How to measure large DC current

LowQCab

Joined Nov 6, 2012
4,026
You attach a Fuse-Holder, containing a 1-Amp-Fuse,
to each end,
of the longest run of wire that is carrying the Current You want to measure.

Then determine the Wire-Gauge, or Copper-Area,
then look-up the Resistance per 1000-feet,
then estimate the number of feet between Fuses,
and then calculate how many Ohms that length of Wire should be,
then use Ohms-Law to figure-out how many milliVolts will equal 1-Amp.

Attach the 2-Wires going to the Fuses to a cheap Digital-Voltmeter,
you will have to use both the Inverting, and Non-Inverting, Inputs on the Meter,
Power the Voltmeter with a Battery.

After the initial calibration with a known-good Amp-Clamp Ammeter,
you will have a very accurate, Wired, Ammeter.
And even if it isn't 100% accurate, it will give very consistent readings
.
.
.
Ammeter Wiring  Flat .png
 

LowQCab

Joined Nov 6, 2012
4,026
The proximity and orientation of a Hall-Sensor to a Conductor is extremely CRITICAL to its operation.
But if the Wiring can be temporarily disconnected,
a Hall-Sensor with a Factory-Calibrated Gapped-Toroid-Core could be slipped over the Wire,
providing a very nice solution.

What will you bet, that they don't want to disconnect the wiring ????
.
.
.
 

Ian0

Joined Aug 7, 2020
9,671
You can buy current shunts on Ebay at very decent prices
https://ebay.us/XVdEFI
The one I bought was perfectly accurate.
Industrial ones are about £15
It produces an output voltage of 60mV at the rated current.
Amplify that with an AD8418 or INA180 to get a usable voltage.
Shunt+shunt amplifier gives a better offset and lower noise than Hall-effect (so if you also need to measure small current in the same circuit it is the best method).

Hall effect
see https://www.lem.com/sites/default/files/products_datasheets/has_500-s_sp58.pdf
are galvanically isolated, and don't require you to terminate the high-current cable twice! (but the output signal is a bit noisy, and there is a higher DC offset than with the shunt. If you are not interested in high frequencies, you can filter out the noise.
If there is a time when you know that there is no current through the sensor, you can sample the output voltage and save it, then subtract it from the reading to null the offset.

@LowQCab 's method using a length of cable can be very useful, because you have probably already got the cable and you can just measure the voltage drop, but the voltage drop is proportional to absolute temperature. (And don't forget the fuse, because if the main cable comes adrift then all the current tries to go through your sensing circuit). You can use a device such as the AD8418 or INA180 to amplify the output.

For 300A you will probably be using 120 mm^2 cable, which will have a resistance of 141μΩ per metre. A metre would give a voltage drop of 42mV, amplify by x50 and you have a usable 2.1V. It will drift by 0.3%/°C. Don't forget you need to arrange the metre of cable to minimise its inductance, otherwise you will have misleading results on any superimposed AC.
 
Last edited:

Irving

Joined Jan 30, 2016
3,845
Assuming DC and a relatively short distance I too like the idea of using the existing cable. If >50V you should preferably do this on the low-side cable to avoid common mode and safety issues and use an instrumentation amplifier or current sense amplifier to give stable fixed gain between 10 and 1000 as needed. You may need some RC low-pass filtering to avoid RF pickup if your sense wire is more than a few CM long.

If temperature compensation is an issue add a couple of k-type thermocouples, one for ambient and the other strapped to the cable. The thermocouple outputs, between 1 & 5mV or so can be amplified by similar devices and fed to a couple of spare ADC channels or use a proper thermocouple compensator/amplifier setup (eg LT1025).
 
Last edited:

Ian0

Joined Aug 7, 2020
9,671
Bear in mind that the resistance of cable probably won't be what you'd calculate from the resistivity of copper, it will be the maximum value that it could possibly be and remain legal. The relevant standard here is BS6360, which specifies 120mm^2 cable at ≤153Ω/km @ 20°C (British "room temperature" is 5°C lower than American).

Probably the best suggestion here is to take the path of least inconvenience! There's no "right" answer.
If absolute accuracy is important, the go for the shunt. Trying to compensate the resistance of cable changing with temperature seems like a lot of bother.

Using a very short piece of cable and a lot of gain will maximise errors due to amplifier offset, which is important if you need to measure low currents, but not if you are only concerned with >100A.

If you use a longer piece of cable, DON'T be tempted to coil it up! The inductance will make you think every slight change in current is a massive spike!
 

Thread Starter

LAOADAM

Joined Nov 21, 2018
862
With a real big multimeter the size of a elephant. You know what you're asking is very very dangerous do you have something specific you wish to measure so we can guide you safely
Thanks.
WOW.
I heard the DC motor has a very large start current, and I gonna deal with a 2 KW dc motor.
 

Thread Starter

LAOADAM

Joined Nov 21, 2018
862
You attach a Fuse-Holder, containing a 1-Amp-Fuse,
to each end,
of the longest run of wire that is carrying the Current You want to measure.

Then determine the Wire-Gauge, or Copper-Area,
then look-up the Resistance per 1000-feet,
then estimate the number of feet between Fuses,
and then calculate how many Ohms that length of Wire should be,
then use Ohms-Law to figure-out how many milliVolts will equal 1-Amp.

Attach the 2-Wires going to the Fuses to a cheap Digital-Voltmeter,
you will have to use both the Inverting, and Non-Inverting, Inputs on the Meter,
Power the Voltmeter with a Battery.

After the initial calibration with a known-good Amp-Clamp Ammeter,
you will have a very accurate, Wired, Ammeter.
And even if it isn't 100% accurate, it will give very consistent readings
.
.
.
View attachment 241537
Thanks.
I'll test the circuit, and probable I can join it into my multimeter.
 

Thread Starter

LAOADAM

Joined Nov 21, 2018
862
The proximity and orientation of a Hall-Sensor to a Conductor is extremely CRITICAL to its operation.
But if the Wiring can be temporarily disconnected,
a Hall-Sensor with a Factory-Calibrated Gapped-Toroid-Core could be slipped over the Wire,
providing a very nice solution.

What will you bet, that they don't want to disconnect the wiring ????
.
.
.
Thanks.
How about the links:


Can it work?
 

Thread Starter

LAOADAM

Joined Nov 21, 2018
862
You can buy current shunts on Ebay at very decent prices
https://ebay.us/XVdEFI
The one I bought was perfectly accurate.
Industrial ones are about £15
It produces an output voltage of 60mV at the rated current.
Amplify that with an AD8418 or INA180 to get a usable voltage.
Shunt+shunt amplifier gives a better offset and lower noise than Hall-effect (so if you also need to measure small current in the same circuit it is the best method).

Hall effect
see https://www.lem.com/sites/default/files/products_datasheets/has_500-s_sp58.pdf
are galvanically isolated, and don't require you to terminate the high-current cable twice! (but the output signal is a bit noisy, and there is a higher DC offset than with the shunt. If you are not interested in high frequencies, you can filter out the noise.
If there is a time when you know that there is no current through the sensor, you can sample the output voltage and save it, then subtract it from the reading to null the offset.

@LowQCab 's method using a length of cable can be very useful, because you have probably already got the cable and you can just measure the voltage drop, but the voltage drop is proportional to absolute temperature. (And don't forget the fuse, because if the main cable comes adrift then all the current tries to go through your sensing circuit). You can use a device such as the AD8418 or INA180 to amplify the output.

For 300A you will probably be using 120 mm^2 cable, which will have a resistance of 141μΩ per metre. A metre would give a voltage drop of 42mV, amplify by x50 and you have a usable 2.1V. It will drift by 0.3%/°C. Don't forget you need to arrange the metre of cable to minimise its inductance, otherwise you will have misleading results on any superimposed AC.
Thanks.
The basic application is a power wheelchair, some of that used 100A breaker still got trigger. That's why I like to take a measurement of it.
 

Thread Starter

LAOADAM

Joined Nov 21, 2018
862
Assuming DC and a relatively short distance I too like the idea of using the existing cable. If >50V you should preferably do this on the low-side cable to avoid common mode and safety issues and use an instrumentation amplifier or current sense amplifier to give stable fixed gain between 10 and 1000 as needed. You may need some RC low-pass filtering to avoid RF pickup if your sense wire is more than a few CM long.

If temperature compensation is an issue add a couple of k-type thermocouples, one for ambient and the other strapped to the cable. The thermocouple outputs, between 1 & 5mV or so can be amplified by similar devices and fed to a couple of spare ADC channels or use a proper thermocouple compensator/amplifier setup (eg LT1025).
Thanks.
I didn't realize the RC low-pass filtering, I'll take it seriously.
 

Thread Starter

LAOADAM

Joined Nov 21, 2018
862
Bear in mind that the resistance of cable probably won't be what you'd calculate from the resistivity of copper, it will be the maximum value that it could possibly be and remain legal. The relevant standard here is BS6360, which specifies 120mm^2 cable at ≤153Ω/km @ 20°C (British "room temperature" is 5°C lower than American).

Probably the best suggestion here is to take the path of least inconvenience! There's no "right" answer.
If absolute accuracy is important, the go for the shunt. Trying to compensate the resistance of cable changing with temperature seems like a lot of bother.

Using a very short piece of cable and a lot of gain will maximise errors due to amplifier offset, which is important if you need to measure low currents, but not if you are only concerned with >100A.

If you use a longer piece of cable, DON'T be tempted to coil it up! The inductance will make you think every slight change in current is a massive spike!
Thanks.
It is not a absolute accuracy measuring. Just for select a suitable breaker.
 
Top