I feel for you, geofizz. You could be describing one of my kids, without the oral apraxia. (No official diagnosis, but pretty much the same profile based on my clinical observations and informal probes.)
Both AAS and Barton are OG-inspired, and likely to be equally effective on your son. We've used AAS with our equivalent child. Prior to using it, we tried a few traditional spelling programs, and found that, although all the spelling exercises looked all right at the time, there was negligible transfer or retention. Also, it was a tremendous struggle to get through any kind of written work (likewise including math). We have now been through a few levels of AAS; spelling is actually an enjoyable subject (love the fact that you can finish a meaningful session in 15-20 minutes). I did a fair amount of poking around before I decided to try AAS; I was specifically looking for a home OG-style program. I actually do know a couple of school-based OG-certified tutors in our area who probably would have done it, based on my personal relationship with them, but since we homeschool, and I have a certain amount of exposure to the approach myself, I thought I would try AAS first. It has been very easy to do, as it is all laid out for you and scripted. It's very hands-on and visual, which has been very suitable, and allows you to practice your spelling separate from letter formation/handwriting, which initially was tremendous for us (you use magnetic letter and phoneme tiles to assemble words). In terms of results, of course I don't have a control! But the retention of spelling words appears to have improved, as has generalization to applied writing tasks. Most importantly, I've seen substantial reduction in resistance to writing. It's also much less common that I see peculiar dysphonetic misspellings now, which is a big step for us.
On placement: I started from the beginning, on the basis of research that indicates that the core deficit in dyslexia is in some level of phonological awareness, and that it is critical to remediate the PA deficit prior to investing too much time in learning actual spelling words, as kids will develop bad habits for word attack and attempts at encoding otherwise. One of the issues when they already have an extensive written vocabulary is that they 1) will overrely on rote memory, and 2) have less incentive to practice good PA, orthographic, and morphological strategies enough to reach automaticity (which is already a primary difficulty for them). We moved very quickly through the first couple of levels, but did not skip any steps/lessons. In the process, I found that a lot of words were familiar and accurate, but quite a number of very easy words were oddly challenging. In the design of the curriculum, those words remained in the review rotation for months, while most of the other Level 1 and Level 2 words were reviewed only a couple of times before we moved on.
On the topic of diagnosis/eligibility: do you remember if he had multiple CTOPP subtests, and if so, how he did? I'm looking in particular for phoneme reversal, which is a supplementary subtest that can be critical in identifying compensated (or stealth) dyslexic/dysgraphics, such as your son may be. It's possible to use vocabulary and cognition to blind the test on many of the standard core subtests, but this one is trickier.
