Monitoring Amps using Hall Effect Sensor

[geoffd]

I have a 3kw Outback Inverter (VFX3024) and I would like to monitor the input current.

As it has a 9kw surge rating this could be as high as 375 amps.  The cables have a cross section of 95mm sq.  Instead of cutting the cable, fitting terminals, finding a suitable 500 amp shunt etc.  would it be possible to monitor the magnetic filed of the cable using a sensor such as an AD22151 magnetic field sensor?

What is the panels opinion?

Cheers

Geoff

[Flux]

I don't know the AD22151 but Hall sensors are a perfectly acceptable and convenient way to measure current as long as you can provide the necessary power supply.

The AD part of the number makes me suspect it is Analog Devices and generally their devices are excellent.

I use Hall sensors to sense current in my buck and boost converters to track alternators to the prop characteristic. They are a lot more convenient to use than shunts when feeding back into circuits. For just metering there are not so many advantages but even so it may be an easier option and even cheaper.

Flux

[jimovonz]

I use the UGN3503U in conjunction with a ferrite core to measure amps without any trouble. You can vary the number of turns on the core to get the sensitivity you require or if you want to make a 'current clamp' type sensor where there is efectively only one turn, you can use an opamp to get the sensitivity you need.

[Ungrounded Lightning Rod]

While you can use a hall sensor - preferably in a clamp-on flux path - as a current sensor, there are other approaches.

For starters, why don't you use your cables as a shunt for a normal meter?

The cables have a cross section of 95mm sq.

Is that the copper cross-section, a measurement you took of the copper bundle, or a measurement of the wire and insulation?

What's the wire guage of the cables?  (If 95 mm sq. is the actual copper cross-section it's somewhere between 000 and 0000, so I don't know what to make of your numbers.)

Here's some wire numbers:

gauge   feet/ohm

0000     20402

 000     16180

  00     12831

   0     10175

   1      8069.5

   2      6399.4

You can find more at this page for starters.

Assuming it's 000, you could open the insulation at two points 1.618 feet apart and solder on two small wires, and tape it up.  That gets you a 1/10,000 ohm shunt (1 mv = 10A).  Use a digital meter on a milivolt scale multiply milivolts by 10 to get amps.  If the wire is long enough, do it 16.18 feet apart for a 1/1000 ohm shunt.  Then milivolts -> amps directly.

Similarly for 00 you'd use 1.2831 or 12.831 feet, and so on.

Solder your instrumentation tap wires so the ends are running ACROSS the big wire rather than along it for best results.  You can slide the wire back-and-forth to get the right position while the solder is being kept melted.  (Don't sweat the resisitvity of the solder.  It's high enough that the blob won't be a significant effect.)

[Opera House]

Two problems that can be encountered are non linear response of the sensor and saturation ofmagnetic material of the flux concentrator (clamp).  Commercial designs get around this by driving a coil on the concentrator to cancel the magnetic field with an amplifier.  The current to this field coil is proportional to the load current.  No manufacturer that I know uses just a hall effect sensor by itself.

[scottsAI]

The AD22151 may just work fine. With 375 amps the field may be large enough without coils and such. Give it a try. If the field is too large, just move away. The AD22151 requires power...

I prefer to use the shunt method like "Ungrounded Lightning Rod" suggest. With some differences. I do NOT like to cut the cables insulation. I use the ends along with a POT to adjust the signal to the DVM to get a nice scale factor. If that does not float your boat, then use a low offset opamp like ADI's OP193, input offset is 150uv, nice and low, draws 30ua of current. Works on 1.7v to 36v. Adjust the gain to give a nice scale factor to get a meaningful reading on the DVM.

Have fun.

Scott.

[tecker]

A lot of things can influence a Hall ic They can usually be depended on for a 1 or a zero though if you can isolate a portion of the current path that will not get too hot .There's a lot of different parameters that you can work with . They are for the most part not able to take much abuse . I use a relay coil to sense voltage rise in a back emf accumulator cap .Makes a nice pulse charger triggering a hexfet string. The coil take the energy release

better than other componants.

[geoffd]

I live in Europe so the cables are sold by cross sectional area, i.e.  area of the copper and I guess looking at the chart http://www.simetric.co.uk/siwire_elect.htm it is halfway between 000 and 0000.

I had not thought of using the cable itself as the shunt.  I could measure the exact size if the cable and use a wire connected at each end.  That would save cutting the cable, althought the maths might get a tad tricky.  

Thanks

Geoff

[geoffd]

It is an Analogue Devices chip, the only Hall Effect sensor thay weem to do.  The gauss calculations fall just within the spec of the chip.

Thanks

Geoff

[scottsAI]

Using the POT or (variable) resistor, you can adjust for the right cable length.

Lets say the cable is 5 ' long. The cable length you want is 2' for a nice 1mv=1amp.

Using the DVM measure the 100ohm POT. The value of the POT does not mater. Remember it has a resistance, you can only load it with a high 100x larger resistance to keep an accurate reading, load would be the DVM, usualy DVM's input resistance is 1M or 10M ohm, nice and high. Take the DVM resistance reading and multiply it by 2/5, adjust the pot to that reading. Verify the other side is 3/5 reading. Mark or fix it so the pot does not move. I'm sure your ratio will be different, so use the right numbers.

Measure your current readings across the 2/5 side of the POT.

Or, skip the POT, just remember to multiply the shunt readings by the ratio, 2/5 to get the correct reading! Might want a chart next to meter, to do conversions quickly.

Last note, copper had a large temperature coefficient. With a change in the cables temperature, the resistance will change, changing the current reading. Few deg doesn't matter, but 10 does, if your going for accuracy.

Have fun.

Scott.

[drdongle]

Can you be a bit more specific on how your current measuring method works, and perhaps a schematic?

thanks

[Ungrounded Lightning Rod]

I prefer to use the shunt method like "Ungrounded Lightning Rod" suggest. With some differences. I do NOT like to cut the cables insulation.  I use the ends along with a POT to adjust the signal to the DVM to get a nice scale factor.

That's OK by me if you remember to do one thing:  Don't crimp the pickup wire into the connector with the others.  The connector crimp resistance and various current-path pathologies can be large compared to the copper resistance.  Far worse, it varies with pressure, which varies with temperature and flexing.  I'd prefer to solder the mater taps to conductors a non-trivial distance inward from the connector, after the current has evened out.

Another approach would be, when you cut the wire for the crimp=on connectors, to leave the centermost conductor long and let it pass through the crimp-on connector, then solder your pickups to that, BEYOND the connectors, and insulate the result (so it won't touch the posts the wire is connected to, on the other side of the voltage drop from any resistance from the bolt-down.

A pot should work just dandy - even if the meter is NOT a large resistance compared to the pot.  You get a thevanin equivalent voltage and series resistance, and since the resistance is linear, calibrating the meter at one current level calibrates it at all currents.

Low meter resistance just means the pot's setting is harder to precompute due to the load effects.  But you don't want to try precomputing a pot setting anyhow, because you'll never hit it right by hand.  Instead you need to hang some other instrument on the wire temporarily and calibrate the ammeter to get the readings to match.

[scottsAI]

Great idea, leave center conductor long going through the crimp on connector.

I'm getting ready to build cables for my battery, I think I will give that a try! Thanks. I may use more than one wire, just in case it brakes.

Do you solder your crimp on connectors?

Have fun.

Scott.

[tecker]

  Well Doc I kinda was digressing a little into a relay I have been using . The hall ic triggers when the coil is energized high enough saturate an adjustable gap between hall face and the core of the coil. I use it to sense voltage rise. I don't think the large conductors will give a reliable field to trigger the ic effectively. I 'm guessing geoffd needs to protect his battery bank and a 400 amp breaker is not practical.

[tecker]

  The ad22151 I think is non polar at the face and has temp compesation .It's a 8 pin flat pack and there are resistor configurations for high and low magnetic fields .Alot of adjustment there but reliabiltiy wonld be dificult as  the flux path would be hard to align.

[geoffd]

I have a suitable breaker installed and the setup is working well.  

The problem I have is that I have no monitoring to see how much power I am using.  I don't have funds for anything like an EMeter / Link-10 and would rather not go thorough the hassle of disconnecting everything, cutting 95 sqmm cables, attaching crimp terminals and installing a 500 amp shunt.  

What I am hoping is that I can get enough information to see what is happening with loads etc.  by using a Hall Effect / Magnetic sensor.

Cheers

Geoff

[Ungrounded Lightning Rod]

I do solder my crimp-ons.

But I'm not sure if it's productive, neutral, or counter-productive.  (I think it's productive, since it gets you away from depending on the tension of the crimp, which varies with temperature and flex.)

I do NOT solder wirewrapped connections:  They have been shown to be better as-is than either a soldered joint or a soldered wirewrapped connection.

Note that I have not done cables this size (yet).

[Ungrounded Lightning Rod]

Have you looked at getting a clamp-on ammeter?

[Ungrounded Lightning Rod]

Don't try to preset the pot.  Instead, adjust it for the right scale reading.  (It's a linear device so if you calibrate your meter for one load level you've calibrated it for all.)

To calibrate it:

 - Attach a moderately heavy load with its own (reference) ammeter and a switch or clamp-on connector between the load end of your "cable shunt" and ground.

 - Hang a voltmeter across your inverter's input.

 - Switch the load on and off, recording the current your new meter reads in both on and off states and the current the reference ammeter reads.

 - (Also check that the voltmeter at the inverter's input isn't changing appreciably, which would make the inverter change the amount of current it draws.)

 - Compare the difference between your new meter's two readings to the current through the reference ammeter in the "on" state.  Adjust the pot and repeat until the difference and the reference current match.

[Ungrounded Lightning Rod]

Last note, copper had a large temperature coefficient. With a change in the cables temperature, the resistance will change, changing the current reading. Few deg doesn't matter, but 10 does, if your going for accuracy.

But note that an analog ammeter intended for use with a shunt also uses copper with the same temperature coefficient, which automatically compensates.  (The current is always divided in the same fraction between the shunt and the meter, as long as the shunt and the meter are at about the same temperature.)

Digital meters intended for use with a shunt typically have a temperature compensation circuit to perform the same function.

If you use a pot and your shunt is in a place where temperature changes widely, you may want to substituted a tapped piece of thinner copper wire for the pot.  Just hang your wire across the "cable shunt" output, then solder your pickup points to the wire and move the joints along that wire to achieve calibration just as you would have if you were doing it directly on the main cable.

[geoffd]

I had originally thought of a clamp on meter, but the cost of them here is not much less than buying an EMeter and I would not get remote monitoring and 24 hour information.  It would be an out of the house, downstairs and into the basement trip each time I wanted to check it.

Cheers

Geoff

[tecker]

  Geoffd

  You might get a hall ic to trigger on a meter pointer ,here's the digikey page if you want to give it a go. You could have the ic parallel in with The eol resistor in burg alarm point and program a key pad to chime in with a trouble. If you don't have a burg system ebay is always good for a couple of oldies .

http://dkc3.digikey.com/PDF/T053/1598.pdf

[RobD]

I use the Allegro ratiometric hall chips.

Here's a link:

http://www.allegromicro.com/techpub2/techpub2.htm#B

Check the 27702 notes.

RobD

[tecker]

Cool site

[geoffd]

Hi Rob,

thanks for that pointer, it is a very interesting document.  I need to digest it and work out if what I am doing is feasible.  Part of the problem is the dynaic range of what I want to measure.  Most of the time it will be perhaps 5 - 10 amp, then when the heavy loads are on it might be 100 - 12- amps , and when I switch in my arc welder 200 - 300 amps  Don't do that often!  But good info thanks.

Cheers

Geoff

 

[jimu]

I too use the Allegro ratiometric hall effect chips in measuring current in my solar and wind Homebrew A/D converter circuits..'

One of the nice benefits over a shunt is the isolation provided by the chip circuitry.

I feed the output of the Hall effect chip into a simple LM324 op amp ( for more isolation) and then into my A/D circuit.

Hope this helps..

JimU

Solar.no-ip.net

[RobD]

You can bias the coil to to increase the range the hall senses.

Allegro is great about sending out samples. I have an electronics business but I think they are even pretty good to hobbiests.

They also sell a current sensor for high current apps.I'm away so I don't have the number on hand but if you do a search on thier site i should come up.

Another chip you might want to check out is the Zetex ZXCT1009. This is a current sensor. I use it for low to medium current sensing in SMPS circuits but it should work in your app. Mouser sell them at low cost.