# Measuring accurate voltage across a shunt resistor for high frequency AC current sensing

I am trying to measure the voltage across the shunt resistor to then be able to measure the flowing current. The output voltage will then be fed to a microcontroller or other ICs for further analysis. The signal has an amplitude of 60V and a frequency of 1MHz.

I was initially thinking of using an op-amp in a non-inverting configuration but I am concerned about the ground connection at the output of the transformer. If no galvanic isolation is present, would it be a cause of concern when interfacing with a uController?

Would a difference amplifier be more suitable for this application?

simulate this circuit – Schematic created using CircuitLab

Which configuration suits my case the best? Are there any better ways of doing this?

Thanks in advance for your help!

## 2 Answers

It all depends on your transformer. Note that the opamp tolerable voltage range and voltage supplies must be larger than the largest voltage swing you expect at the output. This circuit is configured with a 1:11 voltage gain, and a less than +-5 volt (lets say +- 4 volts to be nice to the opamp, as not all opamps are rail-to-rail) output swing based on the power supplies.
The output of the opamp being ~8Vpp means that the transformer should somewhere in the range of a 1:100 kind of thing. (60V / 100 * 11 = 6.6V). Also for a gain of 11 times a bandwidth of at least 2MHz, your minimum gain-bandwidth product for the opamp has to be at least 22MHz.

I guess that this voltage division is done with the transformer plus "LOAD" and "shunt" resistors? If so, the LOAD should represent like 0.99 ohms given a 1:1 transformer. It will be difficult to find those resistors (and pricey). They will also load the circuit you are measuring by a lot!. It would be better to have a 1:10 or 1:100 transformer and a lighter load, as this affects your measurement much less.

Imagine your SIGNAL voltage source, in reality it always has some series resistance, and you don't want to change the system you are measuring. If you load it with 1 ohm (as in a 1:1 transformer with your current schematic) a 10-milliohm thevenin resistance in series with your signal (that's like the contact resistance of a good connector) will change your output voltage by 1%. In other words, this is not the best circuit.

I would drop the resistor divider and just use a self-wound 1:100 transformer, but also add some protection to the opamp input. This would only load the circuit with whatever parasitic capacitances there are in the transformer.

You didn’t state your accuracy requirements.

A current transformer may be an option which handles isolation and amplification (to an extent)but won’t be super accurate.

Single ended measurement is ok if accuracy is around 1% otherwise ground bounce may become an issue and require a difference amplifier topology. This can be an IC or a matched resistor like ACAS with any opamp.

If single ended inverting is preferred unless you have a reason to need non-inverting.

Count your gain bandwidth, you’ll need a lot to accurately track 1Mhz. Decompensated opamps may be useful (though rare).