Mike Keitz wrote:

> Not a good idea, phones work on current rather than voltage.

  I know that.  Use a lamp.  Use a regulated supply *and* a lamp.

> Some won't like having 8V forced into them.  Others won't work on only
> 8V.

  The latter won't pass FCC then!

> The simulated "phone company" you supply should look like a 50 V
> source with about 1.2K ohm DC resistance, and 600 ohm AC impedance.

  Obvious error there.  Firstly, if there is more than 600 ohms in the
line, and there often is, then the AC impedance the modem sees is going
to be *much* more than 600 or 900 ohms (which goose left omega out of
the Windoze character set eh?).  Secondly, you are using the modems in
parallel, are you not?  In which case, you want the source to have
*infinite*, or thereabouts, AC impedance.

  My concern however was simply keeping AC hum and rectifier harmonics
out of it.

>  It can be difficult to supply the full 50 V, so another workable
> model is a constant current source of 25 mA or so with a compliance to
> 20 V.

  Again, since the spec for line equipment is about 8V, this is mucho
overkill.  I could be wrong, but 8V has always done in the past.

> Especially at higher speeds, it's a good idea to not run the loop
> current through the transformer, ...  Many that I have seen
> use a transistor active load to pass the line current, then couple to
> the transformer through a capacitor.

  So you want to minimise the capacitor eh?  You're using an undersize
transformer by avoiding DC, and ...

>  Since the active load needs a proper positive DC bias to work, there
> is a diode bridge in line to the whole affair.  I suppose the bridge
> could just be in line to the active load, but then a non-polarized
> capacitor would be needed to couple to the transformer.

  ... you're worried about introducing distortion but you want to use
an electrolytic capacitor for compactness ...

> The early external modems just had a transformer and a relay at the
> input, so they could be connected directly without needing to
> supply DC bias.

  I'd be looking out for them for this project!  I can now see *why* you
might need DC bias for the "mini" modems, though I still can't see why
you'd need more than 6V or much more than 40mA.  My specification of an
8V regulator was actually a convenient value to use with a 12V wall-
wart rather than anything else, regulated 12V to a 12V lamp would seem
to be pretty good though a nice DC-rated choke as well would be great.

  I personally would be really tempted to "bodgie" the offending
components out of the modem. and connect through a (plastic) capacitor
to the transformer.
--
  Cheers,
        Paul B.