BTW, when you print the balls w/o color management it can be useful to compare what the profile thinks is being printed to what is actually printed. This can be done as follows:
1. Assign the balls image to the printer profile.
2. Convert the image to Lab selecting Absolute Colorimetric.
This is the way soft proofing was done before view proof existed.
Your image should now match your print. At least to the degree your viewing illuminant matches your monitor. For instance if the monitor is set up for 160 cd/m^2 and D65 and your illuminant is D65 at 500 lux the two should match quite closely. You don't have to use that. I use D50 at 100 cd/m^2 for the monitor and D50 at 300 (ideally 314 - its not really a visible difference) lux for the illuminant. But they should both be at the same color temp and the lux level of the illuminant should be Pi times the monitor luminance ( cd/m^2 ).
Another thing that sometimes shows up doing this is when you have a small number of patches. There can be noticeable banding on the balls viewed through the monitor even though the actual print is quite smooth. I've seen this with small patch sets (918) but when I go to 2033 and use that profile they both look quite good. It's a good way to check if you have enough patches making profiles.
EtoA: Note that it's possible to print more saturated colors than your monitor can show. You should see close images if you have a wide gamut monitor but if it's an sRGB monitor you could well see banding on the monitor image that doesn't really exist and is an artifact of the monitor's limited color gamut. A way to check this in Photoshop is to desaturate colors by 35% or so in the Settings menu. This washes out the colors overall but will get rid of artificially induced banding from a narrow gamut monitor. Don't forget to disable this when returning to normal work.
