As far as I understand, a partially oriented sample can be identified from the quadrupolar splitting of deuterium. If you do not have deuterium channel (example in many Bruker instruments with cryoprobe) you can observe the lock channel which should show two different wings. I have used this technique a lot and for an unstable sample in liquid crystal media like PEG, which is losing its partial alignment, the two wings observed in the lock channel will slowly changes into a single one. The amount of quadrupolar splitting directly depends on the degree of alignment. You can find more about these relation in this paper Rueckert, M.; Otting, G. J. Am. Chem. Soc. 2000, 122, 7793-7797. One other way of determining the orientation of the molecule (although this will be more tedious) is to do some C-H or N-H coupling experiments in an isotropic and anisotropic sample and measure some of the splittings. A measurable difference in between two different splittings will give you an idea of whether partial orientation has occured. Although one should be careful in terms of stumbling across those peaks where there might not be any significantly measurable RDC (that is at the magic angle orientation).

There are no hard and fast rule about the acceptable difference in coupling constant between isotropic and anisotropic samples. It depends more on the degree of order, the nuclei for which couplings are measured and also their relative distance and finally the type of nuclei you are looking at. Long range couplings will have significantly smaller magnitudes than single bond interactions in a given ordering media at a given concentration.

The splitting of the D2O signal should be around 20-40 Hz I think.

I saw splitting when I did this I did this with phage and I did several concentrations of phage to finally see splitting however I don't remember what that concentration I used. You have to optimize this part

The quadrapolar coupling between proton and deuterium (2QHD) can be used as a probe for checking alignment in most cases (common exception being polyacrylamide gel aligned samples).

(1) In the case of Pf1 this coupling is dependent upon the concentration of the phage ([Pf1])used, the dependence being, 2QHD ~ 0.89*[Pf1] (2) In case of like poly ethylene glycol (PEG) once again the concentration of the media matters, and for a moderate alignment (i.e. N-HN RDCs range of ~-20Hz to +25Hz) a 5%w/v solution is good. For n-octylpentaethylene glycol media (C8E5) a 2QHD of ~30Hz was observed when I tried aligning my sample. One may consider doping this media with cetyltrimethylammoniumbromide (CTAB - for positive) or sodiumdodecylsulphate (SDS - for negative) for charged proteins. A 5%w/v C8E5 with 30:1 C8E5:CTAB gave a 2QHD of ~34-35Hz. (3) Polyacrylamide gels (PAG) can also be used where the 2QHD also gets averaged to zero. Therefore one must used the protein and collect coupling values to assess whether the protein has gotten aligned. As a general rule, now-a-days 7% 1:1 charged PAG are used for proteins with MW in the range of 8-20kDa. There are two ways of using these gels - either vertically compressed or horizontally stretched - which are expected to give orthogonal alignments necessary for comprehensive RDC data analysis. Larger proteins will require a lower % of PAG while smaller ones might require higher %s, as the alignment mechanism is dependent on sterics and shape of the protein.