I think it's an error in the design, as there is already more than enough BA6110 "grilled"..
Now i desoldered socket and modified pcb (cut the traces around pin3), i think i would also leave it floating for now, until i would figure for sure where it must be connected. When the bipolar transistor base is disconnected, it should be in the "off" state
LINK For now it looks like more probably, according to the common sense and the datasheet, in the VCA mode "input bias" should be connected to -VCC, but im not sure how it should be done here.
So waiting now for the comments from the designer, and if he confirm the problem - it should be included in the build manual asap (to avoid frustration and additional demand on BA6110).
ADD:
Look how it was done in the x0x :
[resize=500]
http://www.firstpr.com.au/rwi/dfish/TT- ... -drive.png[/resize]
BUT please note, x0x is actually have only GND and +VCC !
ADD2:
quote from
HERE :
"pin3 on the ba6110 controls a bias circuit to linearize the input stage.
either leaving it floating or connecting it to ground will turn it off, and make it the same as the ba662. i generally like to connec things to ground, just to ensure against stray flux leakage or hum pickup."
But again, remember that the above statement about "connecting it to ground" is correct only for the unipolar power.
So in the nava case - with the bipolar power the "-VCC" is become equivalent of "GND" with unipolar power.
Who is brave enough with more than one BA6110 to connect the pin3 to -VCC and test it? =)
It should give less possible noise compared with leaving it floating.