Hello, I was thinking about creating a high current arudino multimeter however there are some things I haven't quite been able to figure out or decide on yet and was hoping I'd be able to get some suggestions/ideas from those whom may have done something similar in the past.
I am looking to right now make a prototype that measures about 2-8 contact points in both series and parallel in order to allow me to measure the current and voltage used per contact point, then calculate and display this and the power consumed.
The device(s) I will be measuring have a maximum theoretical value of about 15Amps (16AWG wire) on either 12V, 5V or 3V power rails.
So far I've decided on potentially using the ACS756 Current Sense chip to perform the current measuring, however I'm not sure if there is a cheaper/better chip to do this. Also, I was wondering if i should just use a shunt resistor at these currents for my voltage meter or if there would be a better chip to use instead.
Hmm perhaps I worded it slightly wrong to try to stay slightly vague of what I'm trying to do. When I was referring to contact points, I meant separate power rails. I want to measure active current and voltage being delivered by a computer power supply to different computer components such as the hard drive, case fan, optical drive, etc. The contact points i meant would be measuring the 12V rail from the power supply to the hard drive, the second contact point would be the 5V rail from the power supply to the hard drive. I am not looking to measure multiple contact points with the exact same chip/meter, that would be calculated within the software portion of the design (which at this moment I'm not worrying about).
The reason for obscuring is I suppose hoping no one takes or uses my idea, but that's not here nor there for help.
I know that the 12V, 5V and 3.3V rails should all provide the same amount of voltage, hence requiring them only to be measured/monitored once. Its moreso the current monitoring/measuring that I guess I'm worried about as in order to measure power (wattage) you need to know voltage and current or voltage and resistance.
For the actual device, I'm hoping to actually have it as a removable device, as its something i want to try to proof of concept, then try to refine in to an actual potentially usable product by computer reviewers to allow their readers to know how much power a component component actually uses in their different states compared to just what the wall meter says their power consumption is.
If that doesn't answer your question, I'm looking to see if i can get it working for say one connector (molex as its simple) then with that success move onto the pci-e power connector, the cpu power connector to the motherboard and then the motherboard itself.
That my best course then, i was thinking of the ACS756 which will handle 50-100Amps (even though i need peak 15Amps) since it will sense the current and convert it into a voltage, in which i can then set calibrate my system with a multi-meter to verify that its linear or non-linear (current draw).
I'm not sure how the clothes pins would work still computer power cables as they are insulated, as my idea was to make a little header/connector that would just plug in between the power supply pins and the device it was to plug into.
I don't know if the wiring on those clothes pins, or clothes pins in general would be high enough to handle the current draw i'm talking about, unless I have absolutely no idea what you're talking about...
Those sensors need to be in-circuit, so it would be hard to make the whole arrangement removable, no? I suppose you might rig up some sort of removable connector/adapter. As you say, those ACS756 devices are pretty expensive, $6 each through DigiKey, so if you want to make 8 measurements, that's $50 right there.
I note that the device output is 40 mV/A, so full scale (15A) is 600 mV, which I guess should be your atmega ADC reference, giving a resolution of about 15 ma.
If anyone has a cheaper idea to measure current--that doesn't disturb the circuit very much--I'd like to know about it too.
KE7GKP:
The ACS756 is just a hall-effect sensor with a large wire embedded in the plastic case. You can do virtually the same thing by placing a hall-effect sensor against the insulation of the current-carrying wire. Current flowing through the wire creates a magnetic field proportional to the current. That ACS756 simply senses the magnetic field. But you can do the same thing "non-invasively" by putting a hall-effect sensor right against the wire insulation. The clothes-pin would hold the hall-effect sensor against the wire while you are measuring the current.
Look up "current probe" in Google and look at the pictures. There are lots of these large, expensive, precision probes available, but modern hall-effect semiconductors make it more feasible to create some miniature current probes that would be inexpensive enough even for permanent applications.
Hmm, I'm going to need to dust off my brain and my ElectroMagnetics Course content I took last fall, funny that this project would be a partial practical application of that. I would assume though the shield the wires are in are designed to prevent the magnetic field from going too far and disturbing the other wires, though if both were insulated they wouldn't disturb (aka add noise) to the other.
Well what looked like to be a simple W = V * I is going to turn into something a lot more mathematically complicated. I hope I can wade through it enough to understand the additional variables/terms that are in play and be able to accurate measure them (aka multi-meter as my control variable to see if my hall-effect sensor is working properly and if my math is correct).
I figured as much, though with something like this I'd rather sound the idiot and find out I'm right then blinding proceed and do something wrong down the line
I see http://www.allegromicro.com/en/Products/Part_Numbers/1301/index.asp may do exactly what I want, with this would i place this against the wire and it would measure the magnetic field, outputting a voltage proportional to the field it measured. This known voltage would then be converted using what is known, though that would just be the voltage through the wire to figure out the current that was going through it?
To measure a voltage for such high current devices, would i be best using an IC or a shunt resistor set up for this?
KE7GKP:
Yes, those small Hall-effect sensors are what I was thinking of. I think they are sensitive enough for this application, but I would have to test that theory before committing anything to the scheme. There are several different manufacturers of these and many different models with different specs. That Allegro part may be ideal (or not?) but it is by no means the only available examples.
Yes, I would assume that you would need to "calibrate" the current probes. But that is pretty easy to do with a standard reference current and Ohm's law.
Measuring voltage requires a direct connection. Hall effect ICs and shunt resistors are what we use for measuring current, NOT for voltage.
Yes I know, sorry I kind of went on a tangent. My question pertaining to the hall effect ICs is mainly the fact i have an output voltage, is this actually voltage or is it Voltage/Gauss. Either way, I'm trying to figure out how I'd convert from the output an IC such as that gives me to the current I'm trying to figure out. I know approximately V = I X B, but beyond that i'm a little stumped.
In terms of voltage meter, I know voltage is measured in parallel, not in series, but I'm still worried that the current on these wires could harm whatever I'm using to measure the voltage with if the current is too high. Could I get away with still using just a 1 or 10 Megaohm resistor that's a 1/4W or would i need it a higher wattage one.
KE7GKP:
You don't even have to know what Gauss is (or even be able to spell it). The hall-effect sensors output some voltage for a given magnetic field. It should be trivial to put a known 1A of current through a wire and clip on the Hall-effect clothes-pin and see how much voltage you get for 1A. I believe that the Hall-effect voltage output is linearly proportional to the mag field/current.
The previous paragraph discusses measuring CURRENT. The next paragraphs discus measuring VOLTAGE. The requirements for these two schemes are pretty much OPPOSITE each other. DO NOT confuse them. When measuring CURRENT, the LOWER the impedance/resistance of the measuring device, the less it disturbs the circuit. When measuring VOLTAGE, we want exactly the opposite. The HIGHER the impedance/resistance of the measuring device, the less it disturbs the circuit.
There is no significant current on the voltage measurement wires. Arduino has a very high input impedance and whatever load the Arduino presents to the voltage being measured will be completely insignificant. Of course, for measuring voltages > 5V, you would need to use a voltage-divider circuit of two resistors. And you would want to select those resistors to be a high enough resistance to not put any significant load on the node you are measuring. But that should be very easy to do.
You would need very high resistance values only if you were measuring very high voltages (100s of volts). You would never need high wattage resistors because that would imply that you are (improperly) drawing too much current from the measured node.
Alright, that clears some questions up. I know how to measure voltage and current, i've never actually gone and tried to do it from scratch (always used a multimeter or oscilloscope to do this).
I guess my last little question in this whole mess is how would one use the hall-effect sensor such as the one i linked. In the data sheet all i see mentioned is the input voltage and ground to drive the device, and the output voltage it delivers. From this I'm assuming you just place the hall-effect sensor near/against the wire (though i was assuming it would be a loop not just at a point), it should read me a voltage, then away to the races i would go with calibrations.
Yeah, i knew about the wiring, i was just moreso curious how it would be positioned. The only problem is I wont be able to make and loops of my current wires as they'll be active and most computer enthusiasts like having as neat of cabling as possible, without any looping cables.
I guess the best plan of attack now is to order some cables, pick up my test machine (old donated computer) and have at 'er
Yeah, that's something I'll have to worry about/see, although some power supplies have the wires slightly bundled together, so I could potentially try to add the some of two magentic fields, though I fear that could be another can of worms.
All I can say is I'm glad I have a multimeter rated up to 10Amps and my older test machine power supply can't even reach that maximum so ultimately i'll be able to measure whatever wire i'm using manually to see how the hall-effect device is operating.
They are essentially just a Hall-Effect sensor soldered onto a board and pre-calibrated at about 25$ a pop, which is absolutely useless for me if I'm going to need at least half a dozen of them.
That seems to be a good alternative, and seems to have a good price break at 10 of them, making it actually cheaper then the A1301's beefier brother the A1302 (which i just found out you can buy individually).
I think I may order 10 and have at it.
I'm not sure if I should ask it now, or later, but what parts/components would be needed to run the Uno processor, ATMega8U2 (or its predecessor which I currently own) so it can be programmed in the uno and then run independently on a small pcb.
What is with mouser and digikey charging 20-30$ for an order that'll be under a lb to ship
I think its a good bet is to use the one linked by cyclegadget the CSA-1, FHS-40-P or the one from Melexis. Place it in a small PCB then strap it to the wires. Altho you have to calibrate for each wire size thickness, for small wire it could be tricky.I made one for measuring 300Amps and it was pretty stable with the thick wire. See attached.
If you can cut the line and insert a sensor then the ACS-712 and ACS-714 are aok.
But your best bet is to use open loop hall effect sensors. They are a little bit expensive but is very easy to use. This particular one Ive used a lot and is around 17mv/amp/turn. Just unclip the wire you want to measure from the connector, coil around this sensor 3 times and clip it back in to the connector.
Well i got several variety/versions of some hall effect sensors (some courtesy of my university's labs) and it seems that on average there is not enough current for the amount of voltage that is going through any or a few of the wires to measure a difference.
I've decided I'm going to try to use some current sense ICs instead, I know this will be invasive however it'll give me guaranteed results.
I'm not sure if I'll be able to do it with an arduino board, as right now I've factored i can get away with using 12 of them (to measure the 24-pin ATX) active at one time, so the user can measure different devices separately. The goal/idea would be ATX power supply as one setting, the 2 pci-e power supply cables + the cpu power cable + peripherals as the second setting. For the microprocessor (or arduino that i do use) would measure certain pins of these to see what the voltage was, if it detected a certain line it would know what mode it should be in.
Analog I/O needed:
12 for current sensing
3-4 for voltage measuring ( 3.3V rail, 5V Rail, 12V Rail, and potentially measure the second 12V rail)
Any thoughts or inputs on this would be appreciated
