# Measuring a divided differential voltage into a single ended DAQ

I am looking to log a high sped voltage (20 kHz) +-60 V, into a single ended DAQ at +- 5V.

Is the following circuit a good example of accurately measuring a differential voltage or is it overly complicated? It is derivative of some diagrams I have seen.

simulate this circuit – Schematic created using CircuitLab

On Andyaka's suggestion I present the simplified system below.

simulate this circuit

• @Andyaka I'm not at all following how your arrangement of zeners and resistor would offer a +- 10 V capping on the signal. Surely the 100 Ohm resistor would want to be in series after the amplifier output and the zeners in parallel. Does that make sense? Dec 11, 2020 at 15:47
• Not sure which bit you are struggling to follow - it's called an AC zener clamp. Does that link help? Another link. Dec 11, 2020 at 15:50
• Damnit I think you're right on the diodes, I'm confusing the role of the diodes. It looked like in series they would only ever clamp on +- 0.7V, but that would actually be the case if they were parallel and inverted. I think question still remains on the 100ohm resistor, the amplifier output only wants the 1kOhm resistor between it and the negative input, while the 100 ohm resistor only clamps the output of the amplifier downstream to the DAQ. Dec 11, 2020 at 16:11
• The internal guts of an op-amp could easily be represented by an output impedance of a few hundred ohm yet, due to negative feedback, the output appears to be a great voltage source. The 100 ohm adds to the internal resistance just to provide a little extra current limiting and saves power being dissipated totally inside the op-amp should the output be clamped by the zeners or some fool shorts the output. Hope things are good with ya! Dec 11, 2020 at 16:17

In the 2nd circuit I've removed resistor component values because they are not correct. I've also removed the op-amp type number because it is not a rail-to-rail type. I have also removed reference to +/- 5 volts on the power rails because not even a rail-to-rail op-amp can ever produce an output voltage of +/- 5 volts from +/- 5 volt rails.

• You would choose R1 and R6 to have the same value
• You would choose R5 and R7 to have the same value
• You would choose the ratio R1/R5 to be big enough to cope with the +/- 60 volt input signal
• You would choose power rails of +/- 10 volts and have many op-amp options open to you.
• You WOULD NOT connect measurement ground to machine ground

Protecting the DAQ input

Use a 100 Ω resistor as shown and connect the output (aka measurement voltage) to the DAC with an additional pair of 6.8 volt zeners in series opposition to measurement ground. The value for the zeners depends on the DAQ limits.

• Good point on the power supply, that was an oversight, I have access to +-12 V. The idea of the first amplifier was to buffer the divided voltage before presenting it to the second differential amplifier, perhaps this was a bad idea. Is R1/R7 your effective divider voltage? Also had I connected the grounds or is that just general advice? Dec 1, 2020 at 17:04
• R1/R7 does the same job without introducing an op-amp delay, an extra resistor ratio error or an op-amp offset voltage error. +/- 12 volts is good but you might need to limit the output using back-to-back zeners to prevent possible damage to your DAQ input. Dec 1, 2020 at 17:07
• I wanted to think about Zeners, wasn't sure how they would impact the precision signal. You think they would be fine? Essentially, have two inverted 5V Zeners pulling to measurement ground? Dec 1, 2020 at 17:11
• Back-to-back in series from the output to measurement ground. Feed the op-amp output to them via 100 ohm and then onto the DAC. If the 100 ohm gives you an error, build it inside the negative feedback loop. Dec 1, 2020 at 17:12
• OK done @JCollins - I'd be looking at these AD867x op-amps as a first choice option because they have great bandwidth and are low noise. You might also choose to raise the 11k resistor and 1 k resistors in value by about 3 because they will get a tad warm on 60 volts. I'd probably also consider using 0.1% tolerance values too. Dec 1, 2020 at 17:41