Digital BW, The Print

Ok this is driving me crazy here, I'm past the point of caring about asking a stupid question.

I'm running an 11880 and until now I really haven't used ABW mode all that much, the majority of my b/w work has been done on a K7 9600.  I have built a few ABW profiles here and there, and they've worked reasonably well, so I haven't given it much thought.
However, after recently upgrading to snow leopard, something's amiss in ABW town.  I use Photoshop CS5 for image editing, but kept CS3 just for this reason. However, when I tried my old settings for creating an ABW profile, it was way off.  I downloaded Jon Cone's linearization checker, and sure enough, I'm definitely not linear.  It makes me wonder if my ABW profiles ever were...  I guess I foolishly assumed that as long as I was getting decent results, they must be.  Needless to say, I'm excited to have found a way to precisely confirm linearization, the numbers don't lie!  Don't ask why I didn't just make a chart in excel before... 
So I've been messing around with this all week, and I'm to the point of analysis paralysis- I've tried every combination of workaround and setting I could think of, and I still can't get one ABW profile that falls within the acceptable range for a proper linearization.  I'm sure I'm missing something stupid, so I figured it's time to run this by some other people.  Here's what I'm doing:
I kept the most recent printer driver that still allowed you to use profiles with ABW mode.  I'm running snow leopard in 64-bit with CS3 & CS5 installed.  Out of CS3 I print the 21x4 chart that comes with QTR, leaving it untagged.  I set the printer to no color management, neutral, dark.  I read the results with  measuretool and drop the generated file on the create icc app.  At this point I load the LAB values from the "out" file into Jon's linearization checker just to see where I am.  And maybe I'm getting screwed up from QTR here, but I don't expect it to be perfect at this point.  I thought the true test was to print another chart, this time with the profile.  I leave it untagged still, and send it with the new profile, using perceptual.  I drop the readings from the new chart onto the linearize app, so it can average the 4 measurements, then I load the LAB values from the out file into the linearization checker.  The results show a nice smooth curve, but it's way too light.  After trying every variable with this method that I could think of, I started trying some of the workarounds for CS4/5, even though you're not supposed to need them for CS3.  Some of those results were closer to linear, but always looked like one S curve or another.
So am I making some simple mistake that's throwing the whole thing off?  Am I misunderstanding this tool for checking linearization?  Is the new OS/software combination working against me?  Or is it not even possible to get a perfectly linear ABW profile without using an .acv?
My poor old spectroscan has had a serious workout this week…

Any input would be most appreciated!
Thanks