SE to diff converter

[NiHaoMike]

While looking through the datasheet for the DRV1100, I noticed that the suggested circuit for converting a SE signal to a differential signal (for input to the DRV1100) used two opamps in a strange "hourglass" feedback loop. ( http://www.ti.com/lit/gpn/drv1100 , page 9) Is there any reason that peculiar circuit was used instead of a standard inverting amplifier with a gain of -1?

[DJPhil]

I'm not quite sure, but I'll guess.
The 500 ohm resistors look like a fixed Wheatstone bridge. This makes the resistance between any two points equal and dependent on all four resistors (ratiometric relationship). The upside to this is that it's more stable against low tolerance parts and thermal drift. The bottom opamp sets that whole network up as a virtual ground equal to the input ground, so it's really a very stable way to bias the midpoint in your differential signal.
If I understand it correctly it's actually a solid and elegant solution.

Hope that helps.