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

[Ian Fritz]

Scott --

At 07:47 AM 6/22/2003, Scott Bernardi wrote:
>That's the part of the circuit that I contributed, and it goes back to my 
>early days
>as a analog IC designer.
>Bipolar semiconductor transistors have a Vbe with a negative tempco (our 
>famous
>-3300ppm/deg C), and we are familiar with the kT/q thermal voltage, which 
>has a
>positive tempco that is proportional to absolute temperate (PTAT).  By 
>taking a Vbe
>and adding to it a scaled version of kT/q, there is a "magic" voltage 
>where the
>tempcos will cancel out.

Right. So where is this in Jim's circuit?  I see the "scaled version of 
kT/q  part", but not the "taking a Vbe and adding to it" part.

>  For silicon, that magic voltage is 1.25v and is called the
>"band-gap voltage" (and it's also why the commercially avalable voltage 
>reference
>IC's are some multiple of 1.25v). If you used germanium transistors, there 
>would be
>a different "magic" voltage (about .78v). The band-gap voltage 
>characteristic is
>also what determines the scale factor for our exponential generators: 
>approx. 18mV
>per octave or 60mV per decade for silicon.


>Using the difference between two Vbe's running at different current 
>densities is the
>most common way to genate a PTAT kT/q term. In Jim's circuit we convert it 
>to a
>current to drive an OTA two quadrant multiplier which multiplies the input CV
>voltage times a PTAT term that will cancel the negative tempco term of the
>exponential transistor.

Yes, this is all clear.  But why call it a band gap reference 
circuit?  That's what I'm missing.

   Ian