Tech Forum

That’s a very old circuit and it has at least one glaring flaw: The value of R3 is way too high. I suspect it should be 3.3K instead. Also, D3 and D4 seem unnecessary because there can never be more than about 2mA of meter current. R1 protects D1 and D2, but microamp circuit levels wouldn’t threaten them. The input bias current of a 741 op-amp through a 10K resistor produces up to 5mV offset that’s not temperature-compensated. It’s nullable, of course, but the null can drift.

Figure 1

While it’s pretty easy to make a DC voltage inverter with circuits like the one in figure 1, I believe there’s a better solution: Single-supply op-amps are available with offset null pins, completely eliminating the negative power supply! See figure 2 for my (untested) circuit suggestion. TLE2021 chips are available in plastic DIP packaging from the usual suspects, such as Digi-Key and Mouser.
 

Figure 2

The current sampling resistor should be selected for full-scale reading with 10mV drop, which is 10X lower measurement burden relative to the original circuit.

If you add the input diode protection resistor back in, then you incur only 0.7mV max uncompensated offset because of the TLE2021’s enhanced performance. And, if you do this, then there’s really no reason to keep those diodes -- you can rely on the op-amp’s input pin protection circuit.