PE>> I haven't been playing with my init string at all.  "Left
compression
 PE>> off" must have inadvertently left V42+V24bis off and got
MNP4 instead
 PE>> on your end.

 RS> Nope, that should have given a V42 call, LAPM. Thats attempted first and
 RS> then MNP4 is accepted as a fallback if V42 isnt allowed.

 PE>> This morning you connected at a multitude of speeds (why?)

 RS> Thats just how it happened, not me doing anything. I called, got either
 RS> a 12K or 9600 connect, then mostly very soon after gross problems like
 RS> continuous line errors. The usual cause of that effect is session was
 RS> completely stuffed, one end thinking it got V42, the other no error
 RS> correction at all.

Ah sorry, I didn't mean speed, I meant error correction.  ALL of the
connects were done at 14400 my end.  Some of them MNP4, some V42, some
V42bis some no error correction.

 RS> Same today although the symptoms were a little different. Got a 14K
 RS> connect, then continuous errors and the line hung up. Called next time,
 RS> fine, or atleast until the QWK download with the usual 800 cps thruput.

 PE>> and finally ended up at V42bis and my MR light started flashing again.
 PE>> Also the CTS light.  I have a theory that says that because the CTS
 PE>> light is flashing, OS/2 is defintely sending data fast enough (too
 PE>> fast, in fact).

 RS> You cant be completely sure of that. Sure it indicates that OS/2 is
 RS> stopping sending to the modem when its told to. You cant be sure tho
 RS> that it is actually immediately sending to the modem when its told to
 RS> start again tho. Its unlikely to be not doing it, but you cant be
 RS> completely sure just from the CTS.

 RS> Thats the main reason why it would be worth a local call with real
 RS> booted DOS to completely eliminate OS/2 and all its software and
 RS> config from the equation. If a real booted DOS with Telix say gets
 RS> poor thruput out of your system, you can be totally sure its got to be
 RS> the modem or some hardware detail like the serial port etc.

What will DOS + Telix prove?  I am not having a problem with other people. 
If I'm not having problems sending a file to someone, then I use Telix and
I don't have a problem, that doesn't say anything. If I use Telix and I do
have a problem, it means I have DOS problems. Neither of these results are
useful.  If I was to do a test with you calling me and me running Telix,
then that would possibly be a useful test.  However, not much point doing
that if we haven't got consistent results from the OS/2 connects (too many
things happening at the same time).

 PE>> Now the only thing that could be happening is that my PC is ignoring
 PE>> the dropping of CTS. When the MR light was flashing this morning (in
 PE>> fact most times), I didn't get any Zmodem errors.

 RS> Thats most unlikely because that will cause continuous dumping of data
 RS> into the modem when the modem cant accept it. Giving a gross problem
 RS> very quickly since the speed of the data over the cable is rather
 RS> faster than the modem can put the data up the phone line. Guaranteed
 RS> to give lots of Zmodem errors and the last few days have seen
 RS> absolutely none.

So PC ignoring CTS is out!

 PE>> BTW, my Init string is based on the T. Rodyhouse authorized settings.
 PE>> I used that as the basis at any rate.

 RS> When you say based on, why did you change what you did change from his ?

I disagree with some of the things, such as switching echo off.

 PE>> Since I have used two Spirits, and other people are working OK, this
 PE>> means that the error correction going on either has to be a terrible
 PE>> STD line (thus genuine error correction is happening), or I have my
 PE>> init string set wrong, and it isn't allowing enough time for error
 PE>> recovery.

 RS> No, you are assuming that MR blinking is really line errors and error
 RS> recovery like it supposed to be. But the Spirit will get into that MR
 RS> blinking state with poor thruput in other than line error situations.
 RS> Thats the fundamental point of the problem with Spirit and was
 RS> precisely the problem which has plagued the Spirit owners. It cant
 RS> really be line errors because most people were getting that symptom.
 RS> The cant all have bad lines, and many have a non Spirit modem work
 RS> fine on their line too. It really is a Spirit design problem, the C70
 RS> cut and new cable was supposed to fix that but seems like maybe it
 RS> hasnt.

First of all, Mike told me that the MR flashing was a well-defined problem
which happened on uploads and downloads locally and STD.  I got it fixed
anyway.  Then I got the whole modem replaced with a brand new one which had
the earth problem fixed.  He told me it wouldn't make any difference, and
he was right.  This makes me think I didn't have the problem you are
talking about in the first place.  Two fixes since then must reduce the
odds considerably on me having the problem with MR flashing that was a
Spirit problem.

BTW, when I called my local board today, sending a 700k file, it first of
all went fine, up to about 200k, then had a lot of MR light flashing for a
minute or two, then it resumed all quiet.  The end result was 99%
throughput.  I'd better name these people.  I feed off Dave Hatch, and he
has said it is OK for me to send these large files to him for test
purposes.

 PE>> A lot of my thinking has involved delays in STD, but they don't
 PE>> really exist, do they? I mean, you calling STD is IDENTICAL to
 PE>> someone calling locally, right?

 RS> Yeah, delay wise there is no problem with STD at all.

Since I am having problems with you which are quite different to others,
and consistently like that, I would like to propose a theory that my
exchange is having problems with STD calls.  The problem is that the
exchange is *generating* noise, which my modem picks up and attempts to
correct.  Would this cover all observed phenomenon.

 PE>> And I am now uploading fine (since changing com port).
 PE>> Admittedly I am initiating call.

 RS> Yeah, thats an important point, there are certainly some well
 RS> documented cases with some modems where it matters who calls. So its
 RS> not surprising that you get different behaviour.

This is what makes the Telix test too difficult.  It is difficult for me to
ask Dave Hatch to do this for me, as I am already hassling him about flow
of GTNET_SCIENCE, HS_MODEMS and procuring SKEPTIC and BLAKES_7 and a TIC
file area.  I can do the test with Paul Markham, but he is using a Spirit,
so that may not be a fair test.  Also, his Spirit is unmodified and using
the old roms.

 RS> I get the same thing on calls to you, my uploads go fine. They are
 RS> much shorter so its harder to see the thruput achieve but the look
 RS> fine in comparison with the dud downloads.

Yeah, I wouldn't bother testing that.  I have been downloading stuff for
quite a while and never experienced problems.  Uploading has always been an
untested thing for me.

 PE>> Do you agree that the problem HAS to be either the Spirit or the
 PE>> phone line?

 RS> I would still do the booted DOS test to be completely sure,
 RS> particularly as its quite an easy test to do. But I agree there is
 RS> little chance its anything but the modem.

And with 3 different modems (almost like that anyway) it has to be either
consistently wrong parameters, or a problem with my exchange. I have
another phone line, which is connected to a different exchange (I believe)
and I also believe it is an AXEE, so that can be another test down the
track a bit.  If that was the case, I can ask Telecom to move me to a
different exchange, and my number will change too (not a significant
problem by any stretch of the imagination).

 RS> I think the phone line is unlikely to be the problem, mainly because
 RS> the detail of the behaviour doesnt smell like that. Its quite rare for
 RS> example to see the situation we have seen on occasion with a 800 cps
 RS> thruput all of a sudden pick up thru a download to a full speed one
 RS> when the phone line is at fault.

The hardest thing is if it is a genuine fault in all 3 of my Spirits,
because there isn't much I can do about it.  I could probably swap with
Paul Markham's modem for a while, but that would be the 4th modem, not the
2nd.  I'm hoping some of these S registers that I had to change to get in
line with Trev's settings will do the trick.  From your latest message it
looks like maybe and maybe.  BFN!

Paul

--- GoldED 2.40