Aoyagi et al. (J. Am. Chem. Soc., Vol. 115, No. 24, p 11393) reported that the addition of 2-lithio-1,3-dithiane to the alpha-chiral (S)-2-Acetylpiperidine gives the (2 S,3 S)-configured product. It is stated that the diastereoselectivity of this 1,2-addition (dr = 6.2) is predicted by Cram's model involving complexation of the Li-ion by the piperidine-N. I tried to reproduce the diastereoselectivity in the following scheme:
As you can see I come up with the wrong product by applying Cram's rule with a O,N-complexed Li-ion (right side of the scheme). By applying the non-chelate Cram rule (left side; biggest substituent oriented anti to the carbonyl, attack via smallest group) I get the correctly configured (2 S,3 S)-product.
Where is the mistake?