[Asterisk-Users] Digium FXO Interfaces don't support groundstart???

[Frank Cofer]

He who knows not that he knows not, knows little; he who knows that he knows
not, knows a lot.

1.  Glare is the simultaneous seizure of a common facility using two
separate independent universes whereby a decision at one end is not
immediately know to the other.  As mentioned previously, because any two
points that are not the same in space are separated in distance, they are
necessarily separated in time.  The possibility of seizure of a given
circuit from both ends simultaneously is minimized by careful design, but it
is never reduced to zero.

Thus the original statement, "...ground start eliminates glare..." is simply
false.

You can tell it is false if you have any rudimentary knowledge of physics,
which (currently) holds that the maximum speed any information can travel is
the speed of light.  Electrical circuits are much slower.  Thus the means to
communicate an independent decision made at one end, that is separated by a
distance, is also necessarily separated by time to the other end.  Due to
this propagation delay, "glare" cannot be eliminated on any two-way circuit.
This does not take craft credentials to recognize, it merely requires a
modicum of logic and perception.

 2.  Without maligning your comparative lack of telephony experience of a
scant ten years, it would be useful to bring up how the central office or
the PBX "know" what the other end is doing.
On older central offices that had line and cut off relays, ground start
service was effected by simply removing the tip ground of a bifalor (two in
hand) wound line relay.  There was a line relay for every line in the
central office.  Doing this in effect preventing it from operating unless
the ring conductor (the -48vdc power feed to the line) was grounded, closing
the loop would no longer do.  Until either the line relay or the cutoff
relay (busy at central office) was operated, the central office was
"ignorant" of the change of status of the distant end.  So to operate a line
relay if the ground was removed from it, the PBX had to place an earth
ground on the ring side of the line for a short interval that would not only
be safely more than the time to cause the line relay to operate, but also to
hold it there until a circuit had been established and the associated cutoff
relay operated by the central office to hold the connection.  Typically this
occurs in 20-80ms but, to be safe, the time for the PBX to ground the ring
is set nominally to 150ms or so, or until ground is seen on the tip
conductor at the PBX end.  The latter is not seen until the call set up is
nearly complete at the central office, since it is first extended from the
inter- or intra-office trunk through the cut off relay.

Since a subscriber line relay, modified as mentoned above for ground start,
has only one half the ampere-turns of current to operate as loop start, its
speed of operation is decreased greatly and the mark of busy to the central
office (from the time the line is seized by the PBX at the distant end until
such time the cut off relay operates) is markedly increased.  From (1),
above, it logically and inescapably follows that the time that the central
office "knows" that the PBX has seized the line is increased over that of
loop start because, in the latter, both coils are aiding and the relay
operates much faster (its speed of operation is a function of the flux
density which is a function of the ampere turns).

Thus, for a seizure from the PBX to the central office, "glare" (the time of
not "knowing" what the other end is doing) is increased when ground start is
used.  Though I want bother to describe it here, the same is true of modern
electronic offices using any form of earth ground signaling because they are
arranged to emulate the older offices above.

2.  Now for the other end, for a call from the central office via a two-way
trunk to a PBX, the same applies.  It is true that some older PBX's
(Stromberg-Carlson X-Y and some really old SXS versions come to mind) were
somewhat brain dead and could not recognize an incoming connect from the
central office using loop supervision until ringing was applied.  For these
PBXs, they relied merely on the presence of ringing (AC on tip to ring > 65
volts and wait for a sufficient time to ignore transient spikes) as an
incoming signal.  This is the fault of using ringdown detection instead of
change of state for an incoming connect when loop supervision is employed.
Severely aggravating this, many older central offices did not have
multi-phased "immediate" ring, thus the combination of "ring detect" and
non-immediate ringing could delay up to four seconds (due to the ringing
cadence of two seconds on and four seconds off) before the associated PBX
would "know" the line was seized by the central office.  For ground start,
these older PBX's were arranged to look for presence of ground on the tip
(which was normally absent when the facility was idle in the central office
except for call setup).

However, few PBX's (unless they are ancient or sloppily configured by the
installer) wait for ringing to be applied as an incoming signal in loop
start anymore (they recognize change in state immediately when the
superimposed DC is applied) and even for those that do nearly all electronic
offices have immediate ring.  But even in these cases, the time that the
line is seized by the central office until such time the PBX "knows" it has
happened is discrete and glare nevertheless occurs, be it looking for ground
on the tip (ground start) or looking for a change in state (loop start).

3.  There are numerous schemes to inspect the line for latent seizure as the
call setup process goes on to try to minimize glare.  The central office
routinely does this after call setup is mostly complete, as a final check
before 2nd trial failure on all lines (both loop and ground start).  To my
knowledge, this is not universally known as "pre-seize" in your limited
vernacular and it certainly is not restricted to ground start lines, though
it may be described as such on the machines that you have worked on.

4.  Though your other arguments are patently irrelevant, I will address a
few of them.  Wink supervision is independent of loops (it can be effected
by E&M, reverse battery, loop, etc.) and merely refers to a brief change of
state (from on hook to off hook to on hook); hence "wink"  at the remote end
after the call has been set up.  It is nominally 130ms.  It is primarily
used as a channelized signal between common control switches to indicate
that the receiving end is ready to receive digits.  The first wink causes
the distant sender to send the called number and a second wink (optionally
employed) causes the distant end to send the calling number for use for
Automatic Number Identification (ANI).  Other than the similarity of the
time interval, it has as much to do with the discussion of ground start as a
pig knows about Sunday.

It is also somewhat academic, because most central offices and a growing
number of PBXs that provide Direct Inward Dialing (DID) now use some common
channel form of supervision (e. g., SS7 or ISDN D channel, though it is
still widely employed on channelized DID for some switches (Mitel).

It is important to understand that the DC supervision labels used by
two-wire and four-wire switches were carried over as a label for channel
banks, much in the same way the old cord board "tip", "ring" and "sleeve"
now carries the labels for transmission lines.  Since you only have ten
years of experience, you are probably unaware of this.  These signals were
converted at channel banks and encoded as "A" and "B" signaling bits in the
digital stream; their meaning is simply binary at the lowest level, but
their interpretation differs based on the signaling schemes of the attached
equipment.    "E&M" signaling is a means to communicate supervision is a
very old (>50 years) outband method of signaling (the "E" means "ear piece"
or receive and the "M" means "mouthpiece" or transmit).  Originally, battery
was placed on the "M" lead to transmit an offhook, and ground received on
the "E" lead was interpreted as an offhook from the remote end.  It was
devised on open wire before there was any carrier.  E&M has evolved into
several so-called "types" whereby same source battery and ground is used
(each connecting unit provides its own battery and ground reference and the
opposing end merely provides an isolated relay contact).  Ironically, the
E&M Type II and Type III were developed to avoid ground loops between
equipment, which is very undesirable for digital equipment, and is again
discussed below.

My point that the carry over label used in channel bank terminology as
"ground start"  has advantage because it start with A and B as 11 (an idle
line) and then changes only one bit at a time (Grey code) which relieves it
from ambiguity.  (Normal binary count going from 01 to 10 must pass through
either 11 or 00 before it gets to 10).  The supervision used on T1 should
always be "ground start" because of that reason.  (There is no delay in
connect, however, if loop or E&M supervision is used instead).  Your comment
that Feature Group D somehow made ground start available is mistaken and
misleading.  Ground start has been available for ages and was introduced
with No. 1 SXS PBX's over fifty years ago and also was used with N, ON as
well as with T carrier beginning in the early 1950s.

I am unfamiliar with your "funky" description (which seems to lack any
precision) as to what it means but it may have something to do with
connotated ignorance.

6.  The fact remains that any analog balanced pair transmission line is
severely compromised by use of earth ground connections and its use should
be avoided.   One rarely hears of a 2500 set being struck by lightning and
it is no accident.  It is immune from longitudinal voltages since it is
entirely transformer coupled and is totally isolated from earth ground.
This also applies to circuit cards.  As soon as ground start is added as a
feature, earth ground must be extended to the equipment and through to the
connecting circuits.  And as soon as that is done, all of the circuits are
exposed to longitudinal voltages to that earth ground from the transmission
line.  Believe me, this is a very bad idea.  It is also unnecessary.

7.  Digium is not alone in not offering ground start.  Avaya, Nortel, and
other major manufactures of key and PBX's have eliminated ground start other
than accommodating diehards on some older equipment.   You might argue that
their reason is more value engineering rather than to get away from FEMF
failures that are intrinsic to any such design.  Or, you may argue that they
did so because they no longer care about the benefit to the central office
any more, or because they are no longer controlled by the Bell System and
can do what's singly best for their customer.  Or, you might argue that Bell
Labs (Avaya) was unaware of the dangers of glare that are present when you
don't have ground start as simply lacks the wisdom of your ten years
experience.  I don't know what there reasons were for abandoning ground
start, but I think the reasons I have given above should certainly suffice.

8.  How do you prevent (or minimize) glare on a two way circuit via fiber
optic, since there is no earth ground?  Your answer, if your head firmly is
not implanted firmly in the sand, should make you come to terms that glare
is independent of ground and, indeed, is independent of any supervision
method.  It is merely a function of time.  No amount of signaling schemes
can eliminate it for two-way circuits.  One can only reduce it by speeding
up the communication of the state at each end to the other end or lessening
the probability that the event will occur by changing the hunt order.  Even
ISDN PRI using common channel signaling is still susceptible to glare on
2-way circuits and still employs inverse hunt order to minimize glare.

9.  I have a hard time correlating anything but an overloaded or
misconfigured trunk group that would be responsible for the severe problems
given as horror examples you and the previous writer have related, since any
glare interval is relatively short if the system is properly configured
(inverse hunting sequence) and the probability that a circuit would be
simultaneously seized would be relatively infrequent unless it is severely
underdimensioned.  How large was the trunk group you had, what was the
traffic carried and what was the engineered grade of service? Did you check
the time of disconnect of the associated central office trunk?   What was
the guard interval employed?  Did you have immediate ring from the central
office?  Did you use change of state for an incoming call with loop
supervison?  Did you have reverse battery disconnect supervison from the CO?

Finally, there are two times one should mention their credentials:  shortly
before they're hired and shortly before they should be fired.  Lawyers argue
credentials, scientists argue verifiable facts. Those that mention their
credentials to persuade, don't even believe themselves.

Tone down your rhetoric, forget your credentials, and think.

----- Original Message ----- 
From: "Bruce Ferrell" <bferrell at baywinds.org>
To: <asterisk-users at lists.digium.com>
Sent: Saturday, July 31, 2004 1:27 PM
Subject: Re: [Asterisk-Users] Digium FXO Interfaces don't support
groundstart???


> First of all I use this in humor:
>
> Frank you ignorant slut!
>
>
> I have to disagree on your analysis.  I worked in telephone COs (DMS250,
> Stromberg/Carlson) and with PBXes for over a decade.  Glare can and is
> controlled by ground start signaling.  It does so because the ground is
> tested for (or supposed to be) prior to dialing.  It's called the
> pre-seize condition.  On a T1 using robbed bit signaling, tip and ring
> conditions are converted into A/B signaling states in the channel
> modules of a channel bank.
>
> Ground start was the prefered signaling system for what was called
> Feature Group D trunks between Other Common Carriers and the RBOCs.
> Before FGD was available, we used loop start. We had incoming and
> outgoing trunk groups, hence no glare... Needless to say expensive.
> Because FGD had ground start, to cut interconnect costs, we went there
> as soon as it was made available.
>
> The 150ms pulse you described is called wink start, which was funky.  I
> most commonly say it on systems using E&M signaling.  Gawd I hated those!
>
> YA know, the asterisk list has been for a lot of walks down memory lane :)
>
> Frank Cofer wrote:
> > Glare cannot be prevented on two way trunks (it is physically impossible
> > because the two ends are separated in distance and therefore separated
in
> > time and any independent decision to use it at one end is never seen
> > instantly at the other end).
> >
> > Ground start does not decrease glare at all (it actually increases it)
and
> > use of "ground start to eliminate glare" is a common myth.  This is
because
> > use of ground start (which uses only one side of the pair to earth
ground to
> > "start" a request for service) increases the time to mark a central
office
> > line busy when it is seized from the Customer Premise Equipment (CPE),
owing
> > to its clunky signaling (150ms earth ground on the ring of the line) and
the
> > fact that it uses only one half of the current to start the line as loop
> > start.   Since it increases the time to signal the distant end, it
increases
> > glare.
> >
> > Its only benefit is to the central office because it stops a second
seizure
> > to the central office when a call disconnects from the central office
end
> > first, which would otherwise find a request (loop) as soon as the
disconnect
> > was effected.  This is why "ground start" was introduced by the Bell
System
> > (when they owned both the PBX and the CO) since it would reduce the
attempt
> > load on the central office from large business users by 25% or more
saving a
> > lot of central office gear for a relatively small expenditure on the PBX
> > end.  Ground start has some ugly drawbacks, since it reduces signaling
> > range, requires the normally isolated floating pair to be referenced to
> > earth ground (which exposes the circuit to longitudinal spikes, noise
and
> > lightning) and requires the circuit to be muted during the imbalanced
> > condition that occurs when the ring conductor is momentarily grounded to
> > draw dial tone.  Digium is right to leave it out.   Most other informed,
> > modern manufacturers do likewise.
> >
> > "Ground start" signaling referred to in T1 (which is an absurd label
since
> > there is no ground placed on a T1) is really after the Grey Code (only
one
> > signaling bit transitions at a time) and has nothing to do with glare or
> > ground start signaling and is just a carry over label.
> >
> > Glare can be reduced by changing the hunt order from either end and to
> > employ faster signaling.  The former method decreases the likelihood
that
> > both ends will compete for the circuit at the same time and the latter
> > reduces the window that a commitment has been made at one end and is
still
> > not known by the other end.  Typically, the CO is set to hunt ascending
and
> > the CPE descending and this is still employed even in ISDN circuits.
This
> > is a "terminal hunt" and NOT a "round robin" hunting sequence.  If you
want
> > to absolutely eliminate glare, use one way (incoming/outgoing only)
> > circuits.  I believe asterisk has a feature to set the hunt order
> > preference.
> >
> > The disconnect problems you experienced with your Agilent PBX may be
more
> > likely related to the "guard interval" that a circuit is left alone at
your
> > end after it is used.  Though "ground start" will appear to fix it,
there
> > are some issues of CO message rate three way calling that have caused
grief
> > (the CO interprets the next call as a flash for a three way call and
holds
> > the circuit rather than disconnecting it).  This phenomena may have been
> > misdiagnosed as glare, since the message unit 3-way calling was imposed
as a
> > default feature in certain jurisdictions.  Increasing the guard interval
to
> > 2 or 3 seconds will suffice, or specify to the carrier that the 3-way
> > calling is to be denied for your lines.
> >
> > Hope this helps.
> >
> > ----- Original Message ----- 
> > From: <ghost at silcon.com>
> > To: <asterisk-users at lists.digium.com>
> > Sent: Thursday, July 08, 2004 4:55 AM
> > Subject: [Asterisk-Users] Digium FXO Interfaces don't support
groundstart???
> >
> >
> >
> >>Hi All,
> >>
> >>I was surprised to be told by a Digium support person today that
Digium's
> >>FXO interfaces (X100P, TDM400P FXO modules) don't support groundstart
> >>signalling.  This surprises me because as far as I know in a typical PBX
> >>configuration with analog trunk lines, groundstart signalling is the
only
> >>way to prevent Glare.
> >>
> >>I just purchased two TDM400P's for a system I'm building to replace our
> >>office PBX (Altigen).   Since there are no statements anywhere on
Digium's
> >>website about lack of groundstart support (Actually, to the contrary
they
> >>boast about all the signalling support in their sales slick), I now need
> >>to decide if I want to return the products and switch to a T1 / channel
> >>bank configuration.
> >>
> >>I remember when we setup our current Altigen PBX, we had problems with
> >>glare and disconnect detection and so I went through the process of
> >>figuring out what was going on and learning about groundstart.  After we
> >>switched to groundstart everything worked great.
> >>
> >>In a high use system, it's highly likely that a trunk will experience
> >>glare, which is annoying for incoming callers and system users.   I'm
just
> >>a bit baffled as to why Digium wouldn't support groundstart on cards
> >>designed to be PBX trunk lines.
> >>
> >>Someone please tell me I'm missing something.
> >>
> >>Mark
> >>
> >>
> >>
> >>_______________________________________________
> >>Asterisk-Users mailing list
> >>Asterisk-Users at lists.digium.com
> >>http://lists.digium.com/mailman/listinfo/asterisk-users
> >>To UNSUBSCRIBE or update options visit:
> >>   http://lists.digium.com/mailman/listinfo/asterisk-users
> >>
> >
> >
> > _______________________________________________
> > Asterisk-Users mailing list
> > Asterisk-Users at lists.digium.com
> > http://lists.digium.com/mailman/listinfo/asterisk-users
> > To UNSUBSCRIBE or update options visit:
> >    http://lists.digium.com/mailman/listinfo/asterisk-users
> >
>
> _______________________________________________
> Asterisk-Users mailing list
> Asterisk-Users at lists.digium.com
> http://lists.digium.com/mailman/listinfo/asterisk-users
> To UNSUBSCRIBE or update options visit:
>    http://lists.digium.com/mailman/listinfo/asterisk-users
>