I asked myself the very same question just yesterday, as I'm using a 4110 right now to test some oscillators. As you point out, the specification is not very, well, specific, so I measured it myself using a PXI-4461.
The attached figure shows the spectrum from 10 Hz to 100 kHz of the 6 V output channel set to 5.0 V. The oscillator being tested pulls about 34 mA of current according to the readback - not much, I know. The resolution bandwidth of the acquisition is near 1 Hz, so the noise density is correctly indicated in V/rt(Hz) and the individual spikes are almost correctly indicated (maybe about 10% low). The total amount of noise is shown to be about 120 uV rms in the 10 Hz to 100 kHz bandwidth, and it's clearly dominated by the 3 kHz switching frequency and its harmonics.
I haven't measured above 100 kHz. Nor have I measured the other two channels.
Hope this helps,
Ed L.
