

That's the whole beauty of TCP/IP. Many virtual connections can exist over the same physical piece of wire.

[Andrew Gabriel]

I have ZyXEL P660R-D1 (basic modem with NAT) and ZyXEL P660HW-T1 V2 (more advanced modem with NAT, Firewall, WiFi), and both auto reconnect.

Still have a couple of old Dynamode R-ADSL-C1 modems. These also reconnect, but do periodically lockup for other reasons and so are regularly checked and if necessary rebooted by a computer on site. Eventually they die completely.

Also still have a very old BT-badged Alcatel modem as a spare, which reconnects without any problems. These are infamous for the internal mains PSU dying, but mine's still OK.

[Andy Wade]

It could be either. If the modulation represents a continuously varying quantity (e.g. as in PAM or PWM) it's analogue. If the modulation represents discrete states from a finite set (e.g. OOK, ASK or FSK) it's digital. Even plain old OOK morse ('CW') is essentially digital, although you could argue that subtle timing variations in the sender's 'fist' is additional analogue information:~)

The essential characteristics to consider are these:

Analogue

--------

- transmitted signal has continuous variation, representing an input signal with similar variation;

- imperfections such as noise and many types of distortion arise in transmission and are always present in the received signal to some extent;

- the original input can never be perfectly reconstructed for retransmission. In a chain of relays there will be cumulative deterioration;

- tendency to gradual deterioration with degrading S/N ratio in the channel.

Digital

-------

- transmitted signal represent discrete states;

- received & decoded signal can have a vanishingly small probability of error. Provided the channel parameters are within certain limits, the original input can be reconstructed free of added noise and distortion;

- no cumulative deterioration in long chains or relays;

- tendency to rapid deterioration with degrading S/N ratio in the channel (digital cliff effect).

That's a carry-over from thinking of all RF systems as analogue. Really a carrier is just a sinusoidal signal; it's neither analogue nor digital. How you modulate it determines that.

Not if significant signal processing is performed, as is very much the case for digital modulation schemes like xDSL and COFDM. A line driver doesn't do things like multi-carrier FDM, interleaving, FEC coding and I-Q modulation.

Point taken, but that's just part of the electrical signal processing in the modem chain. In fact it's normal to use DACs to create the I and Q baseband components that are multiplied by the carrier in a balanced modulator, and ADCs after downconversion at the receiving end. The baseband waveforms are certainly 'analogue-like' (no 'square' waves here as they're band-limited) but that doesn't make the modulated signal analogue, in my view. Call it analogue signal processing within a digital system if you like.

Would you consider Ethernet on twisted pair analogue or digital? It's certainly influenced by electrical issues like attenuation, frequency response and crosstalk. DACs and ADCs could be used to send and receive the actual line signals.

And what about high-speed data buses in computers? Noise, overshoot & ringing and transmission line issues such as reflections and impedance matching are all relevant. Are these buses analogue or digital transmission systems?

[Andy Wade]

'Symbols' are not 'carriers'.

A symbol (one ore more bits) is the basic unit of data transmitted. Symbols are transmitted in time sequence to send a data stream.

number of carriers (one or many) is quite separate. Multi carrier FDM systems like DMT and (C)OFDM are used so that a fast stream can be broken into many slower ones, each with a correspondingly slower symbol rate. This, within limits, can make a system much more tolerant to multipath reflections occurring in the transmission channel.

[The Natural Philosopher]

all analogue of course.

All computers are in fact analogue, only the DATA is held in digital format.

Never the SIGNALS.

[tony sayer]

Yes indeed, 'tho not as digital as might be thought.

We have a Digital FM transmitter .. work then one out;!..

[Timothy Murphy]

Thanks for your info. But how do you reboot the modem by computer, as a matter of interst? In my case when the modem goes down I cannot access it from the desktop (to which it is connected by ethernet).

[The Natural Philosopher]

What you all need to appreciate is the data may be digital, electronics is always analogue.

To get a digital data transferred over an analogue channel takes a modem.

even a line driver is a sort of modem

[Andrew Gabriel]

I can still get to the web interface, and hence select reboot. I used to have one on a site with computer controlled power outlets, so I could power cycle it, but it eventually completely died and has been replaced with a ZyXEL P660R-D1 which has been completely trouble-free, and I no longer even poll with the computer.

[Andy Wade]

Hmm, that's a rather unorthodox view. It amounts to saying that there's no such thing as a digital signal. That flies in the face of established usage, and the IEV which gives the following definitions for "signal" and "digital signal":

Signal:

- physical quantity, one or more parameters of which carry information about one or more quantities NOTE ? These parameters are referred to as ?information parameters?.

[IEV number 351-21-51]

Digital signal:

- signal whose information parameter may assume one out of a set of discrete values.

[IEV number 351-21-54]

[The Natural Philosopher]

Information parameter == DATA

What that is saying is that e.g. the information parameter held as 0 or

1 may be expressed by e.g. voltages above below 0.5v and or above say 3v.

It doesn't say that 2v has meaning. Which is why you have a clock..to say 'don't look at me while I am changing, darling'

Neither does it say that the voltages or the circuitry associaed with them are in fact anything other than analogue devices, and they are.

When we talk about 'digital electronics' what we mean is 'analogue electronics that processes digital data'

[Grimly Curmudgeon]

We were somewhere around Barstow, on the edge of the desert, when the drugs began to take hold. I remember [e-mail] (Andrew Gabriel) saying something like:

I've got the v3, as supplied by my ISP. Total pita it is, as it keeps filling the log entries with 'total number of sessions per user exceeded' or somesuch, then craps out, but a reset sorts it for two days. In the Network/NAT settings, it has "Maximum number of NAT/firewall sessions for the router is . To remove the per user limit, set to

4096". I've done just that, but it makes no difference.

Hohum, I can see a Netgear DG834G in my future.

[Andrew Gabriel]

Are you running something that generates lots of network connections or cycles through network connections rapidly? On mine, it's set to 512 NAT/Firewall Session Per User. Not sure what a "user" is in this case - maybe each internal IP address? However, I've not seen what you describe and I've heard no complaints from the users.

[Mark]

I had one similar Zyxel router (probably the V2) and also did not experience this problem. This is no help to GC though. Sometimes things just don't work for a few people.

Don't hear much good about these nowadays, except for the ones with the Broadcom chipset.

[Timothy Murphy]

What do you hear about Linksys routers? I have a Linksys WRT54GL (running wrt-dd) which has worked perfectly for me for a number of years; I'm wondering if this is still the best Linksys offering?

[Mark]

I don't use Linksys routers myself but I believe they are fairly good as a whole.

[Grimly Curmudgeon]

We were somewhere around Barstow, on the edge of the desert, when the drugs began to take hold. I remember Mark saying something like:

Ah; it might be utorrent. I'll try it without for a couple of days and see what happens, thanks.

Noted, ta.