A 9 watts SET for the Maggies, Heavenly marriage

For the heading, I need to start a new thread; this is because the last thread does not do the Audio Note the justice.
The comparison among all those EAR, Pass labs, YBA and the little, David, Audio Note kit one, does illustrate that if we can relate our live musical experience, the SET certainly rules.
For whatever reasons, I like to have advice on higher powered SET amplifiers for the Maggies
Please advice
A friend of mine, just because he likes to experiment, put a single-ended 45 amp on his 3.6 maggies. That amp puts out about 1.5 watts and was, of course, WAY underpowered for these speakers. However, he has also run them with a Wyetech SET that uses a 211 tube and puts out something like 18 watts. That combination sounded quite good to me, so long as the volume was not pushed much. It is possible to be satisfied with high-powered SET/Maggie combination, but one does have to forego some of the really good sounding directly heated triodes (e.g., 45s & 2a3s).
When I was selling my 2a3 I auditioned it for a prospective buyer on my MG 12's. It worked, but just barely - even at low volumes it just wasn't the amp for those speakers in a smallish room.

Subseqently, I heard a guy's maggies with 300b's in a , very very small room and that made him happy for quite awhile, but eventually he kept the 300b's and moved on to a large full range fostex set-up which was just a way better match(IMO).

Personally I don't think you get all the set magic without going to a much more efficient speaker. If you are a maggie guy, then a higher power amp with more grip is what brings out their best.

Different strokes of course....
I have no other choice because I fall in love with the Magnepan sound and also sleeping with the SET.
I've wanted to hear a pair of Maggies driven by one of DeHavillands 845 output tube SET amps. That seems like an combo that might have some potential charms if one isn't going to drive them to headbangers ball levels. In particular, the new DeHavialland GM70s look interesting for this experiment.