david, at 21:53 on Feb 27 1996, you wrote to Bill Grimsley...

db> On Feb 27, 1996 at 07:40, Bill Grimsley of 3:640/305.9 wrote:

BG> # 25 Feb 09:58:12.16 Freq     Level
BG> # 25 Feb 09:58:12.16  150     40
db> [...]
BG> # 25 Feb 09:58:12.60 3750     59
BG>
BG> Basically, once the level drops below around -35dBm for a given frequency
BG> (could be marginally higher at -33dBm, I'm not sure), the modems should

db> So "Freq" is in this case carrier frequency measured in
hertz, yes?  
db> "Level" would therefore be measured -dBm? 

Give the man a cigar!  :)

db> Now is this table telling me that at the time you issued the command to 
db> generate the table, your USR estimated (experienced?  probed and 
db> discovered?  what?) that for a frequency of "x" you would
get (require?  
db> what?) a line level of "y" - yes or no?

That display is basically a hard copy of what the USR does continuously
during a call, and any deviations in line levels for a given carrier
frequency will cause the modems to shift their speed (bit rate) to
compensate, but unlike a full retrain, this is done instantly and
transparently, according to the V.34 specs.  Even the NetComms will speed
shift, as I've since discovered.

db> So poor frequencies (as measured by the line level) are therefore 150Hz, 
db> 2,400Hz, 2,550Hz, 2,700Hz, 2,850Hz, 3,000Hz, 3,150Hz, 3,300Hz, 3,450Hz and 
db> 3,600Hz - yes?

Any value below (i.e. numerically higher than) say -35dBm is rather poor,
but if that's the best level available, regardless of the carrier
frequency, I'd suggest that the modems are going to have problems
maintaining the connect.

db> Is the level the receive or transmit level?  The former, yes?

Not specified, but I'd suggest that it is indeed the Rx level.

BG> shift down to the next highest freq/level pair, and the above stats
BG> indicate that the highest carrier frequency for a successful link is
BG> around 2250Hz...

db> Okay, I can see this.

See?  It's easy once you know how to interpret the stats.  :)

BG> ...say 21600 or 24000 bps (or a step higher on modems
BG> supporting 64-state trellis coding).

db> Does anyone know the approximate frequencies for each ITU-T recommended 
db> speed?  How do you get 21.6Kbps from the table?  All my doco regarding bit 
db> encoding into the carrier signal is dated 1990, and talks about 9.6Kbps 
db> being "high speed".  :-(

Have a partial quote from the on-line USR manual (might be worth freqing it
from Paul, as it has a wealth of non-USR-specific information in it)...
 
===============================
Transmitter Carrier Frequencies
   V.34
        Originate Mode: 1800 Hz
        Answer Mode:    1800 Hz

        Originate Mode: 1829 Hz
        Answer Mode:    1829 Hz

        Originate Mode: 1867 Hz
        Answer Mode:    1867 Hz

        Originate Mode: 1920 Hz
        Answer Mode:    1920 Hz

        Originate Mode: 1959 Hz
        Answer Mode:    1959 Hz

        Originate Mode: 2000 Hz
        Answer Mode:    2000 Hz

   V.Fast Class
        Originate Mode: 1800 Hz
        Answer Mode:    1800 Hz

        Originate Mode: 1875 Hz
        Answer Mode:    1875 Hz

        Originate Mode: 1920 Hz
        Answer Mode:    1920 Hz

Receiver Carrier Frequencies
   V.34
        Originate Mode: 1800 Hz
        Answer Mode:    1800 Hz

        Originate Mode: 1829 Hz
        Answer Mode:    1829 Hz

        Originate Mode: 1867 Hz
        Answer Mode:    1867 Hz

        Originate Mode: 1920 Hz
        Answer Mode:    1920 Hz

        Originate Mode: 1959 Hz
        Answer Mode:    1959 Hz

        Originate Mode: 2000 Hz
        Answer Mode:    2000 Hz

   V.Fast Class
        Originate Mode: 1800 Hz
        Answer Mode:    1800 Hz

        Originate Mode: 1875 Hz
        Answer Mode:    1875 Hz

        Originate Mode: 1920 Hz
        Answer Mode:    1920 Hz
===============================

I didn't bother importing V.32ter or HST specs, as they're not relevant.

BG> The "V.34+" was basically USR's own working name for what
will soon be
BG> ratified as V.34bis by the ITU-TSS (July 1996 is the expected date).

db> Didn't Arthur say that it'll still be "V.34", just a 1996 Annex?

Dunno, I've seen it referred to as both lately.  Logically though, one
would expect 33k6bps to be called V.34bis, in order to avoid confusion.

BG> That was common knowledge, so no surprises there at all.

db> Common expectation, but I wanted to confirm it.  My next question, of 
db> course, is when the heck are Rockwell going to make their DSP Flash 
db> upgradeable.  I guess NetComm finally going to Flash upgradeable ROMs is a 
db> start, and gives some faint hope.  :-)

I strongly doubt it, as the data pump is OEM Rockwell.  Unless NetComm
change to a generic DSP (like AT&T and USR for example), they're at the
mercy of Rockwell WRT what capabilities their chipsets have.

db>> What, no sacrificial ropeless bungee from the window?!?!  ;-)

BG> No, I'm not that cruel.  Unless, of course, you meant for Paul...  :)

db> That's always a possibility!  ;-)

Too much to expect Paul to jump from a 200m building without a rope, but I
wonder if he could be conned into doing so with a 210m long rope instead?

Regards, Bill

--- Msgedsq/2 3.20