hmm, isn't it the opposite?... seems like the perfect scenario/situation to use one.
with the autoformer following the kdac, for every 3db of attenuation the reflected load is halved (e.g. for the naksa, 28k -> 56k), at -12db the naksa would be a nice easy ~200k load
http://www.intactaudio.com/atten_FAQ.html
Yes that is correct....... mostly but in an ideal world with an 'invisible' transformer.
There are a couple of real world mechanisms that can make things less than ideal.
Transformers have core losses and require magnetization current. For every halving of the frequency, the core magnetization level doubles.
This becomes very prominent at frequencies below a few hundred Hz and once you get below 100Hz can be clearly seen (and heard).
It depends a lot on whether the transformer is a high impedance type or low impedance type. Low impedance types are the most
transparent but at low frequencies especially need to be driven from a low source impedance.
High impedance types have generally a bit more 'colour' (distortion) but are much less dependent on driving impedance.
The second thing is that at low frequencies this increased core magnetization current will impose a load on the Killer DAC's high
OP impedance.
So the bottom line is if the autoformer is fairly high impedance, ie lots of windings it may work reasonably well but
it is unpredictable what the outcome is.
If you were seriously considering this approach I'd shoot Slagle an email and ask the question.
T