-=> Quoting Joe Frankiewicz to Richard Town <=-

 -> USR chooses the 3200 symbol rate when answering a call from a modem
 -> that doesn't support the 2743 symbol rate and that indicates the same
 -> maximum projected data rate for both the 3200 and 3429 symbol rates.
 -> This does not match what other V.34 implementations do.  By choosing
 -> the 3200 symbol rate instead of the 3429 symbol rate USRs are giving
 -> up about 2dB of SNR.  On a lot of lines that is the same as giving up
 -> one data rate.

 JF> This entire paragraph was "cut-n-pasted" from a message originally
 JF> written by Dan Moore of Supra, shouldn't you be giving him some credit
 JF> for it?

Quite right.  I'm not claiming authorship, it comes from a help text file
here 3429HUH.TXT  Within it there are the credits:


Here's a confirmation from  Supra's Dan Moore  showing what they've
done to overcome the USR's implimentation method of not recognising
                 the call for a 3429 symbol rate:
 ----------------------------------------------------------------

and

--
Dan Moore
Diamond Multimedia

I now hope you'll retract the implied allegation of plagarism...
That file does however also contain work of my own which does not
require an acknowledgement anytime someone refers to it.  I'm not so
petty...

 JF> It contains some minor technical errors, but it explains
 JF> things pretty well for anyone who is armed with a copy of the V.34
 JF> spec.

 -> However were you to examine your INFO1c and INFO1a sequences you will
                 ^^^
 -> see what has and what has not been deliberately done to deceptively
 -> try and enhance USR's position in the market place.  Shame on Skokie!

 JF> What exactly has been done, and in what way does it "enhance USR's
 JF> position"?

IMO, there has been a deliberate attempt to cause less performance when
a USR is called by a modem easily recognisable as non-USR than a USR
itself.  Serious Rockwell users know that as part of the budget package
they forgoe the 2743 symbol rate for the reason that it only provides
a .11 db difference in s/n drop as against 2800.

Thus, if you're not
going to achieve a 2800 rate connect for link condition reasons, you're
hardly going to maintain 2743.  And to rely on line conditions that've
deteriorated to that extent, it's odds on that the 2400 rate will be
far more stable and not require a further in-connect rate change.

In the matter of difficult line connects it thus is more appropriate to
go streight from 2800 to 2400.

 JF> Be careful, I *DO* know how to examine the INFO sequence data...

Yes, you sure do :)

Now, as you've taken time out to respond to Dan's analysis via my posting,
(without taking issue with any of his salient points) perhaps you'd care
to explain how, when a calling modem does not indicate 2743 symbol rate
(ie Rockwell chipset-ed devices which are in the main, budget devices)
a USR will, subject to line conditions, select the 3200 rate instead
of the 3429 rate?

As a check, anyone in here may disable their 3000 rate (s54=80 on _both_
Couriers _and_ Sportsters,  or on a different chipset-ed modem from the
same stable the 92MHz TI chipset-ed Sportster with s54=16)  Now see those
3429 rate connects come tumbling in!

 -> It's called Modem Apartheid.  ie USRs are deliberately configured to
 -> recognise the colour of callers' money spent on modem in deciding
 -> what performance to achieve.
 JF> This is completely false.  V.34 provides *NO* way of identifying the
 JF> manufacturer of the modem at other end.

Manufacturer no, that it's a budget modem, for the reasons identified
above, yes

 JF> USR's symbol rate/speed selection logic is designed to operate at the
 JF> fastest reliable rates under all circumstances,

Not true.  As above

 JF> even if the remote
 JF> modem does something foolish or out-of-spec.

Those same modems, "even if they do something foolish" do, subject to line
conditions provide 3429 symbol rate connects -- between themselves

 JF> It does not "penalize"
 JF> other manufacturers,

Other manufacturers?  Didn't say that.  It penalises other users.  Including
USR-ers who, if running a busy system, often have to keep one connect alive
longer than necessary

 JF> and it is capable of operating reliably even when
 JF> connecting to some first-generation modems which are notorious for
 JF> handshaking bugs.

These I would enter a special plea against.  Often they're cheeep clones,
or even counterfeit, or produced into the market without a thought for
aftersales, ROM updates etc.  There were specific reasons why the first
symmetric rate V34s exhibited the "3429 symbol rate bug" which was cleared
up by both Supra and Zoom in late August last year.  There's now no good
reason why serious users of Rockwell product, or your good selves, should
have to carry cheepskates.

 JF> I would suggest you re-read Dan Moore's message, you seem to have
 JF> missed the important parts of it.   He does explain why this happens,
 JF> and it *isn't* a bug or secret plot.

I have, and done my own research here together with helpful supporters of
USR (UK) too.  I cannot assume you are berift of programming skills.  Hence
given the clear and precise spec description of V34 as regards INFo packets,
standby my claim.

Additionally would point out that calls _from_ USRs on a good line are
received here well (at the 3429 rate)  Always assuming that standard
factory defaults of &f1 for PCs are employed.  My understanding from Supra
Sonic 336+ users (again Rockwell chipset) is that they too have had no
difficulty with incoming calls from USRs either.  How can two different
makers, working in different sectors of the market, only sharing the
same datapump, come to the same conclusion?

In addition would point to lack of V42bis with MNP, lack of MNP10 on your
domestic models as further evidence (pre-SREJ) of a deliberate and calculated
attempt to show lower performance with budget models as against own.  Not
to mention the real and growing interest in Cellular which ACE caters for
so well with dynamically linked line level adjustment and MNP10

Once upon a time, users of budget modems used to look up to USR, thinking:
"One day I'll have one of those".   Not any more...

rgdZ
Richard

--- FMail/386 1.02