Best way to clip a circuit

Question

I have an analog circuit that takes a ±15 V DC voltage and some input signals, does some things to the signals and outs an analog signal via an op-amp. I would like to limit the output voltage to ±10 V to protect circuitry down stream. I am considering using a rail-to-rail op-amp, supplied with ±10 V to do this. Is there any reason not to do it this way? Or should I just stick with diodes?

Sorry if this is a duplicate question. I searched the site, but couldn't find this exact question.

Update:

Thanks for all the great responses. At least one person asked for a schematic. Below is partial schematic showing the last part of the circuit (there could be IP in the rest of the schematic). w123 is the signal input. It is averaged through an active RC filter and is sent to three op-amps with different gains (final amplification to be determined. May be 1x-100x). The idea being that we can select an appropriate gain based on the signal (kind of like selecting a magnification on a microscope). When the signal is low, the high-gain output can be used, when it is high, the low-gain op-amp can be used. I don't plan on using the signal close to the rails: I just need to limit it to protect a multiplexer and an ADC down stream (partial schematic below).

Answer 1

If you want a simple voltage clamper try this dual zener circuit: -

Image from here. However, if you need precision clamping read on...

This circuit provides a precision voltage clamper. As shown below it has a positive clamp voltage of +5 volts and a negative clamp voltage of -1 volts but, this can be easily adjusted: -

• U2 clamps the positive peak to 5 volts
• U3 clamps the negative peak to -1 volt

The clamp values can be adjusted by V3 and V4.

Image from this answer. Another example of basically the same circuit can be found here: -

Answer 2

I am considering using a rail-to-rail op-amp, supplied with ±10 V to do this. Is there any reason not to do it this way?

It could work and could be a perfectly OK way to do it. There are two things I can think of to consider.

1. You have to somehow be sure that you won't exceed the allowable input voltage range for the op-amp. Typically op-amps do not want to see input voltages exceeding their rails. Sometimes the range is more restricted than that. So check your datasheet.
2. Saturation effects. If the op-amp is driven hard so the output slams into one of the rails, some of the internal circuitry may be in deep saturation, and it may take a bit of time to recover when the input signal comes back in the normal range (the range that calls for the output to be within +/- 10 V).

I am not too sure how much of a problem 2 will be with modern CMOS rail-to-rail op-amps. I am thinking it won't be much of a problem. But it is something to keep in mind.

So I say go for it.

Answer 3

Using a rail-to-rail op-amp is a good idea -- but don't bother making extra supplies, use the 15 already available. This gives a 15V output range, too much, right?

Simply reduce the output with a 1:2 resistor divider (e.g. 4.99k + 10k). (You will need an extra 33% gain going into this limiter, of course.) Then add a second op-amp as voltage follower (buffer). If the output is feeding a high impedance, the buffer can even be omitted. Or if the load impedance is a fixed resistance, it can be combined as part of the resistor divider.

Answer 4

I frequently use the following circuit to clip +/-5V external signals to a "human controllable threshold".

The circuit is configured to x2 amplify the threshold signal (V3, in your case +5V) and then provide the inverted output as well. This gives the rails: Threshold2 and Threshold(-2) Then, the diodes limit the signal voltage to +/-(Threshold*2 + V_DiodeDrop).

CAUTION: Can your op-amp source/sink the required currents to act against the signal source impedance?

I usually want the 200-700mV (There are some ~50mA shottky SMD-diodes out, which only have 100mV drop. I usually go with the cheapest what ever diode rated for the voltages and currents involved) extra range on the signal. If you dont want it, you can just connect the op-amp feedbacks "before the diodes".

NOTE: Usually the threshold-voltage (V3) is generated in my designs via a HMI rotary switch providing a switch-able resistor divider. This is the reason, i invert the already amplified signal and do not use a x(-2) op amp (Lower imput impedance changing the equation for the voltage divider!)

NOTE: This implementation has some issues with very high-frequency signals due too the limited slew-rate. But 100kHz bandwith is not an issue with common parts - I mainly use it for sensor signals or control signals within 10kHz bandwith.

PROS: The actual signal is not feed through an extra op-amp (introducing non-linearity, delay, phase-shift, etc.) and is only affected mildly by the diodes in normal operation (Signal-source has source impedance, but leakage-current thorugh the currents is somewhat low!). Also, by not using a "rail-to-rail" op amp, non-linearty and saturation effects on the signal are avoided - this is can be important as "rail-to-rail" can be an issue even with modern cmos-amplifiers.

CONS: You need extra components, extra board space, extra validation steps and so on. You can only limit signals with a fairly high source-impdance with common op-amps.