Skip to Content.
Sympa Menu

baslinux - Re: [BL] slow upload speed, was Re: multiple systems (among other things) -- was Re: interesting info for those with trident cards (possibly applies to others)

baslinux AT lists.ibiblio.org

Subject: Baslinux mailing list

List archive

Chronological Thread  
  • From: Samual Acorn <sam.acorn AT gmail.com>
  • To: baslinux AT lists.ibiblio.org
  • Subject: Re: [BL] slow upload speed, was Re: multiple systems (among other things) -- was Re: interesting info for those with trident cards (possibly applies to others)
  • Date: Sat, 31 Dec 2005 17:15:10 -0600

On 31/12/05, sindi keesan <keesan AT sdf.lonestar.org> wrote:
> >>>>
> >>>> S15=128 seems to keep the Sportster from attempting to use X2, so it
> >>>> negotiates and connects faster. The non-Sportster boings for a lot
> >>>> longer.
> >>>
> >>> the 'boing boing' is the modem testing the analog frequency response
> >>> of the phone line... the modem on the other end will answer with a
> >>> simple beep when testing is successfull (if testing is not successful
> >>> you will not get 56k)
> >>
> >> So why is the internal (slight newer) modem with the slightly older flash
> >> taking longer to connect?
> >
> > multiple causes for this one;
> >
> > the modem is just trying to be more thourough in its check
> >
> > bad programming causing problems during the check
> >
> > corrosion on the pins in the jack causing noise or signal loss (make
> > sure they are shiny)
>
> You mean the phone jack? I can plug it in and out a few times.

yes.. the phone jack in the modem... had (still have) a gateway 2000
telepath 14.4k internal that had some serious issues till i took a
flat-blade screwdriver and scratched it across the metal pins to get
the corrosion off... worked perfect till i 'took it out of service'...
replacing it with a rather flaky 33.6k (never did find out what was
wrong with that one... wasnt connection problems but power-on
problems... long story....)

>
>
> > since its internal may be getting interference from a nearby card
> > especially if it uses a transformer for phone line isolation (scsi is
> > a bit noisy) or the PSU might be noisy (checked the modem in multiple
> > systems?) note: this one is rare but not unheard of...
>
> I have ISA SCSI card in that computer and this is an ISA modem. (Are
> there any pci hardware modems?).

there are a few... but they use a different means (of course) to
communicate rather than the standard UART... they need software to
make them act like a regular com port in dos which isnt always the
best at compatibility (but it does work) but performace of a hardware
pci vs. isa is neglegable when used with OSs like linux or windows....
when it comes to PCI modems tho buyer beware...

> I try to put the modems as far as
> possible from the power supplies.

PSUs are sheilded in their own case... putting a modem close to one
wont hurt.... what i ment by noisy was the power supplied by the
PSU... an overloaded or tired (near the end of usefull life) PSU can
make noise (low amounts of AC or even RF) on the normally pure DC
power connectors... causing all kinds of random weirdness from modems
not wanting to work properly to "hum bars" on the monitor to
spontanious reboots

putting the modem near a scsi card or video card thats particularly
noisy (RF noise... usually from poor design) can cause interference
with the modem....

these are just possibilities to consider/try... not absolutes...

>
> > if it uses relays to 'pick up the phone' rather than solid state
> > components the contacts may be corroded... (this one is something best
> > left alone).. rare as well but still not unheard of...
> >
>
> >>>>>> S32=34 stops the Sportster from trying to connect with SX2 but has no
> >>>>>> effect on the other USR.
> >>>> It does not stop the Sportster, I was wrong.
> >>>
> >>> x2 and flex should be disabled entirely.... they are obsolete unstable
> >>> and unused by ISPs....
> >> I thought S15=128 was supposed to disable X2 but maybe only for the
> >> Sportster, which is why it connects with less boings.
> >>
> >> Ijust had a 20 sec timeout again.
> >
> > check your manual for an option for 'quick retrains' and make sure its
> > turned on (if available)
>
> I will look online for this. Just found a Supraexpress manual 28.8
> AT+MS=? lists supported options
> %CI MNP 5 enabled
> %C2 V.42bis enabled
> %C3 V.42bis and MNP 5 enabled (Default).
>
> They recommend turning OFF compression when transferred already-compressed
> files such as ZIP, because it may actually go faster. I will try it.

make sure to only turn off compression and not error correction... (i
made that mistake once... things went from slow to useless very
quickly)

>
> %E Do not request a retrain (default)
> %E1 Monitor line quality and automatically request a retrain if line
> conditions are bad.
>
> &K commands - flow control

flow control should be set to hardware (in both the modem and your software)

>
> Gateway Telepath 28.8 - supports full Hayes AT command set
> Data compression: V.42bis and MNP5 bps to 115,200 bps throughout. Modem
> also operates without data compression (maybe this would work better for
> jpegs?)
> ATI 0-7
> No %-type commands like the Supra.
> &K0 - Data compression disabled
> &K1 Auto enable/disable (default)
> &K2 - Data compression enabled
> &K3 - MNP5 compression disabled
>
> Obviously different modems use different commands. I have a few USR
> manuals around some place and will experiment.

hayes standard is supposed to be universal (but its very simple...
similar to the commands used with 9600 bps and less modems)

hayes extended is also supposed to be universal but quite a few
manufacturers dont use it... newer USRs in particular.... zoom uses
it....

>
> >>
> >> This must be why we are finding used 56K v90 for $1 (one of the three
> >> actually worked at about 56K). With people now switching to DSL, maybe a
> >> non-winmodem V92 will show upused.
> >
> > er... i woudnt count on that... the ppl buying hardware v92s are those
> > who are -stuck- with dialup...
>
> But people around here have a choice and many still use dialup.

well... keep an eye out then.... most anything dated 1998 and pretty
much -everything- dated 1999+ will either be v92 or v92 flashable...

>
> >
> > also many newish hardware v90s are flashable to v92 (USR has linux
> > flashers... might want to look into this one)
>
> Mine are 1997 and I already flashed them all.

no newer flash updates available?

>
> >>
> >> The 0459 came in that case, I meant that the case was dated 1995 because
> >> they were first made then and used with 14.4K modems.
> >> I may have read this wrong, upside down with a flashlight. We have
> >> another similar one that came as V90 not X2.
> >
> > check for V92 updates on the one that came as V90... if that was its
> > original configuration...
>
> I think that is the one that got knocked out by lightning.
>
> >>>
> >>> those 20sec freezeups are speedchange retrains.... if you use M2 you
> >>> will hear a full retrain during those pauses...
> >>>
> >>> 14.4k is technically the most stable speed a POTS line can handle....
> >>> anything faster is pushing the laws of physics...
> >>
> >> If I dialed in at a slower speed would it freeze up less often? For
> >> writing and reading email the speed is less important than a lack of
> >> freezeups. I get freezeups even with a 33.6K modem, should I try 28.8K?
> >
> > 28.8 would freeze too (26.4 is a more common speed than 28.8... phone
> > lines simply werent ment for data... and i have the unfortunate
> > situation of only being able to reach 24.0 with analog (28.8 & 33.6)
> > connections/modems...
>
> A friend who thought his phone lines only supported 28K discovered he
> could get 40K with a USB modem (after the onboard one died).

i can get ~45k/s down its the upspeed i have problems with.... checked
not to long ago to be at ~21k (always gets worse when the temperature
cant make up its mind [frigid nights and hot days].. its just
something ive come to be used to...)


> >>
> >> 3.3K. With a 33.6K modem. 3.1K with 56K - but variable. Would a 33.6K
> >> speed actually upload faster overall, because of fewer freezeups?
> >
> > yep... if you do more uploading and general traffic than just surfing
> > the web a 33.6K might be better than a 56K...
>
> I could try doing email online at a slower speed.
> If I told it not to retrain what might happen?

if things got bad the connection would instantly drop or become very
unstable... (ive tried that one too)

>
> >> Should I set connect speed to not 115200 but 57600 for 56K 1998 modems?
> >
> > if the modem supports 115200 use it... if not use the next highest
> > speed... even the modem makers say this...
>
> >>
> >> I will check our 14.4K ATT (if I did not already recycle it).
> >> Does anyone but Rockwell and USR make 56K internal hardware modems?
> >
> > i didnt even know rockwell was still making hardware internals... only
> > hardwares i can find are externals... (zoom and usr) PCI hardware
> > modems arent... compatible enough for my tastes...
> > ISA cards and ISA mobos are specialty items in todays market (if we
> > are talking new hardware anyway)
>
> Mine are not new. I have a Hayes external that used to be 56K.

used to be?

>
> >> We had one external 56K Sportster that produced garbage at 152000, worked
> >
> > bad programming...cant remember what the bug was in (i think there was
> > an Sregister that had to be changed) but look for an update... (altho
> > you did say 'had' so i guess you dont have it anymore)
> I have not kept good track of the modems.
>
> >
> >
> >> at 57600. Maybe broken somehow. They keep working despite problems (such
> >> as taking the phone offhook as soon as you turnthem on,
> >
> > if they dont take the phone off-hook while turned off but still
> > plugged in check for Sregister settings that may have been changed to
> > cause this.....
>
> Thanks - I will look for a way to set defaults. But it is probably
> damaged by lightning (before we got it).
>
> >
> >> only running at
> >> slow speeds,
> >
> > corrosion, noise, bad programming... incorrect Sregister settings...
> After a lightning storm.
>
> > needing to be told to go offhook before you dial....)
> >
> > that one is definitly software related....
>
> >
> >> Lightning damage?
> >
> > possible on the one that stays off-hook and the slow one but modems
> > usually have lightning protection... if the strike actually makes it
> > through its usually the end (cant pick up the phone at all or
> > completely dead) ... i have a USR 5686D flashed to V92 that was struck
> > by lightning... ATH1 (go off hook) ATX3DT (blind dial) both cause a
> > click noise and "ERROR" .... it never dials out or can even pick up
> > the phone.... AT&T1 (internal analog loopback test) works tho.... so i
> > have a 90$ paperweight that responds to AT commands and nothing
> > else..... ugh... (i can even do flash upgrades on it... not that it
> > does me any good)
>
> We once fixed a 1200 bps modem by replacing diodes. The company sent them
> to us free. That was back when they cost $150 used.

problem with new equipment is that new = less parts and very small
parts... i can barely see surface mount let alone solder it... also
new = in-house parts... not standard/interchangable ones... if i
narrowed it down to a replaceable part it would be nearly impossible
to find a cheap replacement for much better of a price than to just
buy a new modem... and if the flaw is in the single main chip i cant
do anything... welcome to disposable america... honestly tho i never
really liked that modem because of its upload problems (sound
familiar?) but it was good for surfing the web...

>
> >
> > a friend once asked (quite some time ago) if it would be possible to
> > use a modem as an ad-hoc soundcard (play music through its speaker)...
> > i told him i would need some sort of low-level access to the
> > DSP/datapump to do that.... guess i have a candidate for
> > experementation... (USR modems -are- rather extensivly documented)
>
> The speakers are even smaller than the PC speaker, why bother?

just to see if it can be done... the 5686D has an analog volume
control on the side and i could desolder the speaker and put
line-level conversion circutry in its place if i wanted to get serious
with it... but mainly just to see if it could be done...


>
> >
> >>
> >> Thanks for all the helpful info. Where did you learn so much about
> >> modems?
> >
> > "hacker" mentality stuck on dialup since 1997?.... (i read in the
> > jargon file that those with a hacker mentality tend to absorb and
> > retain massive amounts of "useless" data)
> >
> > mainly i just like taking things apart to see how they work and if
> > they can be improved in any way or made to do things 'they arent
> > supposed to do'.... and watching clocks slowly count backwards all the
> > time has lead me to learn everything i can about modems and phone
> > lines.... and ways to try to speed those clocks up.... (and as a
> > side-effect... learning a few other things related to modems/phone
> > lines.... ever called anyone by using the hook switch to dial their
> > number? ;) )
>
> No, how?

pulse dialing... switch your phone to pulse and put the headset to
your ear and dial your own phone number (so as not to bother anyone
else)... if you have good rythm make a mental note of how the 'pop -
pop - pop' noises are spaced... (the technical term for this is 'make
and break ratio')... then... emulating the tempo of pulse dialing to
the best of your abilities its possible to quickly tap the hook switch
to dial phone numbers.... for instance tapping 5 times at the pulse
tempo then waiting for the amount of time normally taken before the
next number in the string is dialed then tapping 5 times again would
dial the number "55"... tap once for 1, twice for 2, and so on... 0 is
dialed using 10 taps... its very tedious but it does work... (expect
to get a few "wrong numbers" and "this number cannot be completed as
dialed" automated messages)

> >>
> >>>>>> Sindi
> >> _______________________________________________
> _______________________________________________
> BasLinux mailing list
> BasLinux AT lists.ibiblio.org
> http://lists.ibiblio.org/mailman/listinfo/baslinux
>


--
--sam
http://mephitus.renamon.org/
"When you've done something right, no one will be sure you've done
anything at all." -- Futurama
--




Archive powered by MHonArc 2.6.24.

Top of Page