10A+ DC current sensing circuit ideas?

[pigman]

Hey guys, I'm building a little automated pump project and I was wanting to install a current sense circuit in line with the 12V DC pump so I can output it's current consumption to the IC using an ADC pin.

My issue is I am only really experienced in sensing lower currents using a low value resistor with an op amp. And a resistor with a current rating of 10A-15A is a little bit rediculous.

Any suggestions on what sort of circuit I could use?

Cheers

 

[ericgibbs]

hi,
There are a range of Shunt current measurement resistors , from 1 to 1000's Amps.

**broken link removed**

 

[jpanhalt]

Another option is an integrated Hall effect sensor, such as these from Allegro:
**broken link removed**

Although, for just 10 A, the load resistor is a simple and most likely cheaper solution.

John

 

[Reloadron]

As can be seen you have a few options. The current shunts as Eric points out being the old tried and true. Each method has it's good and not so good points. Using a current shunt will generally require some signal conditioning off the shunt as you generally get for example a signal of 0 to 10 amps or 0 to 20 amps = o to 50 mV and we need to really get that 50 mV up to something useable. When doing circuits like this I like using a good Instrumentation Amplifier along these lines. Only because they are easy to work with and achieving a fixed gain is easy. Any IA will work but you will need signal conditioning and amplification off the shunt.

Then as John suggest there are integrated HALL Effect sensors that take care of some signal conditioning for you with a voltage out proportional to the current. These can be a very good way to go and offer isolation from the power driving the motor in your case.

There is also a canned complete Hall Effect solution in devices like these. Easily installed and with a nice scaled output of 0 to 5 volts if that is what you want for their range.

All a matter of budget combined with the accuracy you may need.

Ron

 

[Boncuk]

Hi pigman,

I guess a shunt is the worst way to measure high currents with DC. I consider it being an "emergency solution" since it requires a high power shunt resistor, converting current into heat, thus reducing the effectiveness of the load.

Here is a better way to go.

Look into Allegromicro and find the ACS713-ELECTR-20-T current sensor.

It is a unipolar current sensor (DC only) and has a full span of 0 to 20A with an output voltage of 185mV/A.

So at 20A load current it will put out 3.7V, good enough to be fed into any A/D converter to calculate for current.

It's internal resistance (for V+in and V+out) is 1.2 milliOhm and the voltage drop across it will be negligable (24mV at 20A).

If a full range of 20A won't suffice please refer to the ACS713-ELECTR-30-T which has a range of 0-30A with reduced output voltage of 133mV/A, resulting in max 3.99V at full range.

Both chips are available in an SO-8 package with a minimum of external components.

A typical application circuit is added in the attachment.

Boncuk

 

[ericgibbs]

hi,
A Rshunt of 0.01R at 10Amp will have a volt drop of 0.1V, amplify this by 50 and it will give 5Vout for 10Amps.

The Power loss at the Rshunt will be 0.1V * 10A= 1 Watt.

 

[MrAl]

Hi there,


In addition to the nice modern Hall Effect current sensors out there (we didnt have them in my day working in the industry, we had to make our own or buy elaborate DC current transformers or else design AC current transformers when possible) and the commercial current shunts, you can make your own current shunt using ordinary hobby grade brass strips. If i rem right, the brass has resistivity very close to 4 times that of copper, so you can use it the same way you would use copper but expect 4 times the voltage drop per unit length.
I've made my own brass shunts and they work pretty nicely. You have to calibrate them yourself but that's usually not too hard. One relatively short brass strip makes a nice 20 amp current shunt.
If you are interested i'll look up the spec's on one of the ones i had made in the past.
Total cost, about a dollar or so.

 

[4pyros]

Hi there,


In addition to the nice modern Hall Effect current sensors out there (we didnt have them in my day working in the industry, we had to make our own or buy elaborate DC current transformers or else design AC current transformers when possible) and the commercial current shunts, you can make your own current shunt using ordinary hobby grade brass strips. If i rem right, the brass has resistivity very close to 4 times that of copper, so you can use it the same way you would use copper but expect 4 times the voltage drop per unit length.
I've made my own brass shunts and they work pretty nicely. You have to calibrate them yourself but that's usually not too hard. One relatively short brass strip makes a nice 20 amp current shunt.
If you are interested i'll look up the spec's on one of the ones i had made in the past.
Total cost, about a dollar or so.

brass nice idea.

 

[nsaspook]

Most good current shunts are made from a special alloy called MANGANIN® that has a very low Temperature Coefficient of Resistivity. Brass is about 150 to 200 times worse and only twice as good as platinum which is used in RTDs to measure temperature by resistance change.

**broken link removed**

 

[bountyhunter]

Hey guys, I'm building a little automated pump project and I was wanting to install a current sense circuit in line with the 12V DC pump so I can output it's current consumption to the IC using an ADC pin.

My issue is I am only really experienced in sensing lower currents using a low value resistor with an op amp. And a resistor with a current rating of 10A-15A is a little bit rediculous.

Actually, it's very easy. You use a type of wire called manganan (or constantan) which has zero tempco and very accurate resistivity. I have made dozens and dozens of current sense resistors using it. It comes in many different diamaters.

https://en.wikipedia.org/wiki/Constantan

 

[Boncuk]

hi,
A Rshunt of 0.01R at 10Amp will have a volt drop of 0.1V, amplify this by 50 and it will give 5Vout for 10Amps.

The Power loss at the Rshunt will be 0.1V * 10A= 1 Watt.

Hi Eric,

if you compare prices you'd probably tend to buy a hall current sensor.

The shunt counts $23.95 per piece while the ASC713 counts $3.82 (Digikey) per piece.

Additionally the ASC713 puts out a voltage which can be directly connected to an A/D converter while a shunt requires signal conditioning. (amplification and filtering)

Moreover the shunt measures (LXWXH 2X1.25X1.68" = huge!) while the ACS713 comes in 8-SOIC package.

Regards

Hans

 

[ericgibbs]

hi Hans,
A 'low value' reasonable shunt can be home made from a small block of aluminium, by cutting slots using a hacksaw and trimming with a file.
Drill a couple of fixing holes for the cable termination connections, use eyelets on the cable ends and screw down onto the shunt.

I have never bought a shunt resistor, home made shunts work OK.
The important detail is try not to make it too small, to ensure that it stays cool at the max current
Eric

 

[MrAl]

Most good current shunts are made from a special alloy called MANGANIN® that has a very low Temperature Coefficient of Resistivity. Brass is about 150 to 200 times worse and only twice as good as platinum which is used in RTDs to measure temperature by resistance change.

**broken link removed**

Hi,

yes good point, you do have to know how to deal with the temperature coefficient if you expect a big change in temperature from the temperature at the time of calibration. I've had no problem but yes it's a good idea to be aware of this. interestingly, copper is even worse.

 

[nsaspook]

The calibration temperature difference offset can be easily accounted for, it's the drift from self heating that really causes a problem. If the shunt is in a control circuit the feedback from resistance change due to current change could reduce the measurement accuracy by a huge amount.

Copper shunt curve from NIST shunt measurement document.
View attachment 60781

 

[Mosaic]

I hope this isn't off topic, but can anyone suggest an approach to make a circuit tracer ? I want to be able to trace currents of perhaps 10A or more by injecting a known PWM signal into wire harnesses.
I think using a decent hall sensor can act as a pickup for following the wire in a harness. Is that a reasonable approach?