[sdiy] Patchell's Boggling 13700 Expo Converter Thing???

[Jim Patchell]

	Scott Bernardi is responsible for part of that, so I am only half 
boggling you all... :-)

	If you haven't looked at 
http://www.silcom.com/~patchell/synthmodules/100-1007.pdf

	I will be refering to it.

	I suppose you can break it down to several small blocks.

	There is a VCA (made up of both pieces of an LM13700) U9A and U9B.  The 
purpose of the way it is connected is to make a VCA whose gain only 
depends on the control currents.  A normal CA3080 VCA will have its gain 
depend on both Iabc and Temperature.  The way U9A and B are connected 
make the temperature dependency of both stages cancel out.  Also, by 
connecting the LM13700 sections up like this, you get a very linear 
response.  I get about 0.1% linearity which is about twice as good as 
say a CA3280 using the internal linearizing diodes.  And, also a lot 
cheaper.  LM13700's being about $0.80 and the CA3280 being $3.50.

	The second block is the temperature sensor.  The temp sensor is made up 
of Q6A, Q6C, U10 and U11B.  The configuration is a Band Gap Reference. 
Q6A and Q6C are biased at two very different currents, 1uA and 0.5mA 
respectfully.  U10 is used to provide the constant currents for both 
transistors.  U11B is a differential amplifier.  It takes the difference 
between VbeQ6A and VbeQ6C, and is it turns out, the tempco of this 
difference is exactly what is needed to change the gain of the VCA 
mentioned above to compesate for the scale drift in the expo converter. 
  The output of the differential amplifier is fed into Iabc of U9B.

	The Iabc of U9A is used as the master control for the Volts/Octave. 
U11A is just a simple voltage to current converter with R52 adjusting 
the imput voltage to control the scale factor.  There are actually 
several ways the scale factor could have been controlled, but I chose to 
do it this way.

	Now U8B and Q6B are not really part of the temp compensation.  They are 
use to generate the voltage needed to compensate for the Bulk Emmiter 
Resistance.  I did it this way as a means of getting rid of the Vbe of 
Q6A rather than using a diode...just sort of my choice.  The reason I 
did it this way was to remove as many things that vaired with 
temperature as posible...maybe I was over doing it, but the Vbe of Q6 
does vary with temp...so...

	The rest of the oscilator is just you regular every day Terry Michaels 
oscialtor (aka ASM-1)...

	Hope that maybe makes things clearer...if not, I can try again...and 
hopefully, I am going to be running this circuit through its most 
challenging test very shortly.  I had hoped to do this a couple of 
months ago, but, things just kept getting in the way.  To test the 
stability of the oscilators, I am putting them into a polyphonic synth, 
which is a very demanding application.

Cynthia Webster wrote:
> Hi!
> 
> Jim, while we're talking about 13700s...
> could you shed some more light on your mystical
> 13700 exponential converter circuit?
> 
> It's really boggling!  I look at it
> and just don't *get* it...
> 
> Please enlighten us on just what is actually
> going on in there?
> 
> Thanks a bunch, I feel that this is of
> *major* usefulness, but here I am trying to apply it
> to a discrete oscillator design and I keep
> scratching my head over it?
> 
> (please help?)
> 
> LOL!
> 
> Hey, it's a Sunday!
> (hope no school busses are trying to mow you down today!)
> 
> Best Wishes 
> 
> oh, did I mention...
> 
> < HELP! >
> 
> LOL!
> 
> Thanks 
> 
> Cynthia
> 
> 
>