Can an op amp Vin+ and Vin- be outside the supply voltage so long as their difference is inside it?

Question

An op amp amplifies the difference between signals on Vin+ and Vin-. If I wanted to sample the differential voltage of a pair of signals whose common mode is outside and potentially significantly outside the range of the supply voltage, can op amps natively support this?

I understand professional oscilloscopes can accommodate this; what circuitry exists in oscilloscopes that allows referencing a seemingly arbitrary ground and comparing a signal to it? Will an oscilloscope 'clip' or have some other behaviour outside a range or is there a circuit that is essentially immune to common mode?

Answer 1

Generally speaking, no. The datasheet should clearly tell you what is permissible without damage and what might work and what is guaranteed to work (six different numbers in general).

Both inputs of the op-amp should be inside the supply range. Many op-amps can tolerate inputs slightly outside the supply range (slightly being hundreds of mV) and not be damaged and some still work for small excursions, especially below the negative rail for "single supply" op-amps and outside both rails for rail-to-rail input op-amps. But only slightly.

For example, the popular MCP601 op-amp (at 25°C) is guaranteed to work with inputs as low as -300mV wrt the negative rail and as high as 1.2V lower than the positive rail.

The limits on damage (not functional) are found in the absolute maximum ratings and involve scenarios such as with or without current-limiting resistors.

A very few boutique op-amps such as the LT6015 will work with inputs much higher than the positive supply rail (up to +76V wrt the negative rail, regardless of the positive rail in that case).

Many op-amps, especially older ones (LM308, LM741 etc.) , won't reliably work unless both inputs are 1 or 2V higher than the negative rail and 1 or 2V lower than the positive rail. That should be your default assumption unless the datasheet provides better news (and the numerical part using min/max headings, not the 'typical' figures or the advertising blurb at the top)

---

Most oscilloscope inputs have an attenuator ahead of the internal supply voltages and the input is relative to the circuit ground which is typically also earth. So the input may be +75V relative to ground, but it's only a few volts by the time it gets to the active amplifier. A few, such as my Tektronix TPS2024, have fully isolated channels and can tolerate hundreds of volts with respect to earth. You pay dearly for that feature.

Answer 2

No, op-amps cannot operate correctly with either of their inputs outside of their supply rails.

The internal transistors must be operating from a supply higher than their base/gate voltage to function correctly. Driving the inputs above or below outside the supply rail range most likely damage the op-amp, though it depends on the part and the driving circuit.

Op-amps not specified as rail-to-rail in both their description and their electrical characteristics need their input voltage range to be inside their supply voltage range by a margin.

Oscilloscopes have potential dividers on their channel inputs to bring the applied voltage into an acceptable operating range for their internal circuitry. It is the divided voltage that is measured.

Answer 3

Typically, op-amps do not work properly if either of the inputs lies outside of the range of the power rails. [In fact, many op-amps are not rail-to-rail, and their proper operating range is smaller than the range of the power rails].

However, there is a circuit that can be made from an op-amp called a differential amplifier, which can amplify a differential signal where the voltage of one or both of inputs lies outside of the range of the power rails.

A differential amplifier typically looks like this:

^{simulate this circuit – Schematic created using CircuitLab}

R1 and R2 serve as a voltage divider, so that even if Vin+ is outside the range of the power rails, the voltage at the op-amp non-inverting input may be within the proper operating range of the op-amp.

Because the op-amp is configured with negative feedback, as long as the differential input is not large enough to saturate the op-amp, the inverting input of the op-amp will be at virtually the same potential as the non-inverting input. So, if the non-inverting input at the op-amp is within the proper operating range, the inverting input will be as well.

Answer 4

Definition: A channel is the hardware input on the scope a probe plugs into.

I understand professional oscilloscopes can accommodate this, what circuitry exists in oscilloscopes that allows referencing a seemingly arbitrary ground and comparing a signal to it?

No. Most scopes do not have isolated channels. Not even the $20k scope I have at work with 8 channels. And students blow scopes thinking they do by trying to measure across two different components simultaneously where they connect the probe ground clips to two different circuit nodes. The ground clip on all the probes is the same node unless you know otherwise.

So beware when connecting passive probes on a wall powered scope to a circuit powered from the AC mains that does not use a transformer. The reference node has already been decided for you since probe's reference clip is already connected to something in the circuit through the scope power supply.

What you are thinking of are either isolated probes or differential probes. Isolated probes actually do float.

Differential probes do not but instead amplify the voltage difference relative to the scope channel's reference and pass it on to the scope, often stepping it down first with a divider so high input voltages can be brought into range to be measured. So for my work scope we had to get a bunch of differential $4k differential probes for each channel we wanted to measure with a different reference. Ouch. This is what is inside a differential probe:

https://circuitcellar.com/research-design-hub/high-voltage-differential-probe/

See link for circuit description. Enough is provided to build your own. Basically a symmetrical or balanced resistive divider in parallel with a capacitive divider of the same ratio for frequency compensation (so the R in the divider doesn't kill bandwidth by making an RC filter with parasitic and input capacitances) followed by a differential amp.

---

Or the scope power supply is outright isolated from the circuit supply so they can float to whatever you want for ONE reference among all passive probes since input channels still share the same reference. (Of course, differential probe or isolated probe connected still allow you to measure with different references).

1. Either the scope or circuit is battery powered and the other is not
2. Scope and circuit are powered from different batteries
3. Either the scope or circuit is powered from an isolated supply and the other is not
4. Scope and circuit are powered from different isolated supplies

---

Or, in the rarest of cases, the scope has isolated channels and each channel can be connected to a different reference node even if all passive probes are used. Battery powered, portable scopes have a higher chance of being this due to their expected usage.

Answer 5

Generally no, what you're describing is points 2 and 3 being raised higher than Vs+ if you were using a BJT op-amp. In this case, Q1 and Q2 allow current/voltage to flow upwards, but you have Q8 as a diode-connected PNP. This will act like a revere biased diode and will not allow any current to flow. Q9 also will act like a revere biased diode so nothing really happens. If you go high enough, you'll damage the part.

On the other hand, what you've effectively done is turn this into a current controlled amplifier because your input currents will flow through the bottom portion of that stage. Everything else in the amp would work as expected. In this case, there's actually a good possibility what you're after will work to some degree. It's outside the spec and I wouldn't recommend it, but if you're looking for a current-current amp (vs a voltage->current amp) then this could still work. Also note that your gain may suffer greatly because you've lost any amplification at the first stage.

Note this will definitely not work with a MOSFET based op-amp because the input gates don't allow any current to flow and the raised input voltages will simply pinch off the supply currents which will prevent anything from happening.

Further, you would need to review the internal spec of whatever BJT op-amp you have to determine if it's feasible just as I did in the second paragraph. Still... I would not recommend doing this.