How to optimise this dual (bidirectional/bipolar) opamp+BJT voltage controlled LED driver?

I'm working on a dual, voltage-controlled LED driver circuit, where one LED's brightness indicates positive voltages at the input, and the other LED indicates negative voltage.

The purpose of the circuit is to provide visual indication, so the current/voltage relationship does not need to be incredibly precise.

This is where I am at so far:

• There are no microcontrollers present in the design.
• The control voltage for the LEDs are a combination of 2 or more analog signals which once summed by OA1, range from -2.5V to +2.5V for driving the LED current sources. Thus, a summing node is necessary in the circuit (somewhere) before the LED current drivers. I have shown this simplified in the schematic below.

simulate this circuit – Schematic created using CircuitLab

A number of bidirectional VCCS schemes were discovered, such as:

...however, the project which this indicator circuit is for implies some constraints, which seem to indicate the above ideas will not work:

1. The +5V rail for the LED supply current must be isolated from the lower current analog areas of the circuit. While the control voltage signal is not terribly high in frequency (maximum is within audio frequency range), the switching of that amount of current on the main analog +/-12V supply cannot be tolerated in the project.
2. There is no negative supply rail available for the LEDs' current sourcing/sinking.
3. The maximum of 8mA current cannot be supplied by the opamps themselves (unless somehow supplied by the isolated, single +5V LED supply) as this circuit will be present in multiples throughout the overall project, and the +/-12V rails cannot be used to supply all that through the opamps.

Is there a more elegant and optimised solution to this problem?

Perhaps one that negates the use of so many opamps, or a VCCS (opamp+transistor, or otherwise) combo which rids of the extra inverting buffer in the positive section?

• Your circuit is quite "crude" and misses some essential details like: no current limiting resistors for the LEDs, no base resistors for the NPNs, when the opamp's outputs outputs a negative voltage the output is shorted by the diode. You're mistreating your opamps, LEDs and NPNs. – Bimpelrekkie Apr 24 at 14:01
• @Bimpelrekkie Thank you for pointing that out. I will add base resistors for the NPNs, and current limiting resistors on the LEDs. Do you have some advice for me on the overall solution? – lhibouambigu Apr 24 at 14:06
• @lhibouambigu One opamp could do it, I think. You won't like it, though. Use something like the LT1677 and put the LEDs in series with the +5 and ground power points. So the opamp itself is running on 5 V and not on your sensitive sources. Force the output to sink or source what you want using a divider. Feedback with resistor (and cap) to (-) in. Set (+) in to 2.5 V. A few more details. But it could be made to work. (Just simulated it and came up with this, without thinking, for the two LEDs vs your source voltage: opamp with LEDs in power rails.) – jonk Apr 25 at 5:51