I have not looked for a buffer, but I wanted to explore the details of the problem as an academic exercise to see what one could do in this situation.
It was a bear finding out what the input impedance of the analog input of the Wiim Ultra actually is- Wiim does not say, but I finally found it from
a write-up of independent testing at soundstagenetwork.com. It's about 22K ohms.
And looking at the output impedance of the circuit (shown below) driving the tape out and main output of the HH Scott 355
from the schematic I downloaded for free from ElektroTanya, they used a standard 12AU7 cathode follower circuit with a cathode load resistor (Rk) of 47K and an output coupling capacitor of 0.1uF.
There is a great discussion of
the AC coupled cathode follower at the valvewizard.co.uk website, and in that discussion indicates that for this type of circuit, the most ideal performance is if the external load resistance is much higher than the Rk load resistor. Back in the day, power amplifiers were designed to have the higher input impedance necessary to work with tube pre-amps like this one, so 47K was probably completely appropriate. This output circuit can generally accept external loads as low the resistance of the cathode load resistor itself, and somewhat lower, but in doing so it shifts the slope of the load line more steeply vertical, which drops the signal swing in the output can definitely change the distortion characteristics of the sound for sure.
Another concern is that the output capacitor value of 0.1uF, which was completely appropriate for the expected high input impedance of the amplifier of the day, is probably too small to drive the loads of today. Using
a quick and dirty F3db RC calculator for 0.1uF driving into 22K, the roll-off F3db frequency is about 72 Hz, which would be quite noticeable in frequencies well above that.
I don't know if modifications to this Scott model are a "non starter", but it is worth considering a minor modification to the model to help it to better cope with the modern equipment it is now needing to play with. Following the simple example circuit in the valvewizard article, you could drop the 47K Rk resistor down to the 18K used there, which would increase the drive current of the 12AU7, which it can certainly handle, and lower the output impedance to at least be on par with the input impedance of the Wiim. You would also drop the 1K bias resistor to more like 470 ohms, and increase the size of the output coupling cap to say 1uF or so, depending upon what high quality poly or foil cap you could fit in there. I know modifying a Scott is probably heresy, but it's a simple enough mod, and it would get you at least 90% of where you want to be in terms of performance. Perhaps less heretical than having to pipe the precious signal through yet another set of interconnects, through another set of coupling caps, and adding the noise from yet another (probably switching) power supply to power it.