Hello, could answer my comments in your question? Also I am not sure what you mean by "being able to get a saturated HSQC". What is the composition of your sample in terms of the concentration of peptide and the protein? Do you have any idea about the Kd? What is the molecular weight of the complex?

If your HSQC of the protein changes after adding peptide (and you are sure that pH and salt concentrations do not change as you add peptide) - then you have a chance to solve a structure of your complex. - Because what you see in that case is the spectrum of the complex.

Roughly, you should be able to record a high quality HSQC within 20 minutes of instrument time using a standard HSQC pulse sequence. If you need substantially more instrument time - then you will not be able to record 3D data on the same instrument with sufficient signal to noise ratio, because conventional 3D data takes anywhere from a day to three days to record (depending on the experiment).

On the contrary, if the spectrum does not change - that is all you see is the unbound state of the protein - then your chances are signigicantly lower. You might still be able to use T2 and T1 relaxation data to map "points of contact" between the molecules and infer chemical shifts of the bound state (see for example this JACS paper by Lewis Kay and co-workers).

Have you recorded F1,F2 and F2-only filtered 2D NOESY? The F1,F2 filtered experiment will allow you assign resonances of the peptide in the bound form, while F2-filtered NOESY - will have intermolecular NOE's. If you don't see spectrum of the peptide then you won't be able to assign resonances of the bound form. Intermolecular NOE's may be quite elusive, but fortunately they are not the only possible source of intermolecular distance restraints - there are methods based on use of paramagnetic labels - PRE (paramagnetic relaxation enhancement) and Lanthanide Pseudocontact Shifts. However assigning resonances of both components is a pre-requisite to obtaining structure.

You should be able to tell whether the most sensitive instrument will help you by running HSQC on your instrument with more scans. Sensitivity scales as B^3/2. Cryo-probe vs RT probe gives a further 2-4x enhancement, so after switching to 900 cryogenic probe from room temperature at 700 MHz you can expect a factor of 3-6 gain in sensitivity (most likely 3 - for buffered samples).

So if you do see peaks from the complex after in HSQCs recorded in 3 to 12 hours on 700 then it will be possible to record adequate 3D data on a 900 cryo.