Okay, finally getting back to this.
Now that I'm understanding this all a bit better (after reading the manual/schematics, another book, and these posts a few times), I understand the input buffer feeds is what's flipping between zero, ref, and signal (I think I said earlier this looked like zero, zero, ref, sig, but I can't find where I got this from). Anyway, I was measuring this on TE1 and TE2, which is BEFORE the input buffer, not after. So, measuring pin 6 of the input buffer, with a 2.5V input signal on the 3V range, I get the following:
N1: ~0V
N2: ~0V
N3: -3.3V (I think this should be -2.8V)
N4: 2.97V (shouldn't this be 2.5V)
Double checking with my Fluke, rather than scope, these check out.
Checking Q32, which according to the diagram should be what switches on the reference -2.8V, I do indeed get -2.8V on its drain (it's a JFET). Also checked on TE1 and TE2, and yes, getting the correct 0V, 0V, -2.8V, and 2.5V on each of the N stages. So I guess that was the correct place to test before.
So, why is the input buffer adding voltage to these? I do see the exact same numbers on U20 pins 5 and 6. So the LF442 seems to be working (though I can't really say what it does in the grand scheme of things other than act as a buffer...)
Checking U45 pin 14, I get 0V, 0V, -2.8V, and 2.5V perfectly. So it's working.
That said, measuring the A/D side of R58, I get 0V, 0V, -2.54V and 2.6V on N1, N2, N3, and N4 respectively. So that seems to be working as well. Where I get lost is after that...
Looking at U19, though, I'm a bit confused. Pin 6, the integrator output, shows 14.4V on N1, N2, and N3, and then shows integration on N4. I would think it should show integration on N3 as well... but I'm not sure. Alm? Codeboy2k? I will say it's not a very stable integration. See photo.
One thing I have learned... the accuracy of a scope is relatively poor to a DMM when it comes to voltage. Off a decimal or two doesn't seem to be that big a deal).
-------
On a side note, and maybe this is my issue... I switched my scope over to AC coupling, and measured the filter caps on the supply side of the voltage regulators. On C59 and C65 I'm getting about 300mV of ripple (see picture) on the positive side (furthest from the stripe). But oddly, I get no ripple on the negative side of C59 and C65, but I do see the same amount on the negative side of C63. So what the hell? I don't see any ripple on the outputs from the LM7805 (5V Analog), LM7915 (-15V analog), or LM340T15 (+15V analog). Not sure if it's interesting that the +15V is a LM340T15, when all the diagrams say it should be a LM7815... guessing they're basically the same.
I also see about 550mV of ripple on the negative side of C72, and about 750mV of ripple on the positive side. I thought I was seeing ripple on the +5V digital regulator too, but then I realized I was referencing analog ground... referencing to digital ground, there's no real ripple (about 120mVpp of noise).
Should I just recap and see where I am at that point? I can't see caps being an issue though since the VR's all seem noise free.
Thoughts?