Fellow treggers:

I have been away from my email for a few days, but I was pleased to return
and find such a lively exchange on tax pulses.  I was particularly impressed
with David Drori's lucid description of the underlying issues and the
considerations that go into the filter design.  I fully agree with David's
remarks.

After reviewing what has already been said by all of the contributors on this
subject, I have only a few comments to add:

1) In Germany today (under BAPT 223 ZV5), there are only two requirements
related to tax pulses that modems must meet.  The first is that the impedance
of the TE must exceed 220 ohms at 16 KHz.  With some care in the design, this
requirement can be met without adding a filter.  The second requirement is
that the modem must not drop a connection in response to the application of
tax pulses.  As Duane Marcroft has pointed out, some fiddling with the modem
control registers can sometimes allow this requirement to be met without a
filter.

2) Tax pulses are not provided by default on all lines in Germany, but they
may appear on some lines when they have not been specifically requested.  In
some central offices, if any one customer on a given channel bank has
requested tax pulses, all lines on that bank receive them.  This is the
justification the BAPT used for imposing an immunity requirement on all TE.

3) In Switzerland, there is no requirement for modems to be immune to tax
pulses.  However, there is a requirement for TE to present a minimum
impedance at 12 KHz.  This requirement, which is misleadingly referred to as
an "insertion loss" requirement, can be found in section 4.1.2(CH)1 of NET 4.
 The test signal for this requirement is +20 dBm.  With a "dry" transformer
and a solid state DC hold circuit, it is extremely difficult (impossible?) to
meet this requirement without a series inductor.   

4) While it may be possible to get products approved in Germany and
Switzerland with absent or inadequate tax pulse filters, there is a risk in
doing so.  Tax pulses show up on a significant percentage of the phone lines
in these countries, and modems that are not immune to the tax pulses will
have problems.  Unsophisticated users are not likely to recognize the cause
of the problem.  All they will know is that the modem can not maintain a
connection on an outbound call, or that the throughput is quite low.  This
may generate problems for the technical support staff, not to mention the
image of the vendor.

5) I recognize the design headache that the tax pulse filter creates for
designers of PCMCIA cards.  Filters that perform well tend to require
inductors that do not fit.  On the other hand, a mediocre filter is probably
better than no filter at all.  I would be interested to hear from anyone who
knows of a good solution to this problem.


Joe Randolph
Telecom Design Consultant
Randolph Telecom, Inc.

Reply via email to