Guten tag,
Sorry I will reply in English today as my German is 35 years old and too basic for the topic at hand.
DFB1 does also not pay any attention to the /EXPAND line. I don't know enough about the CT60 (I have never owned one) to know why this matters for it but not for DFB1. I remain slightly skepitcal about EXPAND being a problem at all and tend to agree with Rodolphe that it should have no effect really.
BUT there is one area where I can see it could. Perhaps CT60 generates its own BERR signal for the lower 16MB range rather than relying on the motherboard's 128 clock cycle timeout? [nb. from memory it's 128 cycles -- about 8 microseconds, but I may be misremembering and it may be only 4microseconds]
DFB1 does not synthesize its own bus errors apart from at the top of TT-RAM. It relies on the motherboard to assert BERR and EXPAND is, of course, still connected to the motherboard.
Perhaps CT60 is not like this and, for some reason, Nova requires a very, very slow bus access from time to time?
BW