[sdiy] Buchla 194 Fixed Bandpass Filter analysis

[Donald Tillman]

   > From: Aaron Lanterman <lanterma at ece.gatech.edu>
   > Date: Wed, 16 Apr 2008 22:50:23 -0400

Hey Aaron,

   > To get some practice analyzing circuits with transistors, I started  
   > looking at the Buchla 194 Fixed Bandpass Filter, as redrawn by Marjan  
   > Urekar:
   > 
   > http://members.tripod.com/urekarm/synth/buchla_bpf.pdf
   > 
   > 1) The highpass S-K has a "feedback resistor" and a "resistor to
   > ground." When I'm considering the "resistor to ground," should I
   > think of the 68K and 100K biasing setup as both being resistors
   > in parallel to "AC ground?" If so, my "to ground" resistor is
   > actually 68K || 100K = 40K.

To AC and DC ground, yeah.  It's called a Thevenin equivalent
(Wikipedia has an entry for Thevenin's Theorem).  A resistor divider
between two voltage sources such as the power rails is equivalent to a
single resistor, of value equal to the parallel combination of the
two, wired to a single voltage in between.

So a 68K/100K resistor pair across a 15 volt power supply is
equivalent to a 40.5K conneted to a 6.0 V supply.

   > With this figured out, I can compute the cutoffs and Qs of the
   > various filters.

Yes.

   > 2) All of the bases of the voltage buffers, except for the first one,  
   > have 10K resistors going into the bases. Why are those 10K resistors  
   > there?

Discrete emitter followers often include base resistors to hold down
high frequency oscillations.  Similar to the dominant pole in opamp
circuits.

   > 3) There's a 2K2 and 100K voltage divider right at the beginning
   > that attenuates the signal by a factor of 50, presumably to get
   > the signal into the "linear small signal analysis"
   > range. But... I don't see anywhere where the gain is brought back
   > up at the end! Is there something I'm missing?

The resistor values are actually the other way around; the circuit
only attenutates the input signal by 1/50th.  The 100K resistor keeps
the input cap from floating and prevents loud pops when you connect a
ground-centered input, and the 2.2K protects the transistor from nasty
hight current transients.

   > I thank the list, as always, for its collective wisdom!
   > 
   > I'm thinking it would be fun to create a modern version of this
   > that just uses op amps and a split supply.

No no no; the discrete version is lots more fun.

   -- Don

-- 
Don Tillman
Palo Alto, California
don at till.com
http://www.till.com