BL> The frequency response gives you a good idea of how possible HS is

BL> The frequency response looks like this...

BL>   The frequency response looks like this...

BL>      _______________________________
BL>                                      \
BL>                    |               |   \--------- -35db
BL>                    |               |     \
BL>                    |               |       \
BL>                    fc              fo

Pity that phone lines normally dont look anything like that.

BL> The significant part is the roll-off frequency,

There *IS* no 'roll off frequency' in this situation
Bob. Thats something that makes some sense with audio
stuff, and is NOTHING like real world phone lines.

BL> and how sharply it rolls off compared to the carrier frequency used.

Still crap. Not with a modern trellis protocol. Soorree.

BL> The shape of this curve will vary with the modem, according
BL> to how much preemphasis is used in the transmitter.

Nope, thats crap too.

BL> The USR seems to manage a lower carrier frequency,

It doesnt, and what difference there is is quite trivial in fact.
Thats just ONE of the parameters that can be varied to get the
maximum thruput out of a particular measured line. And unless
you have a grossly non ideal line, isnt very important at all.

And even on a particular session which has a very lumpy characteristic,
its primarily adjustable so your can move WRT the lumpy bit.

BL> plus 2dB more compensation at 3450Hz.

You dont even know that. Waffle.

BL> A reasonable limit seems to be -35dB.

You cant say that either, modern trellis protocols aint that simple.
Which is WHY they can get so much more out of a particular line.

BL> If you can't get that under 1.5 * fc, then you
BL> are basically going to lose 28800 connects.

Tripe.

PE> It was this that Bill kept on posting, and telling me my lines
PE> were shit. I didn't hear you say that he's misreading the stats.

BL> Bill is not misreading the stats. Unlike you, he reads a
BL> frequency response as second nature, and understands that
BL> once you get under 30-odd dB you are going to be in trouble.

Pity we DONT EVEN KNOW IT IS UNDER 30db,
its MUCH more likely its a pack of lies Bob.

PE> Certainly we have all been using these as a measurement
PE> of line quality, that's why we were originally reporting
PE> that my lines were so bad. We weren't saying "The Netcomm
PE> isn't loud" we were saying "The lines are bad".

BL> I naturally assumed that all Austel modems would use the same specs.

You dont even know the DONT Bob.

BL> It has turned out that *BOTH* your lines ar crook and the USR
"shouts".

You dont know EITHER of those. Pauls line ISNT crook, and the USR
DOESNT shout anyway. Nice theory Bob, pity about the reality tho.

BL> This is the usual situation in engineeriong. It is rarely just
BL> one thing causing a problem; it's usually a combination of things.

Or Bob in waffle mode. Again.

BL> Even so, the basic problem is that both you and Bill have lousy lines.

You dont know that, in fact we KNOW that Paul hasnt.
From the calls Russ made, Courier/Courier. Poor old Bob.

BL> The lines *are* bad. That's the main thing.

Pity its crap, only Bills is rather down.

BL> But there are other things beside, including
BL> noise, about which we do not have a clue.

Pity the stats list it Bob. YOU certainly dont have a clue, thats right.

BL> The would is not as simple as you see it.
BL> This is caused by your lack of knowledge of the world.

Says he who hasnt got a fucking clue how V34 works.

PE> My lines, *I* mean what's between the modems. If it were the comms
PE> link, ie between the two computers (or between the two RS232 cables),
PE> someone should have mentioned that, and not started blaming lines
PE> for what could well be an adjustable parameter in the modem.

BL> The comms link consists of a modem on either end, and a line between
BL> them. The modem won't change the line, but it can certainly change what
BL> it sends up the fucking line. Even an idiot like you should realise that.

You dont even know that what is sent up the line changes.
In fact on the transmit level, we know it doesnt. Poor old Bob.

BL> The *stats* are created when one modem (the USR) asks the other to send
BL> a range of signals and then measures the levels at *its* end of the line.

Yes.

BL> This tests the line *plus* whatever the other modem is doing.

Yes, but the whole POINT of V34 is that the other end is SUPPOSED
to be doing a standardised thing in that situation SO the result
at the other end is meaningful Bob. BEFORE they scrutinise the
results and decide what to do with the various adjustable parameters.

Its pointless doing it any other way.
@EOT:

---