QTR-Quadtone RIP

Hi all,

I recently replaced my 2100/UT3D setup for a 3800/UltraChrome one. 
Obtaining a neutral QTR curve for a given paper with the UT3D inks 
was pretty straightforward. However it appears a bit more tricky with 
the UltraChrome inks. So, I would appreciate some guidance.

I've extensively looked at the UC curves supplied with QTR and the 
basic recipe seems obvious to me:
1) create a warm curve using MK, LK and LLK
2) add some C and LC as toner 1 curve (use ink limits for coarse 
adjustments) 
3) add some M and LM as toner 2 curve (use ink limits for coarse 
adjustments) 
4) micro-adjust the toner 1 and toner 2 using the curve point feature.
5) linearize

However, the following is not fully clear to me:

- First, there have been several threads in the past on defining 
a "neutral curve". I took from that that it is common sense to strive 
for fractional Lab a and Lab b values up till 50% density and from 
there on bring the tone smoothly to paper white. So, here I am with 
my draft curve adhering to the guideline above, at least that's what 
my spectro tells me. But visual inspection still shows some color 
cast in some areas. My assumption is that this is because the Lab a 
and Lab b are very small indeed, but fluctuate a bit around zero, 
leading to one color hue in one area and another one in another area. 
Perhaps a humans eye is still very sensitive for these local 
variations. As a matter of fact, I do have a draft cuve with higher 
Lab a values that actually subjectively does look more neutral than 
the objective "best" curve. So perhaps I should consider to allow for 
a tiny overall, identical, color cast to avoid the fluctuation around 
zero. Does this make sense?

- Do I assume correctly that one FIRST brings Lab b to 
(approximately) zero by adding C and LC and only then start adding M 
and LM to adjust the Lab a value (since magenta is predominantly Lab 
a)?

- It is impossible to add sufficient C and M to neutralize the 
darkest tones (85-100%). Frankly, I consider to ignore that 
altogether, my reasoning being that 1) any color hues in these dark 
tones are virtually invisible anyhow and 2) perhaps Dmax reduces 
since I'm using ink load capacity for inks that only suboptimally add 
to Dmax. Am I right?
  
- The curve point feature is a bit of a puzzle to me. I have figured 
out its general use via the standard curves. First of all, there is 
apparently a curve fitting algorithm under the hood as often when I 
change a value on a certain point, one runs the risk that the whole 
curve is affected. Are there some general guidelines for defining the 
curve points?

- On the curve points, there's another tricky part: if I ONLY use LM 
(and not M) in a toner curve (density set to 25) I can BOTH increase 
and decrease the ink load for the LIGHTER tones but for the DARKER 
tones I can only DECREASE the ink load. If I try to increase it, it 
decreases nevertheless. My assumption is that the algorithm just 
assumes there is an M ink as well and tries to increase the load of 
that non-existing darker ink (and simultanuously reducing the LM ink 
load). This assumption is confirmed if one adds a M ink. But how to 
circumvent this odd behaviour if one does not need/want an M ink?

Thanks for any light that could be shed on this,

Joost