Low Current Sensor IC

[ransiluj]

I need good current sensor IC. I'm going to measure current change of the PC when software is running. Do you have an idea about how much current should it take. I mean the current that is changing. Using a good accurate IC is good way to measure that change of the current ? IC like ACS 712 is good or bad?

or is there any current sensing IC for measure micro current change or mA change?

and you can tell me what is most accurate way to measure the current.
1.using hall sensor
2.using a current sensing resistor.
what is the most appropriate way?

 

[alec_t]

Do you have an idea about how much current should it take.

That will depend greatly on which PC you have. Do you want to include fan current, monitor current, disk drive current?........?
If it's taking an amp or so you will be hard pressed to sensibly measure uA/mA changes.

 

[MrAl]

Hi,


I've actually done this. I wanted to know at least approximately how much power different programs would cause the computer to draw. I found that programs that use the CPU cause the most power draw, and if you have a system with multiple cores and the cores are turned on and off as needed, as the cores come into play one by one more and more power is drawn. For example, with one chess program running (while the computer is 'thinking') the CPU might use 30 watts, but then if two chess programs are running 60 watts, etc. That happens up to the number of cores being used. So with 8 cores and 4 chess programs running the computer is using about 50 percent of the total power it might use, but with all 8 cores running it would use the max power of the CPU.
Chess programs make good test programs because they are constantly using the CPU when it is the computer's turn, and you can make settings in the program to keep it thinking for a long time so it constantly uses one of the cores. If you happen to get a more advanced chess program that can use multiple cores then it may use more cores at the same time just for that one program. Setting the program to make sure it is thinking on the human's turn too keeps it running in the background even if it is not the computer's turn.

For the measurements i used two different ideas:
1. 0.2 ohm 1 percent non inductive resistor, rated 50 watts. The power rating is to keep it very cool not because it needs that much power.
2. A standard "Kill-O-Watt" power monitor for the line power.

Both of these methods gave reasonable results, but i think the second was the simplest and also allowed a direct power reading.

Note not all motherboards allow the power down of individual cores and some do it a little differently too.

 

[kinarfi]

https://www.lem.com/

 

[ransiluj]

Hi,


I've actually done this. I wanted to know at least approximately how much power different programs would cause the computer to draw. I found that programs that use the CPU cause the most power draw, and if you have a system with multiple cores and the cores are turned on and off as needed, as the cores come into play one by one more and more power is drawn. For example, with one chess program running (while the computer is 'thinking') the CPU might use 30 watts, but then if two chess programs are running 60 watts, etc. That happens up to the number of cores being used. So with 8 cores and 4 chess programs running the computer is using about 50 percent of the total power it might use, but with all 8 cores running it would use the max power of the CPU.
Chess programs make good test programs because they are constantly using the CPU when it is the computer's turn, and you can make settings in the program to keep it thinking for a long time so it constantly uses one of the cores. If you happen to get a more advanced chess program that can use multiple cores then it may use more cores at the same time just for that one program. Setting the program to make sure it is thinking on the human's turn too keeps it running in the background even if it is not the computer's turn.

For the measurements i used two different ideas:
1. 0.2 ohm 1 percent non inductive resistor, rated 50 watts. The power rating is to keep it very cool not because it needs that much power.
2. A standard "Kill-O-Watt" power monitor for the line power.

Both of these methods gave reasonable results, but i think the second was the simplest and also allowed a direct power reading.

Note not all motherboards allow the power down of individual cores and some do it a little differently too.

from which area you measure the current ? after tha power supply of the pc or before the power supply of the pc? how much current it drawn when program is running ?

 

[ransiluj]

i like to know your method that use to measure power consumption of the program?? did you make any seprate device to measure that? and i like to know the watt that you telling is apprent power or a real power(active power)?

 

[MrAl]

Hello,

With the Killowatt device you can read off the wattage from the device, no problems there, reading just like a digital volt meter but it gives you the watts. Now if your computer draws 100 watts without running the program and then it draws 150 watts after running the program (and keeping it running) then you know it caused a 50 watt increase in power. This doesnt always happen though. Sometimes if you run one program it will draw 50 watts more but then another program will not cause another 50 watt draw because it did not use another core (with powered down cores that is). You have to experiment.

With the resistor you only get a rough idea of the power draw unless you use an oscilloscope to measure the voltage drop across the resistor. When you measure the voltages drop across the resistor the current through the resistor is:
I=V/R
where V is the voltage and R is the resistance in Ohms.
Then knowing the line voltage E, the power is approximated as:
P=I*E
so if you measure 0.1 volt with a resistance of 0.1 ohm that means I is 1 amp, and if the line voltage is 120 volts then the power is:
P=1*120=120 watts
The catch here is that the current may not be in phase with the voltage (it's AC) so we would have to know the phase difference to calculate the power more accurately, but we would need to use a scope to do this. The Killowatt device does this automatically though and provides a readout without having to use a resistor and do a calculation.

 

[Reloadron]

Like MrAl I also tried this little experiment and I also used the Kill A Watt to note changes in line power to the computer. Originally tyhe little Kill A Watt units only worked with 120 VAC 60 Hz mains power but I guess they now have Euro versions on the market. Since you do not show or mention a location I can't tell you much more on the Kill A Watt.

Another method I used was to monitor the CPU 12 volt power and placed a hacked adapter cable on the CPU power connector. Similar to what MrAl did I used a 10 amp current shunt. Knowing the cpu voltage and current I applied the math MrAl mentions. Using a 4 core CPU I got each core to 100% by running some software designed to run a CPU core at 100%. Been years since I did it so I forget what my numbers were for the CPU I was using. I also did a similar experiment measuring the case inlet and outlet temperatures to see how much additional heat was generated when running the cpu at 100%.

Ron

 

[Jaguarjoe]

I have a pair of LEM current sensors I will never use. I don't recall the specs or model #, they're about the size of an ice cube. Contact me off list, they're yours if you want them, just reimburse me for the shipping.

 

[CmdrRickHunter]

How accurate are you looking to measure? Measuring to the nearest 1/10th amp (12W) for long durations to see how much running SETI@Home costs you is very easy with off the shelf hardware like the killawat. If you were looking to do timing attacks on an encryption algorithm where you're measuring far smaller currents and far shorter durations (microseconds or less), then you need a completely different set up.

I got a similar device to the kilowatt that uses induction to measure power. That plus a custom power cord with the live wire pulled away from the others has been all I need, and it's been useful in other cases as well (like measuring amperages of my A/C unit inside the circuit breaker box).

 

[ransiluj]

thank you everyone for your help.

 

[ransiluj]

I have a pair of LEM current sensors I will never use. I don't recall the specs or model #, they're about the size of an ice cube. Contact me off list, they're yours if you want them, just reimburse me for the shipping.

can you please tell me the price of it ?