# Differential voltage measurement using Arduino

I have a electrolysis cell in which the anode is hanging using a Structure in a electrolyte. The cathode is at the bottom of the cell with electrolyte above it and anode being dipped into this electrolyte (not completely, a portion of the anode is dipped). The cell operates at 7 volts DC and 90000 amperes of current flowing through it. This is a typical electrolysis process of producing aluminium from alumina powder in a aluminium smelter. The cell is called a pot. Now the question is I need to measure a potential drop on the anode rod. We usually take a meter to measure the potential drop on the anode rod and typically the reading are in millivolts. As the readings are in millivolts, it just indicates the drop in rod voltage due to resistance of rod. Now can I measure this millivolts drop using Arduino provided the voltage at one point of anode is 5 volts with respect to ground and the voltage at the other point is 4.95 volts with respect to ground, however the difference between these voltages is 0.5 volts and this is voltage I need to measure using Arduino and some arrangements as add on to it. How can I do this? Can anyone help. If you want more details, please comment. I will also provide more details.

So far I have done is tried measuring 1.5 volts on a pencil cell using Arduino by connecting A0 analog pin on Arduino to positive directly to the positive of 1.5 volts battery and ground of Arduino to the negative of Arduino.

Can I do the same for this millivolts measurements? I understand that the ground of Arduino and the 2nd point of anode measurement will not be at same potential, so shouldn't be connected together. I am scared to burn my board, so didn't try practically and wanted to get some opinion.

• 90000 amps?! Are you sure? That's just over half a megawatt of power! – Majenko May 27 at 12:54
• Lacking experience of high current applications, I'd never put an Arduino anywhere near something capable of pushing 90000 amps through it. I suspect it might be a problem to interrupt the arc current through the carbonised remains of the Arduino and nearby objects (/buildings). Maybe better qualified people can be bolder though. – RedGrittyBrick May 27 at 13:03
• Yes. An aluminium smelter typically consumes 1400MW of power for 0.5MTPA capacity. Current wouldn't flow from Arduino as the anode is a conducting material made up of aluminium solid rods. The current flowing through Arduino would be really small i think. Because we can measure the voltage on the cell with a multimeter as well. So high amps shouldn't be a problem as far as I think. – surendra choudhary May 27 at 13:15
• @Majenko Smelting! Big power. – Dave Newton May 27 at 13:16
• High current will always pass from the least resistance path and as the rod itself is highly conductive in nature, the dc current shouldn't pass through Arduino, however we will have to have some safety as well. I can't figure out a method to measure this potential difference. Or maybe I will have to open the existing meter and connect the measurement pin to the Arduino. – surendra choudhary May 27 at 13:18

This is obviously a significant project, not something to economize on. I would recommend you use a separate difference amplifier to get a single-ended output referenced to the Arduino ground and scaled to maximize the 10-bit range use. The diff amp can use its own supplies (+/-7.5 or +/-15, say) with the ground tied to the Arduino ground. That way the Arduino only sees the difference, never the full 7 V no matter what.

As you do it with the multimeter: connecting ground to the lower potential (4.95 Volts in your example) and the volt probe to the higher potential (in your case 5.0 V), you can also connect the Arduino between the two potentials: Arduino GND to the lower (!) potential and A0, A1, ... or A7 to the higher (!) potential.

As long as you use the correct polarisation and the difference between the potentials is less than VCC of the ADC (5V Arduino ADCs) you should be save.

The resolution of the Arduino might be a problem. I don't know what resolution you need; 5V divided by 1024 is about 5 mV per step.

You might want to provide a reference Voltage at Vin of -let's say- 1 Volt if you need a better resolution. That would improve the resolution to about 1 mV per step. But then the meassured voltage must no exceed the 1 Volt reference voltage.

EDIT (Gerben's comment)

You could use the 1.1V internal reference of the ATMega328P (analogReference(INTERNAL))

And please let me give you a security advice: If the resistence in any part of your meassuring setup is significantly lower than the resistance between the messuring points (e.g you shorten the probe wires), then there is a chance that 90000 A running through this setup. That could burn away your hand, arm or even your life. Amperes kill; not Voltage. I've seen a wrench that was vaporized (ok, melted, ok glowing ;-) ) in a second, when it had fallen between the poles of a car battery (12 V; 300 Amps). An you operate with 90000 Amps. !

• You could use the 1.1V internal reference of the ATMega328P (`analogReference(INTERNAL)`) – Gerben May 27 at 15:30
• @Gerben Yes that should also be mentioned. But I think even the 5 mV resolution is good enough in the 500 mV range. I just mentioned that there could be a problem with the resolution. Oh, and you brought me to another potential problem. If the measured Voltage can be higher than 1 Volt the idea of using an alternative reference voltage may be not the best. Thank you! – Peter Paul Kiefer May 27 at 15:41
• Note that the Mega2560 has an ADC that can do differential measurements, with a programmable gain of 1×, 10× or 200×. This should solve the resolution problem. – Edgar Bonet May 27 at 20:12
• Thanks all of you. I will read more into it. Thanks peter for clarification. This cleared a lot of doubts. Further there's no need to worry as entire potline in smelters is kept highly insulated and besides it's all DC, not AC. We can touch the DC terminal of any end as long as there's no potential difference and we are good to go with proper insulation. I have seen people directly working on these structures. Only thing is that, it gets really really hot as the electrolysis happens at 960-980 degrees, otherwise molten metal will not be produced. – surendra choudhary May 28 at 1:30
• I now have another question, we use a manual meter with an lcd. Can we read the values from that meter's lcd pin using digital pin read. I know it has to be pwm and the frequency may not be the same, however can anyone explain that part as well. @peter. – surendra choudhary May 28 at 1:33