Eeek, somehow this slipped thru and I missed it...
Well, I don't recall the exact conversation, however it has matured now (with some help from those more experienced with DAQ than myself)...
There was indeed a problem with the way I was trying to use the board, but there was a minimal effort fix for it involving the voltage dividers for the high side blocks. At ~50/50 ratio (not exact because I didn't have enough of one value), the input was going outside of the range of the ADC's op amps. The inputs are differential, but that was only part of the puzzle... even divided, the two upper blocks (batteries 2 and 4) were taking the input past what the amps could reference, causing some strange inversion effects that at first I thought could be covered with some magic math. As proven time and time again, there really never is a such thing... while it looked good when they were discharging, voltage during charge proved that I was dead wrong about *some* aspect of the ADC.
Ross found the issue in the schematic/datasheet for the ADC/opamps respectively, and suggested some changes for it. I think I really ended up somewhere in between what was suggested and where it was, as the full mod meant redoing both ends of each set of leads (I think LOL). I found it sufficient to only involve the positive lead from each pair in the modified divider, and change the mode the board operates in.
The end result meant not having to touch the battery side connections... They are set up with 680R resistors as bulletproof short protection, heat shrunk and reinforced with a chunk of zip tie embedded, soldered to lugs... a real PITA to redo considering what makes them up. The ADC operates in 0-600mV mode rather than 0-10V, and I actually even gained resolution over the original attempt.
I didn't redraw the original schematic sent up during troubleshooting, but here's what that first drawing looks like...
... what it looks like battery-side (1-4 from left to right)...
... and the mods made on the ADC side (replace the simple 560R across the inputs of the above schematic with the pairs as shown here)...
All said and done, it's now working well, and the limitations of the fluke were even clearly visible in the "departure" plots during calibration!
There's a single issue remaining, hard to track down, it's either a ground loop or EMI that so far only seems to happen during generator operation while they're charging. Still investigating this, but it's rare enough to need to run the genset that it hasn't taken any priority. The more pressing issue is the balancing between battery 3 and 4 (at the moment I suspect 4 has a capacity deficit).
As for op amp vs different ADC etc, that really wasn't much of an option at the time, I had to either make this work or forge on with the fluke and single delta plot (which showed when there was a problem but pretty ambiguous as to exactly where).
One other note on "Method 1" vs "Method 2" in the two departure plots...
Method 1 was intended originally to give finer and exaggerated details about what was going on, but the units (set as "mV" because that's the closest representation) - aren't accurate. Each line is the average of the other 3 blocks, less itself... eg, 1 = ((2+3+4)/3)-1 .... 2 = ((1+3+4)/3)-2 ... etc
Method 2 is a true deviation from average... 1 = ((1+2+3+4)/4)-1 ... 2 = ((1+2+3+4)/4)-2 ... etc
Until next time...