Well. I ended up with a Topping DX3 Pro+. I am achieving a peak voltage of ~3.02V, which we round to 3V and at <1% error is suitable for our needs and use, for example, 3V is 30 engineering units, i.e. 0.1V/EU. Two dummy values are added to the raw data eg +/-30, then the data is converted to a wave file so 0dBFS or 32,768 references the value 30.
Due to the various incoming files being various sampling rates, I used Audacity which I assuming is it resampling on the fly to 44.1kHz.
Audacity allows me to move a file to the left or right channel which outputs to channel 1 or 2 on the data acquisition unit, clearly see the waveform and/or spectrum and other nice controls over playback. With the DAC on dual mode the line out goes to the voltage in of the data acquisition unit and the headphone out goes to speakers with their own volume control. Everything always at 0dBFS.
The left channel can sometimes produce very slightly higher levels than the right channel when fed the same signal, I think this is frequency dependent. I also noticed when only using the right channel, the left channel gets some cross-talk, more so than the other way around.
Some very low freqs eg 3 hertz near full voltage cause a HP err on the DAC, but these amplitudes at low freqs I would never expect to see in our data so no issues there.
More testing to be done, but this looks to be a suitable method for now. Thanks for all the input and discussion.