Dear Treggers,

On the topic of taxometer filters for Germany, I have some additional
technical comments to make:

The problem as I see it lies mainly with an active DC holding circuit,
where used, leaving its linear region during the taxometer pulse peaks.

A DC holding circuit is needed whenever a "dry" isolation transformer is
used. (A transformer that cannot pass the line current used for signaling
an off-hook status is known as a "dry" transformer, whereas one that
performs the additional function of passing line current is known as a
"wet" transformer.)

DC holding circuits may be active or passive. In my experience, active
circuits are more commonly used, and these are designed to simulate the
behavior of an inductor, by presenting a low, defined resistance to DC,
while appearing as a very high impedance (tens of kilohms or more) to AC
signals in the voice band. The active circuit is biased so that it will not
enter cut-off or saturation for the largest AC levels expected on the
telephone line in the off-hook state --  maximum DTMF signaling power, or
maximum superimposed transmitted modem signal plus received modem signal on
the line. However, the taxometer pulses as defined by BAPT are close to 10
volts RMS, which is about an order of magnitude greater than voice-band
signals, and the active circuit cannot be biased to cope with these.
Therefore, as I see it, the only options are:
a. Use a wet transformer - generally a problem because it is hard to
achieve high-speed data modem performance (easy for a fax modem) with a wet
transformer;
b. Use a passive holding circuit - I haven't explored this option but am
aware that designs are available that implement this;
c. Use an active current hold circuit, but add a taxometer pulse filter to
attenuate the 16 kHz sine wave bursts before they reach the holding
circuit, to a level that is sufficiently low that the holding circuit will
not depart form its linear region, when the maximum taxometer signal is
superimposed on the maximum voice-band signal level.

It appears to me that to be quite certain of meeting the requirements for
worst-case component values (5% inductors and capacitors in the filter) and
worst-case 16 kHz limits from the telephone exchange, a 2-stage filter is
necessary. Since the first stage has the inductor in series with the line
(to ensure that the subscriber's equipment presents an appropriately high
impedance to the network at the taxometer frequency), this part needs to
have the correct inductance while passing DC levels of up to around 60 mA,
which is the maximum nominal line current in the German network (60-volt
battery with 1kilohm feed). The second stage inductor need not be capable
of passing DC because it is connected in series with a capacitor across the
line.

Incidentally, if the equipment you're designing is series-connected before
other telephone equipment, DO NOT place the taxometer filter in the series
path! I have seen this done in a design that someone was selling for good
money. However, your equipment would have a huge insertion loss at the
taxometer frequency, which is not what the Deutsche Telekom people have in
mind at all, and would prevent a taxometer from being located beyond your
equipment!

One more point: remember to take into account both the DC resistance of the
filter inductor, and its AC behavior in the voice-band, when designing for
return loss and trans-hybrid loss. Also, don't place the filter after the
polarity protection bridge (the bridge rectifier that ensures that the
active holding circuit always receives the same DC polarity whichever way
the line is connected). When the pulses are huge, the line voltage may
actually reverse during the peaks, and the 16 kHz sinusoid will longer be
sinusoidal. The result will be an impressively distorted waveform and, I
imagine, loads of problems with data reception!

I would be very interested to hear experiences of others who may have
passed worst-case BAPT conditions using an active circuit with fewer than
two filter stages, or who have used or encountered arrangements based on
the other two alternatives (wet transformer or passive holding circuit) and
are happy with V.34 performance, cost and size. I believe information of
this kind would be enlightening to quite a few of us.

Finally, as I mentioned, it is easy to design for a fax-only modem, because
fax data transmission is half-duplex and the distortion requirements much
reduced. In this case, wet transformers are suitable, and I would recommend
the Midcom 671-8001 for the North American market (where taxometer signals
are not a problem) or 671-8067 for the European market (where they are).
The difference between these transfomers are, of course, associated with
the widely differing safety requirements of these two regions. Both of
these Midcom transformers represent cost-effective solutions.

I hope this information is useful and look forward to further comments on
this topic and the associated topics of THL and return loss optimization.
All these topics are linked and need to be considered together when
designing a DAA circuit that meets the regulatory requirements and
functions well for a large proportion of short and long lines, and lines
with inductors added (for improved high-frequency response), and lines with
worst-case taxometer signals.

Best regards,

David Drori

---------------------------------------------------------------------------
------------------
Novarex Enterprises Ltd.,
POB 2833,
Jerusalem 91028,
Israel.
Tel: +972 2 581 0995
Fax: +972 2 581 3750
Mobile: +972 5 067 8686
E-mail: [email protected]
---------------------------------------------------------------------------
------------------

Reply via email to