Myke Predko wrote:

> Hiya,

  G'day!

> I just sent this reply back to Paul directly when I meant to send it
> to the entire list:

  No worries, I caught both, will forward to the list only.
 
>> Three output states for each resistor makes nine, enough for quite
>> detailed analogue sound output.

> Paul, could you explain how to do this with a Resistor ladder?  The
> Best that I can do is n + 1 (where "n" is the Number of Resistors):

  * diagram of rather assymetric circuit bearing *
  *    negligible resemblance to R-2R ladder!    *

  Your circuit is a little confused.  Let me propose again the use of
two outputs, each with a resistor to a common output point (coupled to
the LM386 by a capacitor) which is the centre point of an equal voltage
divider between Vcc and ground.  I specify that the resistors from the
PIC outputs are related in a 3 to 1 ratio, and the values in the voltage
divider are each equal to the lesser value of the other.

               Vcc
                |
     |          R
     |          |   DC Block
  o1 |----R-----+------||--- Output
     |          |
     |          |
  o2 |----3R----+-- "common point"
     |          |
     |          R
                |
               Gnd

  If neither PIC output is enabled, the common voltage is ¸Vcc.  If the
output coupled to the lesser resistor is enabled in either direction, it
delivers a voltage bias of Vcc/6 to the common point in that direction,
while the other output produces a third of this bias in whichever
direction.  Calling this lesser bias increment "i" and referring this
to ¸Vcc, the effects of the two outputs are:

            o1
       Gnd O/C Vcc       Nine states in total, four positive
   Gnd -4i  -i +2i     and four negative increments relative  
o2 O/C -3i  0  +3i     to the zero point.  If you REALLY want
   Vcc -2i  +i +4i     twenty-seven states, use a third output.

  Even a four bit (81 state) system would fit really nicely packed in
a byte using the SWAPF instruction to extract first the port, then the
TRIS value, and a lookup table might of course be used to read such
values from a "phase" or syllable table.

  Something tells me an "R-3R" ladder would also be usable for this
task.  For the moment though, stock resistor values are sufficiently
accurate.

> Now, I suppose that multiple Outputs could be turned on to get
> parallel resistances, but I would be concerned about this with uneven
> current drive capabilities with different pins.

  I wouldn't!  The PIC "Maximum Current Sunk" by an output pin being
quoted as 20mA (though this is actually a maximum safe rating rather
than a claimed drive ability) at 5V suggests an output impedance of 250
ohms.  Since the input impedance of the LM386 is about 250k, the "R"
above should be picked in the order of 270k (giving 3R a convenient
value of 820k) and the current drive capability of the PIC pins is
trivial.

> This method would make it very hard to calculate the actual Voltage
> Outputs.

  That's what scientific calculators or PCs are for!

  Cheers,
        Paul B.