What is the purpose of D1? .... apart from reducing headroom, drawing extra current, and introducing a potential extra noise source?
Thanks for the question. I admit that starting with the OPA Alice schematic is part of the reason for including D1, but I think the issues you mention are not that major.
- "reducing headroom"
- the max swing is ¡Ö12V pk-pk at each op amp output, which means ¡Ö24V pk-pk at the XLR pins, which is an absurdly high amount for most interfaces (e.g. Focusrite Scarlett 18i20 mic inputs have a maximum input level of 9dBu at minimum gain, for a sine that would be 6.2V pk-pk.)
- "introducing a potential extra noise source"
- as you mentioned, it's filtered, and also the op amps have very good PSRR
- "drawing extra current"
- this is a good point, since I am typically powering this from batteries in an XLR transmitter (Sony UTX-P03)
- however, not drawing the extra current (even when it's not used, the current through D1 decreases as the current through the op amps increases) introduces issues with transient response:
- whenever there is a sudden increase in input amplitude (e.g. going from silence to screaming), the current drawn by the op amps will rise sharply (because they need to drive more voltage into the XLR input impedance)
- because of this, the common mode of the XLR input will drop, this needs to happen with very low slew rate to place it in a sub-sonic spectrum or it will cause a "DC thump" (so we still need to have the 100u caps to ensure this), and also VCC will drop
- we need to manage the time constant between VCC and VMID, because VMID has its own cap and if VMID does not drop at the same rate as VCC/2, then we reduce the headroom.
- this is kind of a headache I didn't want to manage so I was conservative and kept D1 as in a know and well-regarded implementation (OPA Alice).
?
