PE> When you are doing the yelling, you should be yelling
PE> at a constant loudness level.  Therefore the second
PE> number you have should be consistent.  It isn't.

BG> No, because the USR doesn't default to fixed
BG> transmit levels.  It continually monitors the line,
BG> and adjusts itself accordingly.  This is why they vary.

He's right, tho its not so much varying thru the call with
the transmit level, its more varying from call to call.

PE> 1. Can you fix the transmit level?

Its pointless, its just ONE of a number of parameters that a
fancy V34 modem varys to get the best out of a particular line.

PE> 2. Of what possible benefit is it for
PE> the Courier to yell at 12 instead of 11?

When you have a handful of things you can vary like the carrier freq,
the symbol rate, the preemphasis, shaping, etc etc, its just ONE of
the variables, and with some sessions you get a better thru put if
you say reduce the transmit level a notch and increase the shaping etc.

PE> 3. How can the Courier possibly decide how loud
PE> is best to yell, without some hint from the M34F?

Thats what V34 is all about, MEASURING the line, and adjusting those
parameters to get the best thruput out of the particular measured line.
It appears to be getting rather confused with an M34F on the other end,
and doesnt with a Courier on the other end. Its less clear if thats
got a damned thing to do with the handshaking failures tho, likely not.

PE> 4. What is the form of the hint from the M34F, or
PE> from the Courier if only Couriers can report the
PE> best thing for the originator to be transmitting at.

You are getting obscessed with transmit levels, largely because Bob
hasnt got a fucking clue about modern trellis protocols and has been
raving on about JUST ONE of the parameters which is varied with M34.

V34 MEASURES the line and adjusts a variety of parameters to get
the best thruput out of the MEASURED line characteristics. No hints.

Its less clearly why the Courier is fucking up the measurement
so comprehensively when an M34F is on the other end of the line.
The most likely explanation for that is some wart in the Courier code.

PE> 5. Where does line noise fit into all this?

Its not that important on normal calls. Even when its quite audible
like when Russ calls the bulldog, V34 is STILL fancy enough to get
very impressive thruput anyway. Tho as you can see from his latest
results when its mostly under water, it can get bad enough to drop
the thruput. BUT thats thruput, NOT handshake failure.

BG> Bear in mind also that we're talking about decibels, which
BG> are logarithmic in their progression, and NOT linear.  The
BG> difference between say -30 and -33 is a hell of a lot more than
BG> just the 10% you seem to imagine. Indeed, it's more like DOUBLE.

PE> So I come through twice as loud?  Do you notice that on voice calls at all?

You wont normally hear that. It really is quite interesting to do
a series of voice calls with a direct measurement of the line loss,
and thats trivial to do with a constant tone level at the other end,
and see what variation there can be call to call, particularly in the
older system before all the fibre optic links, and see how variable
they have to be before you notice much on a voice call. Its got
to be a HELL of a lot more than 3db for various reasons.

Again, Bob is raving on from ignorance there too.
@EOT:

---