Scored a 89410a from the army for 500 €...

Zotino v1.3 output voltage noise

after 15 cm IDC cable, IDC-SMA, and 100 cm coax (~50 pF) , and 500 k || 150 pF
3.5 V DAC output
on a running but otherwise idle Kasli
in a shielded crate
connected to RJ45 Ethernet
class I power supply
regular office environment
after a few hours of thermalization
c.f. previous https://github.com/sinara-hw/Zotino/issues/21#issuecomment-380472116
c.f. CMC effects and crosstalk https://github.com/sinara-hw/Zotino/issues/21#issuecomment-376119235
c.f. temperature coefficient https://github.com/sinara-hw/Zotino/issues/21#issuecomment-380401138
look at trace A

tl;dr:

~~stability is excellent to better than 10 µV peak over an hour, no discernible drift~~
up to the filter bandwidth of 75 kHz about 200 nV/rtHz as expected
at 4.6 nV/rtHz at 1 MHz. There is a bump up to 2 MHz that shouldn't be there from the model. My guess is that it's a joint effect of output driver, impedance (500 R || 2.2 nF), CMC-IDC-CMC, coax and analyzer impedance (500 k || 200 pF). That bump definitely exists. Might even be something intrinsic on Zotino. It would be nice to understand and model this. Also regarding Fastino and CMCs for these applications in general.
above 4 MHz it's < 1nV/rtHz as expected, less then I care to measure now
no unexpected spurs (the 0.9 Hz is a sideways mode of the table and the 89410a is pretty sensitive to vibrations, (2n+1) 50 Hz is expected and well behaved, 30 kHz is a CRT, stuff coming up towards 10 MHz is related to grounding).

0.69 mHz bandwidth, 5.0 µV/rtHz at 1.6 mHz

The analyzer 1 Hz AC coupling is in fact a fixed analog lowpass and not an active subtraction: The data below 1 Hz is affected by that lowpass