Ian & Stuart's Australian Mac: Not for Sale/Another.not.for.sale (Australia).iso/Dr. Doyle/Re AustHell Approval

Text File | 1994-05-01 | 14KB | 260 lines

          Saturday, 30 April 1994 4:09:43 AM
          aus.comms Item
  From:           [e-mail] (David le Comte),internet
  Subject:        Re: AustHell Approval
  To:             aus.comms
>(David Maddison) wrote:

>> I guess I'll be flamed for saying this, but I still fail to appreciate
>> why AustHell can't accept modems which have already been extensively tested
>> and/or approved by certain other regulatory bodies (e.g. FCC or European
>> ones). Our market is too small to require such expensive testing for every
>> single modem or modem variant. As long as the modem meets basic electrical
>> safety requirements and certain other conditions, the performance
>> of the modem should be up to the consumer and marketplace to decide.
>>

>Michael K. Murray (wrote):

>Thats what I thought but when this discussion occured before
>it was claimed that technically our telephone system is different.
>Well thats  how I read the claim. Can anyone who  knows spell this out?

I was the one who attempted to explain the issues, but it seemed to have
no affect on the debate.  I simply assumed that facts were simply
confusing the debate amongst the arguers, as they all ignored them.

Now to put things in perspective, you must understand the loathing
I have for Austel.  Having worked as a technical advicer (unpaid) for
them for four years, I feel I have got to know them. Almost all those
who I have dealt with are honest, sincere, and pleasant people, but
collectively they are "evil".  The issue at hand, though, is not related
in any way to this view.

Firstly, the changes required by US companies to their modems for
Australia are fairly minor.  Having seen though, the incredible (and
unnecessary) convolutions that certain companies have gone thru to have
their phones, and/or PABX systems "Australianised" has
re-inforced my theories of how many bad design engineers there are in
the world. (I have similar views about S/W developers these days, and
not just in MS domains).  Thus I wish to separate the debate, into
design mods done, as against design mods needed.  After all, we are
supposed to be arguing about the latter.

Now the Australian network is indeed different from other networks.  Let
us explain the issues again.  Australia's network is based on the use of
0.4mm TWP cable with long average cable runs from the home to the
exchange (referred to as the CAN - The Customer Access Network).
Because of this, many years ago, telephones were designed to be
"loud", I'll come back to that, and to have complex input impedances
which matched the reactive nature of about 2 kms of 0.4mm cable.
(N.B. for those non elec engineers out there, "complex" is a
mathematical term, and has nothing to do with "difficulty")

By "loud", I mean that the telephones had a high sensitivity to
electrical signals (a high electro to acoustic gain), and produced
large electrical signals from low acoustic signals (a high acoustic
electro gain).  This was to compensate for the loss incurred in the
0.4mm cable.  Note that this loss is dispersive, ie higher frequencies
in the voice band are attenuated more than lower ones, so telephone
equipment compensates by having a pre-emphasis in its gain (ie it
amplifies signals in the 1-2kHz band wrt to lower frequency signals.)
I'll come back to this issue.

The reason that CE (Customer Equipment) have an input impedance which
looks like the complex conjugate of about 2km of cable (a typical cable run),
is to minimise the amount of energy coming back to the exchange from the
cable and the CE, caused by transmission from the CE.  Remember that the
cable contains both transmit and recieve signals. The exchange has an
analogue 2 to 4 wire converter (as does the CE), which produces its
receive signal by subtracting a scaled and phase shifted version of the
transmit signal from the signal detected on the line. The phase shift and
the scaling are set by assumptions of the average complex impedance
expected on the cable and as modified by the level of DC current
flowing.  Thus certain assumptions are made as to the impedance seen
from the current flowing.  Now the signal on the cable will contain not
just the signal transmitted, and the signal sent from the CE, it will
also contain the signal echoed from the CE.  This latter component will
not be balanced out by the circuit described in this paragraph.

This circuit is repeated inside ones CE, and it will expect that their
will be minimal echo from the exchange, and from the CE at the other end
of the connection, so as to improve its hybrid balance performance.
This, though, is important only to that CE (ie customer), and it can be
argued, is their concern.  The input impedance of their equipment,
though, affects the performance of the exchange, the other item of CE in
the connection, and the performance of the link.  In the case of the CE
which is paying for the connection, they might expect a certain minimal
level of performance from the item of CE they are connected to.  hence
we have a return loss test (as do all authorities, including in a
minimal way the FCC - I'll come back to this).  This test differs from
country to country because the average cable lengths, and cable
diameters vary.  Europe eg has short runs of 0.5mm cable as does NZ.
Canada and the US mainly use 0.4mm cable (26AWG) as do Aust.

Note that modern modem protocols expect that the DSPs used to implement
them, perform echo cancellation and adaptive equalisation.  This helps
them (ie it improves their signal to noise ratio), but doesnt help the
hybrid balance circuit in the exchange which doesn't use such
techniques.  (A mite too expensive for telephone lines after all)

A second issue is the relative importance of having a satisfactory
return loss.  In Australia, it is more important than most countries,
because the decision was to have a "loud" network.  This means that the
average signals transmitted from CE equipment are higher than almost
anywhere in the world.  This means that on short run connections (common
in the centre of the city eg), signal levels arriving at the exchange
(including from echos) are much higher than in other countries, reducing
the performance of the exchange's hybrid balance network, and hence the
performance of the connection overall.  The decision to have a loud
network, is historic, related to the desire over many decades to give
rural communities services that would normally be given to urban
dwellers in most economies first.  Be that as it may, with 20-30 million or
more items of CE designed around similar specifications, it is
impossible to consider significant change (actually due to my lobbying,
the overall loudness of new CE has dropped by 3dB).  There are important
reasons for the exchange to have a good hybrid balance performance.
These are complex, but relate to open-loop stability and saturation of
digital or FDM channels (Comms engineers should know about these things,
but they are too complex (difficult) to explain here).

This then constitutes one of the changes that US based equipment must
make.  They must swap a 600 ohm resistor for a network of an 820 ohm
resistor, a 220 ohm resistor, and a 120nF capacitor.  Thats it!!!

An issue of some importance in Australia, is the amount of currrent that
CE must pull when they first go off-hook, in order to "pull-in"
relays, on old relay based (Step by Step) exchanges.  We still have
many of these (the US has an even higher % of these than we do!).  This
has proved to be a problem for some European CE in the past.  It is
not a problem, though, for US manufacturers, because their requirements
under RS-470 (I'll come back to this) are more stringent than here.
Unfortunately the Aust standards are often misinterpreted in this area,
and I wouldn't be surprised that modem suppliers were guilty in this
respect.  The Aust spec indicates that one only needs to draw this high
level for at least 300ms, after that, one may draw less (or not
bother!).  Thus many overseas suppliers have unnecessarily put in
"kick-start" circuits to do this.  This clause is only there, so that
"intelligent" phones, which need more voltage so as to power
microprocessors. These may need 5V or so to work.  By drawing less
current, the voltage on the phone can increase to a level whereby 5V can
be stabilised for these micros.  It was a concession, not a
requirement!

Australia has an insulation resistance requirement of 1 Megohm.  this is
extraordinarily generous by world standards. The US requirement(FCC) of 6
Megohm means that they have no problem.  Australia requires that dial
pulse overshoots be limited to 220V.  The voltage is higher in the US,
all this means is that they need to change the spec of the overvoltage
device (a MOV or transil).  This involves no more than looking up the
suppliers catalogue and choosing a different (pin compatible!) part.

The US UL requirements for CE require the addition of micro fuses on
each leg of the connection.  This poses no problems here, but is
unnecessary here.

The most sticky point for US manufacturers generally is their relaxed
(by world standards) requirements for "galavanic isolation".  This is
the safety test which tries to protect consumers from common (not all)
lightning strikes to the mains or to telephone lines.  CE are usually
considered to be what is referred to as Category A equipment.  This
equipment may expect voltage surges to 3.5kV once a year or so.  As a
consequence most standards bodies expect there to be isolation from any
"gerfingerpoken" bits to mains or telephone lines of at least 3.5kV.
For some reason the USA has decided to set their level at 1.5kV, why I
don't know.  Australia has adopted European standards in this regard, as
they tend to do in most things.  This implies that US companies,
intending to export, have to change their mains and line transformers to
bigger (larger air gap) transformers.  This applies to all US equipment
BTW, not just comms equipment.  There is a requirement that handsets to
phones (nothing to do with modems!) must be isolated from the line or
mains by 7kV (The figure is 6kV in the US).  The garbage references to
this issue in this newsgroup have been particularly galling!.

There are a few other things that US manufacturers need to change.
First they need to default decadic dialling to the 66/33 milliseconds
break to make ratio used by almost all countries except USA/Canada, and
Germany which use 60/40ms.  Secondly they often find that STD pips stuff
up their data pumps on Auto Answer, so they need to change the S/W to
hold their data pumps back for a few seconds after auto-answering.  They
have to default the transmit level to -10dBm (the Euro and Aust std for
data transmit levels: note that this is to minimise cross talk in FDM
systems), and default to not auto-answering the phone for at
least 2 seconds.  (A silly Telecom Aust requirement that Austel accepted!)

There are probably a few other minor things such as Aust power plugs,
and voltages, but none are significant.

I mentioned earlier a spec called RS-470.  This was the original spec
for phones and the like, issued by the "evil" A.T&T.  When the FCC took
over the regulatory role, they abandoned RS 470, and produced their own
spec.  This spec was very high on safety (but not on isolation?), and
had little or no requirements on transmission.  Whilst they have specs on
maximum voltages, and signal powers to line from CE they dont have any
specs on electro-acoustic efficiency, frequency response of handsets,
not even dialling (decadic or dtmf) performance.  They have a fairly
loose spec on return loss, but it covers more than the vf band.  They
were regulating only the absolute requirements for connection of CE to
the network.  They had no interest in end to end issues, such as
loudness eg.

The early FCC specs, being so generous, caused a flood of imports of
dubious quality.  Asian suppliers could bring equipment in, designed for
other networks, without modification.  Since then, they have stiffened
the safety requirements, by adding UL safety ratings as well.  Thus new
phones in the USA cannot use paper-phenolic boards, they must use FR-4
fibre-glass or similar.  (This has no relevance to modems, which
typically use multi-layer (FR-4) boards these days so as to improve the
board density and minimise RFI problems, a requirement which is
virtually identical between the US and Aus).

In the absence of any sensible specs from the FCC, suppliers generally
conform to the specs in RS-470 voluntarily.

These so far have been the facts of the matter.

The salient points are:

    - For reasonable reasons each country has a duty
    as well as a right to set analogue communications standards
    because international standards DO NOT EXIST for historical and
    practical reasons.  They do exist for some new RF and digital
    standards, hence the automatic acceptance of some overseas tested
    equipment in these areas.  Note that many CCITT recommendations
    have a plethora of choices, so eg OSI on one piece of gear
    may not interoperate with OSI on another; ISDN has similar
    problems.

    - Australia has set some standards, most of which are
    reasonable, some of which are stupid (but usually innocuos)

    - The requirements for the approval of modems and faxes in
    Australia, are probably the easiest in the world.  Certainly
    easier and far less expensive than those in the US.

    - Passing US specs which are based at no more than safety, and
    guaranteeing network integrity (not interoperability!), is
    irrelevant to satisfying any performance issues relating to the
    Aust network, much less satisfying the network and end to end
    interoperability issues deemed important in Aust (and all other
    specs in the world).


            david le comte

            [e-mail]

--------
Newsgroups: aus.comms
Path: macconn!lsupoz.apana.org.au!spectrum.apana.org.au!warrane.connect.com.au!yarrina.connect.com.au!news.uwa.edu.au!munnari.oz.au!newshost.anu.edu.au!sserve!usage!metro!extro!davelec
From: [e-mail] (David le Comte)
Subject: Re: AustHell Approval
Message-ID: <davelec.767678983@extro>
Sender: [e-mail]
Nntp-Posting-Host: extro.ucc.su.oz.au
Organization: Information Services, Sydney University, Sydney, NSW, Australia
References: <[e-mail]> <2pndem$[e-mail]> <[e-mail]>
Date: Sat, 30 Apr 1994 04:09:43 GMT
Lines: 240