How to have an opamp's input voltage greater than the supply voltage of the opamp itself

I'm making a board that has to measure some voltages, let's say a very simple multi-channel oscilloscope with a very very low sample rate (about 10 Hz).

To measure voltages I use opamps as buffers and the integrated ADC of an MCU. The power supply of the board is 3.3 V, both for the analog part and for the digital part (separate, but both at 3.3 V).

The voltages I have to measure vary between 3 V and 30 V. My first idea is to use a voltage divider that with the maximum input (30 V) makes the opamp input at 3.3 V.

Now, the ideal would be to have a very high resistance to draw very little power from the analog input, but if I use a large resistor for the divider (say 1 Mohm) I could introduce a lot of thermal noise.

The ideal would be to be able to enter the opamp (powered at 3.3 V) directly with the 30 V input. This would need a circuit configuration which allows an opamp's input voltage to be greater than the supply voltage of the opamp itself.

Is there any circuit configuration that allows to do this?

• I'm trying to think of a configuration that wouldn't either require a voltage divider on the input or the opamp to be able to raise its output voltage to 30V (and therefore be powered from 30V) and I'm not coming up with anything. What are your requirements for maximum current from the 30V rail and noise?
– vir
Apr 5, 2023 at 21:00
• What is a lot of thermal noise to you? A voltage divider using a 1M input resistor with a 120k resistor is around $40nV/\sqrt{Hz} = 0.13\sqrt{120k || 1M}$. Since the bandwidth is low, a capacitor across the 120k resistor will reduce the total noise to perhaps something tolerable.
– qrk
Apr 5, 2023 at 21:54
• Can it only sample at 10 samples per second or could it sample more often and average the readings so as to reduce the noise? Apr 6, 2023 at 17:06

What you are looking for are called "over the top" opamps. These are opamps that can work with input voltages well above their rails. They have a special input stage that allows this to work.

The LT6015 is one example.

My current project uses a couple LT1490 to monitor battery voltages ~ 24 V using 5 V rails.

Please read the datasheet carefully to see if something like this will work in your application. Keep in mind, the output of the opamp will be limited to its rails.

That said, a resistor divider into a buffer isn't necessarily the wrong way to go here. Without your design requirements (and other circuit details), we actually don't know if that will contribute too much noise or not. HINT: The ADC integrated into your MCU will probably be the biggest noise source, and will make your resistor noise look like nothing.

• How do those work, internally? ...Perhaps I'll make my own question about that, actually. Apr 6, 2023 at 1:54
• If such an "over-the-top" op-amp were to be used in a design, would that eliminate the need for a protective voltage divider between the input and the op-amp? What would be the advantage of using an "over-the-top" op-amp, compared to the more traditional design of a voltage divider followed by a regular op-amp (i.e., your "resistor divider into a buffer")? Would the single "over-the-top" op-amp be less noisy? (Obviously, I understand that the noise introduced by a resistor divider is almost certainly irrelevant compared to the MCU's built-in ADC, but considering special cases where it isn't.) Apr 6, 2023 at 8:00
• @CodyGray A typical use-case has the output driving the gate of a MOSFET, which allows the feedback network, along with the inputs, to be at voltages much above the rails. Apr 6, 2023 at 13:38
• @Hearth You may find this interesting: analog.com/en/technical-articles/… Apr 6, 2023 at 13:40

As a previous answer from @evildemonic stated, you can use these special kinds of op-amps.

But lets look deeper:

(1) Input resistance in a voltage divider scenario (1MOhm) will introduce no significant noise into your application. Please see Paper on R-Noise IV. Discussion.

(2) Significant means compared to your On-Board ADC. Per the Paper, Fig. 4 values of -50dB can be expected for real-world resistors. I would - without sources - suggest, that a common 12Bit MCU-ADC will have far greater effects. Please see Analog Devices Article

(3) I would suggest using a 900k/100k input-divider using "good Temp-Co, low noise and low drift resistors". I would then feed the signal into an active RC-Low-Pass Op-Amp and then into the ADC. Using the active RC-LP can reduce effects of thermal noise, as the bandwith is decreased. Make sure to select a "good" Op-Amp if noise is of concern.

• very interesting link Apr 6, 2023 at 19:59