I'm not sure what the solution might be, but given that both your preamp and your amplifier have specified 3db frequency responses that go down to an extremely low 0.1 Hz, and given that your amplifier is very powerful, I suspect that the explanation of what is going on may be what Atmasphere (Ralph) described in
this post, and his next post in that thread. Except that in your case the low frequency instability he refers to is presumably in the power supply of the phono stage, rather than the power supply of the preamp.
In that thread, the problem disappeared when the load impedance seen by the tube-based preamp was reduced from 100K to 47K, for reasons that are explained in Ralph's post. Although he felt that was not a good long-term solution, as an experiment you might try reducing the load impedance seen by the phono stage.
I see that your preamp has an unbalanced input impedance of 47K. A convenient way of substantially reducing the load impedance seen by the phono stage from that value, as an experiment, might be to use y-adapters at the outputs of the phono stage, and routing its outputs into two components at once. If you have some other unused component handy that has line-level inputs having 47K input impedances, for example, having the phono stage connected to both components at once would reduce the load impedance it sees to 47K/2 = 23.5K. The second component needn't even be turned on, or connected to anything else.
Alternatively, you could solder a couple of resistors onto RCA connectors. Or perhaps find online an equivalent termination that is already prepared, that you could order. Something like 10K might be a reasonable value to try.
If that eliminates the problem, while it wouldn't be a suitable permanent fix it would provide confidence in the explanation. The explanation seeming particularly plausible in this case, as I said, because of the extreme low frequency bandwidth of the Krell components, coupled with the high power capability of the amplifier.
Hope that helps. Regards,
-- Al