1
\$\begingroup\$

I am looking into using a shunt resistor to measure a current up to 30 A, that may be constant, or vary with a sine wave, square wave or even a PWM signal (up to 10kHz).

I understand there are Rogowski coils and Hall effect sensors, but if the resistance is acceptable for my uses, what other considerations should I be giving to the choice of current sensing?

\$\endgroup\$
5
  • \$\begingroup\$ Seems a little backwards. You look at all the considerations, and then determine if a resistor is right for your needs. Also, do you mean "series" resistor, which is the usual configuration for current sensing (for reasons)? \$\endgroup\$
    – Scott Seidman
    Commented Nov 12, 2020 at 14:09
  • \$\begingroup\$ It's a fair point, though I guess I'm not writing this question at the intuitive start of the design process. I am part way through specifying a system and starting to wonder if I'm missing any critical variable such as inductance/impedance, emitted noise, crosstalk etc. \$\endgroup\$
    – J Collins
    Commented Nov 12, 2020 at 15:04
  • 1
    \$\begingroup\$ Isolation from high voltages would be one consideration. \$\endgroup\$
    – user16324
    Commented Nov 12, 2020 at 15:18
  • \$\begingroup\$ You may wish to consider kelvin leads on your sense resistor. You mention elsewhere that your system is around 2kW and your resistor is absorbing only 200 mW. Kelvin leads help reduce the inaccuracies introduced by lead wires. Whether or not you need them will depend upon the level of accuracy you need. \$\endgroup\$
    – Math Keeps Me Busy
    Commented Nov 12, 2020 at 19:50
  • \$\begingroup\$ For the shunt resistor solution I am looking at something akin to uk.rs-online.com/web/p/shunts/8103267, this has the two sensing terminals that would satisfy the Kelvin lead condition. However I have updated my question with the current range, and have found something like this: aliexpress.com/item/32680512185.html which isolates implicitly. \$\endgroup\$
    – J Collins
    Commented Nov 13, 2020 at 14:58

2 Answers 2

Reset to default
4
\$\begingroup\$

You forgot fluxgate magnetometers and magneto-resistance based sensors for measuring current non intrusively. So, if you're measurement must be non-intrusive then the resistor is out, as it will introduce power and voltage losses in your circuit etc. However, if you must also measure down to DC then the resistor and Hall effect sensor might be the only choice, but the others are out. Then there is the complexity consideration vs accuracy required, and closely related the linearity required for the application in mind. Again, a resistor is very suited to linearity and simple (its rather the post-amplifier that introduces the non-linearities etc.). The others are going to add non-linearities at several stages of the measurement. And finally, if not the last, is the sensitivity (ie S/N) requirements that may or may not just be solved by high amplification. There may be others but generally it depends on the degree of "intrusivity" your circuit can bare, the complexity, the DC vs AC requirements and the sensitivity requirements (ie S/N). In any case it sounds like you may have already made up your mind for a good reason ...

\$\endgroup\$
1
  • \$\begingroup\$ I am working with a system that is around 2kW and the shunt resistance is only absorbing around 200 mW, so I think that is acceptable for my needs. You may instinctively say 'that's never going to work', and that's the input I'm interested in. Thanks for the other measurement types, I was unaware of them. Linearity is not a massive issue so long as it is repeatable and can be reasonably calibrated for. I think the main concern is with the 10 kHz PWM case whether the amplified output will remain accurate. \$\endgroup\$
    – J Collins
    Commented Nov 12, 2020 at 15:08
4
\$\begingroup\$

Isolation is often a factor, but it can be added externally.

Also, the voltage across very low value current sense resistors (shunts) is extremely sensitive to inductance at higher frequencies. Low inductance resistors made for sensing current are available, but everything of finite size (even a straight piece of wire) has some inductance.

If you only need to know the AC component of the current, you may get a much cleaner and more reliable signal by using a small CT (current transformer) as is often used in switchmode supplies. Isolation comes for free.

\$\endgroup\$
5
  • \$\begingroup\$ Since the first thing I'll do with the small voltage signal is put it into a differential amplifier, will this eliminate the need for isolation? Does the inductance come from the resistor being wound and building up a field in itself? Would the be mitigated by simply using a linear bar-type shunt or does the current even in that condition induce spurious voltages for amplification? It almost sounds like the belt-and-suspenders solution would have an AC solution and DC solution combined with low and high pass filters. This starts sounding like a proprietary component is an easier choice. \$\endgroup\$
    – J Collins
    Commented Nov 12, 2020 at 17:08
  • \$\begingroup\$ Further what would it take to model and correct for the inductance error? \$\endgroup\$
    – J Collins
    Commented Nov 12, 2020 at 17:12
  • \$\begingroup\$ Differential amplification is (probably) necessary but does not provide isolation unless it's a special isolation amplifier. \$\endgroup\$
    – Spehro 'speff' Pefhany
    Commented Nov 12, 2020 at 17:19
  • \$\begingroup\$ You're referring to isolation from the perspective of the potential of the high-current wire being very different to the potential of the amplifier/DAQ? \$\endgroup\$
    – J Collins
    Commented Nov 12, 2020 at 17:30
  • \$\begingroup\$ I don't have a crystal ball and cannot see your design, however isolation is frequently a requirement. \$\endgroup\$
    – Spehro 'speff' Pefhany
    Commented Nov 12, 2020 at 19:08

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.