[sdiy] Blackmer cell

[jhaible at debitel.net]

> Interesting.  I hadn't learned about the Blackmer cell before.  I notice 
> that if you remove the PNP's (Q3 and Q4), the I_CC and the V_B sources, 
> then it's just our usual current-sink expo converter.  Similarly, removing 
> the NPN's gives the current-source one.

Yes. Our ordinary expo converter with opamp servo loop to control the emitter 
current of one transistor is a multiplier x * e**y, where  x is the collector
current. That part is very (perfectly?) linear, so you choose it as your audio
signal input. y input is exponential - just what you want for an expo VCA.
Only problem is that it's 1-Quadrant only.
For VC ADSRSs etc. you can simply stack a npn and pnp circuit together, sharing
the servo opamp which drives all four emitters. For Audio, you want Class A or 
AB operation, so you have to keep the two emitter pairs at a certain distance, 
to get some bias current.
As we know from this type of expo converter, there's always a tradeoff between 
stability and response time. (I think you contributed some very interesting 
insight in tiny oscillations that would slightly set off the tracking in VCOs,
if memory serves.) So while a Blackmer cell is wonderful in terms of offset 
(you see the direct offset of the servo amp, not the vastly amplified offset of 
a differential pair as in OTAs!), I think it may be not as good in terms of 
speed.

JH.

-------------------------------------------------
debitel.net Webmail