Please see schematic tested in EveryCircuit:
The positive-side current sensing works, both inputs are pretty high, the amplification of 66.6 times works, and the output is 2.54v, well within the 15.3v that the opamp is hooked up to.
Now the reality:
LM324 is connected to full battery voltage.
Built, as in the drawing, the opamp's inputs in respect to it's GND are:
when idle:
+input is 224mV (in respect to opamp's VPP)
-input is 214mV (in respect to opamp's VPP)
so input is only 20mv from VPP, and the difference is only 20mV
the output is about 78mV (relative to ground)
(All good, I do not need a very precise measurement.)
with 10 A load:
+input is 224mV (in respect to opamp's VPP)
-input is 235mV (in respect to opamp's VPP)
so input is only 21mV "in the other direction", and the difference is only 20mV
the output is about 78mV (relative to ground)
What do I do wrong ? - Too close to VPP ? - I've used the same opamp on the low side, having the -input pulled almost to the ground, and +input 1v above, and the difference works fine then..