John Aldrich wrote:
> >
> I have to wonder if this isn't a symptom of the default to
> K56Flex by the Rockwell chipset and the default to X2 by
> the TI chipset. I seem to recall from past emails that you
> (Ramon) use USR products there at your ISP. I know this can
> be fixed by turning OFF K56 flex on these modems, just like
> people have to do here for USR / 3Com modems. I know that
> our customers with 3Com/USR modems who don't turn off the
> X2 protocol have a DEVIL of a time connecting and staying
> connected. I suspect that the same is true of modems which
> default to K56. While these are TECHNICALLY both compatible
> modems, due to the fact that they both default to trying
> their mutually-exclusive PROPRIETARY protocols FIRST,
> there's a conflict and therefore you tend to get lousy
> connections, IF you connect at all.
No, they both connecting with V.90.
My terminal servers are Livingston Portmaster PM3. They
use Lucent chipsets. Their protocols are K56Flex/V.90. The
server will tell the client first to try V.90. That is the
default and is shown in the logs.
I tried Cisco 5200 and Ascend 4048, these both have Rockwell
chipsets. Neither would negotiate a connection using either
V90 or K56Flex. Max speed with them was 31,200.
On the client side, the X2/V.90 modems using the TI chipset,
such as the US Robotics or Phoebe connect fine to the PM3
at V90 protocol with speeds ranging as high as 53,000, although
it is rare to see over 52,000. The PM3 will not do X2.
I have never really seen any K56Flex connections. The code
for this is rather old now, and the K56Flex and the Robbed
Bit signalling/D4/AMI/CT1 used by the Telephone Company in
their DMS-10 Nortel switch is not conducive to their working.
There are worse switches, I understand from some ISP's that
what their phone company uses has never allowed any of the
56K protocols. This all has to do with how the phone company
changes the signal from Analog to Digital in their Codecs.
My observation has been that the USR/TI chipset client modems
have better throughput, and much less susceptibility to noise
and adverse line conditions; fewer disconnects. The Rockwells
are aggressive in their connection speed and frequently hook
up at too high on their speed. In that case, the connection
usually fails -- either a disconnect, or worse yet, just
pokey throughput regardless of reported connect speed.
The best of the WinModems, for what it is worth, is the Lucent.
They are now better than the US Robotics WinModem. The Lucent
needs a driver of Version 5.66 or better. With the new driver
it is a good performer.
Keep in mind that the DSP in a WinModem consumes a tremendous
amount of CPU power. It requires a Pentium 166 in order to
even work. With a Pentium 166, if you need to move the mouse
pointer 1/8th of an inch, or scroll a window, or do some
other computational operation, the Modem generally has to
be told not to transfer data. For this reason older pentiums
do not do well at all with a WinModem. On top of that, most
older Pentiums have old WinModem drivers, so the problem is
compounded.
As much as we bitch about Win Modems, the truth is that since
the Lucent 5.66 the situation has gotten better. The latest
Rockwell drivers are getting passable now. In most cases I
can connect a new 450 mHz Pentium to Nook Net via a 56k
Rockwell and have it work. The problems is with out of town
phone lines. The Rockwells simply fail over longer wires.
My laptop with a Phoebe External just connects fine. Most of
those Rockewells NEED extra settings to tame the connect
speed, or restrict it to V.34 protocol (or have it dial my
analog modem bank which I still operate for those with
RECALCITRANT modems).
Such is the life of an ISP.
--
Ramon Gandia ============= Sysadmin ============== Nook Net
http://www.nook.net [EMAIL PROTECTED]
285 West First Avenue tel. 907-443-7575
P.O. Box 970 fax. 907-443-2487
Nome, Alaska 99762-0970 ==== Alaska Toll Free. 888-443-7525
