Simple dual supply adjustable voltage reference

Hello.

I'm stuck working on a circuit, and I need some advice. I have a protoboarded circuit with a bunch of dual-supply amplifiers, and so it happens that I need a stable adjustable voltage reference that can span a few volts around GND i.e. +/- 5V. I don't need accuracy up to a millivolt, the important part is that the voltage is constant, stable and adjustable via a variable resistor (potentiometer). I've done some searching and I mostly found schematics for single-supply references. For example see the attached image below.

Circuir 1.a is a simple voltage divider. While this type of setup does offer an easy adjustable voltage, the problem is that it's very inaccurate and unstable especially when the system starts drawing higher current from it (drift). It gets better when the resistor values become lower, but also draws more current. So some people suggested that an additional buffer (voltage follower) be added to it.

Circuit 1.b shows such buffer present behind the voltage divider. The circuit has some additional decoupling/bypass caps for better stability, and it has also been suggested to add a resistor to the negative feedback loop to prevent oscillations when dealing with capacitive loads. There should be similar resistors on both opamp inputs. We can also use higher resistor values since the current is no longer an issue here. This setup works much better, but the circuit still has stability issues. For example, if the positive power supply rail encounters some noise, it will get amplified onto the reference volrage output.

Circuit 1.c shows another improvement. A constant current source is used on the voltage divider to eliminate positive rail supply noise.

Now all the circuits decribed so far are for single power supply. It happens that I am using a dual power supply of +/- 9VDC, and thus I need an adjustable voltage reference that can span for about 5V in each direction from the GND terminal. Circuit 2.a shows a dual-supply equivalent of the unstable circuit 1.a. Circuit 2.b is a dual-supply equivalent to 1.b with similar stability issues. What should I use to achieve dual-supply equivalent of circuit 1.c that has similar stability? I believe I can't use a constant current source between V+ and V- to eliminate power supply noise because if I eliminate it on the positive rail, it can still enter via the negative rail and vice-versa.

Suggestions?
Thanks in advance!

Hi,

If you need it as a reference, then consider to use a fixed voltage reference IC.
The voltage is not that important.

Lets say you have a 1.25V reference IC.
And you want an adjustable output of +/-5V. (Range = 10V)

Then use an inverting Opamp circuit.
You need a gain of 10V / 1.25 V = 8

Instead of connecting the +In to GND --> connect it to 0.555V.
Generate the 0.555V by a voltage divider from 1.25V Ref. Resistor ratio 1 : 1.25.

Use a pot. One side to GND, one side to 1.25VRef. Wiper to the inverting input resistor.

To suppress nois I recommend to use a capacitor at In+ and a second capacitor across the feedback resistor.

Klaus

Hey, thanks! That's an interesting proposal, and I just learned something new today.

For an operational amplifier here's the equation I came up with: Uout = U2 * (1 + R2/R1) - U1 * (R2/R1)
If we connect U1 to GND, the equation reduces to Uout = U2 * (1 + R2/R1) which is an equation for a non-inverting amplifier.
If we instead connect U2 to GND, we get Uout = - U1 * (R2/R1) which is again the equation for an inverting amplifier.

In my case I have two options - I can connect either input terminal (U1 or U2 on the image below) to a reference voltage and the other one to a potentiometer with a constant current source. If I want the output voltage to span from -5 to +5V, and the adjustable voltage on the potentiometer is 1 to 4V, I get two solutions.

A) U1 = 3.57V, R2/R1 = 2.333
B) U2 = 2.00V, R2/R1 = 3.333

The problem is I can't find any +3.5VDC voltage reference IC's, so it's likely I'll go for the second setup where U2 is connected to a +2.0VDC voltage reference. We'll see how this setup behaves. I also got the idea to place the potentiometer in place of R2, and I could control the output voltage by changing the amplifier gain.

Cheers!

Two zener diodes, joined with 0V ground between them. You now have two stable V references, relative to ground. Then a potentiometer can dial any level between +6 and -6V.



Some 'noise' is simulated on each supply rail. However the positive reference is regulated by its zener, and the negative reference by its zener. Therefore no noise gets through to the regulated output.

Neato, that's an even simpler solution. I've already implemented it on a protoboard with two 5.1 zener diodes and a buffer amp (TL071).
It works perfectly, but I'm using two 4k7 resistors in place of the 600 ohm ones to reduce the current a bit and a 100k potentiometer.

Cheers!