🚀 go-pugleaf

In article <VUVrVCALjRjyEwbp@tuppambr.demon.co.uk>,
Mike Tuppen  <mgtuppen@iee.org> wrote:

...
 
> I noticed a lower connection level with the Ascends and wrote in about
> it. I received nothing but scorn from people in this group - "another
> Newbie who can't configure his system" - and Demon seemed not to take
> notice. No one accepted that the Ascends were at fault.
> 
> I measured line levels between USR and Ascend connections and found the
> final level was 6dB below the USR level.
 
> I sent reports and Demon (Jim Seagrave(?)) eventually seemed to take an
> interest and a few weeks later the levels seemed to go up on the Ascends
> and I had no line drops or failures to connect.

Segrave actually. Your report was not ignored, in fact it went to
Ascend within a day or so of your posting it. After mcuh toing and
froing, Ascend asserted that they were already at the BABT maximum
transmit level. Later releases of software provide a way of adjusting
the transmit level, but only downwards. To the best of our knowledge,
the Ascend transmit levels have never changed in any of their
releases - they have certainly never said so.
 
> I still have few problems but have noticed that whenever I get an Ascend
> the transfer speed can be marginally (insignificantly!) slower. Looking
> at the modem ATI6 report shows that on average I get a 28/28K link with
> 3 BLERs and no LINK NAKs for about 3MByte of data. However with an
> ascend connection the modem reports connection speed of 26 (lowest
> 22K)/28K 30 BLERs and up to 4 LINK NAKs for similar amount of data. 
> 
> I realise that this could be due to line noise or that USR Sportster
> internal modems are faulty, or that I haven't tweaked the modem init
> string sufficiently. I know I could junk the sportster and buy a courier
> instead and all would be well *but*...
> 
> ... There is *no* problem USR to USR, increasing output levels on the
> Ascends improved performance a year ago, USRs at Demon still output a
> greater level than Ascends, so how can "requirements of the BSI for such
> equipemnt" not apply to the USR. USR must be Approved by BT.
> 
> Why prevents the output of the Ascends being increased to the same level
> as the USRs?

According to Ascend, the conditions for approval. They claim never to
have changed the transmit level. It is possible that in fact USRs do
have a higher output level but, if so, Ascend maintains that they
(USR) would be out of spec.

The modem statistics you quote for BLERs and what I take to be
indications of retrains does suggest that the combination of your
modem, your telephone line and the Ascends is working less than
optimally. We can't increase the signal level at our end. You may be
able to get BT to remove AGC and set your line level higher. You may
be able to improve your connections by trying one of the various init
string modifications which attempt to over-ride the known errors in
negotiating connections between Sportsters and Ascends. 

I can't think of any other useful suggestions that don't cost money up
front - i.e. changing modems. I am not in a position to say whether
the problem lies entirely with the Sportsters, the Ascends or is
divided between them - I have seen various statements by seemingly
knowledgable people backing each of the previous explanations.