I want to measure a bidirectional current using a Shunt and Current-Shunt Amplifier in High Side topology.
The ADC will work at a lower sample rate (<10 kSpS) than would be required for the faster parts of the signal (up to 500 kHz). Therefore a low pass filter between Amplifier and ADC will be used to prevent aliasing.

  1. Would an Amplifier with a Bandwidth smaller than than the fast signal components but larger than the sampling rate suffice (without a noticeable impact on the measuring)?
    The bandwidth of most Current-Shunt Amplifiers seem to be in the range of 100 kHz to 400 kHz.
  2. Or do I need a Current-Shunt Amplifier with a Bandwidth of >= 500 kHz for this?
  3. I am not sure here if the Amplifier behaviour can be regarded as a low pass filter like the frequency response suggests. See the Gain vs Frequency plot of the LTC6104 for example (Page 4).

One reason I am asking is, that so far I only found Current-Shunt Amplifiers with a Gain of >= 50 but low bandwidth available in stock. The LT1999 for example has a high bandwidth and is available with a Gain of 50 but needs to be sourced directly from LT or as non-stock from Digikey.


2 Answers 2


What you need to do is put a low-pass filter on the input of the shunt amplifier. Then use a shunt amplifier with a bandwidth greater than the bandwidth of the signal you're going to send to the ADC (roughly 5 kHz in this case). Note that the amplifier should not be right on the edge in terms of frequency response. If you feed high-frequency signals into an amplifier without the necessary frequency response, you can get all sorts of weird, disgusting, and puzzling problems with the output. If nothing else, you can get input diodes or transistor junctions actually rectifying portions of the signal, and things just go downhill from there.

  • \$\begingroup\$ Following this a amplifier with lower bandwidth would need a larger footprint for the filter. The INA225 documentation shows the implementation of a input filter. What attenuation would be needed for the unwanted high frequencies? Possibly connected to a specific parameter of the amplifier? Out of curiosity: Could you recommend a source for further reading on those effects you described? \$\endgroup\$
    – Grebu
    Sep 21, 2015 at 18:37
  • 1
    \$\begingroup\$ There is no obvious way to tell what filter is needed. It depends on the input spectrum. The most common description of problems along this line is RF pickup, and I don't have a source handy. And yes, adding an input filter is a pain, but I've personally been burned on a current sensor on a suborbital package, and have become something of a fanatic on the subject. Of course, our package had a couple of RF transmitters which may have been the culprit(s). At any rate, it worked fine on the ground and was nearly useless in flight. \$\endgroup\$ Sep 21, 2015 at 20:30
  • \$\begingroup\$ Thanks for sharing your experiences. My sensing application will move slower and stay on the ground. But as the faster signals will be inherent, filtering will be inevitable. \$\endgroup\$
    – Grebu
    Sep 21, 2015 at 20:49
  • \$\begingroup\$ @Grebu - Actually, it grieves me to say it, but I've been burned twice, the second time on a thermocouple amplifier. As the saying goes, "Fool me once, shame on you; fool me twice, shame on me." So now I put a simple lowpass (using a ceramic cap for good high-frequency performance) on ALL inputs. And haven't been burned a third time. \$\endgroup\$ Sep 21, 2015 at 21:02
  • \$\begingroup\$ One never stops learning in a complex field like electronics. Luckily my PT100 input to an amplifier is already filtered. However it is naturally slow and therefore easy to filter. \$\endgroup\$
    – Grebu
    Sep 21, 2015 at 21:39

Your ADC running at 10ksps will only support a bandwidth of <5kHz in the signal.

To ensure your signal is clean for the ADC, you should select an amplifier that has a bandwidth higher than this (I would normally use a minimum of 10x, so 50kHz). There is no point in trying to 'follow' the highest frequencies in the current as you cannot measure them.

Make sure that the amplifier has sufficient slew rate to drive the ADC range you desire. Slew rate and Gain-Bandwidth product are two different things and are not synonymous with each other.

This is, of course, true unless you are using undersampling to capture the signals within this range where you would need an amplifier with sufficient bandwidth and slew rate to follow the signal.

If your intention is to measure the average low frequency currents, then your option 1 should suffice.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.