I have two methods of achieving this and am trying to decide which one will be a better option for a stable output. The supply voltage is a linear regulator output, so it is pretty much constant.

Option 1: Using op-amp with voltage divider using 1v Zener diode (TZMC1V0-GS08) enter image description here

Option 2: Using op-amp with voltage divider circuit Using op-amp with voltage divider circuit

The adjustable range should be at max +/-2V, at this voltage having an accuracy of 0.01% is much preferred The output in both cases will be connected to one of OPA875 inputs which has an internal Av=+2. I guess in this case, the output current demands are negligible

Let me know if you have any questions


  • \$\begingroup\$ another isse in the top one, is R10,R11 = 50ohm look like they will reduce range of adjustment too much \$\endgroup\$
    – Pete W
    Feb 4, 2021 at 13:59

2 Answers 2


See the Update below.

Nowhere in your post do you say what your accuracy and stability requirements are, or anything about power budget, cost budget, output current, noise bandwidth, etc.; so all of this is guesswork.

The first circuit will not work; the opamp output shorted (through R12 and R13) to what I think is a GND symbol drawn incorrectly. Also, low voltage zener diodes are not nearly as stable as 6.2 V - 6.8 V parts. But no zener diode can compete with a high-accuracy and high-stability reference IC from National Semi, Analog Devices, Linear Tech, or Maxim. Yes, I know some of those companies have been bought and sold, but the old names are better.

In terms of accuracy and stability, the second circuit is much worse. It depends on the performance of two different voltage regulators, neither of which is designed for reference duty, plus the thermal performance of three resistors. Compared to an old, cheap LM4040, its output will wander around every time the door opens or someone walks by.

I would start with the first circuit, go with only one reference diode, and (if you really need a bipolar output voltage range) a different opamp circuit with gain that ranges from +1 to -1 with a single pot.


Here is a schematic from an old issue of Electronic Design for an opamp circuit that goes from positive to negative gain with a single pot adjustment. For your application, ignore the A2 circuit

enter image description here

The gain of the A1 stage goes from +0.5 to -0.5. With a 4.096 V reference IC, the output will swing from +2 V to - 2 V. Note that the source impedance at V1 must be zero ohms, or very low relative to the other resistor values. And, of course, use something other than a 741.

You still have not told us the output voltage range you need, so, again, this is mere speculation.

  • \$\begingroup\$ Thanks for your response, R12 and R13 are there only as place holders, they are not actually being used. In other words with this op-amp they are DNI. Sorry for not being clear there. Also these circuits are old ones that I am reviewing while upgrading the electronics in our systems. We do need bi-polar reference output. \$\endgroup\$ Feb 4, 2021 at 14:15
  • \$\begingroup\$ Updated the answer with a schematic. \$\endgroup\$
    – AnalogKid
    Feb 4, 2021 at 15:29
  • \$\begingroup\$ The ideal range we need is +/-1V, but +/-2V would work too. Please refer to my original post as i updated with more details. \$\endgroup\$ Feb 4, 2021 at 15:57
  • \$\begingroup\$ Also, thanks for the reference schematic, will give it a try \$\endgroup\$ Feb 4, 2021 at 16:06
  • \$\begingroup\$ Since the output is adjustable with a pot, the accuracy of the initial value of the reference is almost irrelevant. Way more important are temperature coefficient and long-term drift. The LM4040 (and just about any decent line of reference ICs) is available in 2.048 V and 2.500 V to meet your range requirement. Other than that, shop around for stability; there are a ton of parts to choose from. Here are some thoughts from TI. ti.com/lit/an/snva510b/snva510b.pdf National used to make the LM399, a single-chip oven-stabilized voltage reference. Excellent. \$\endgroup\$
    – AnalogKid
    Feb 4, 2021 at 18:01

Frankly, they're both pretty awful compared to what can be achieved with the cheapest of components.

I suggest using 2x TL431 (cheapest) or LM4040 (better-est) and a similar circuit to the top one.

The 1V zener has a tempco of ~`0.25%/K which is -2500ppm/K, so it makes a better temperature sensor than a reference. The LM4040 and TL431 are 10 or 20 times better and much more common parts.

You should also try to reduce the influence of the pot element resistance by shunting it with a resistor or connecting it to 'stiff' voltage sources and sorting out the voltage range with your op-amp. Your resistor values don't make a lot of sense, so I am assuming they are placeholders.

Using the supply voltages as a reference (bottom schematic) is possible, but usually undesirable.

You also need to decide whether you want to bypass to ground or to the minus supply. Usually (almost always) ground is the correct choice but the bottom schematic does not show that.

  • \$\begingroup\$ Thanks for your comments, as I was stating above, these are old schematics that I am taking over to rebuild the electronics that these are being used in. Will look into your suggestions \$\endgroup\$ Feb 4, 2021 at 14:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.