The challenge is that the scale factor (the magnitude of the utilities) in choice experiments (whether CBC or MaxDiff) is inversely related to the response error. And, response error is generally much lower in MaxDiff compared to CBC, leading to utilities with much larger magnitudes in MaxDiff than CBC.
So, you cannot just compare utilities directly from a MaxDiff to a separate CBC. If most of the levels involved in the MaxDiff are also involved in the CBC, then it might be possible to do a fancy data fusion where the two sets of data are stacked (and recoded to have a common X matrix). That takes some work!