Your favorite current sensor?

[madlabs]

Hi All,

I need to find some new current sensors. I have been using some Allegro parts, but I am having issues with paralleling them. I would like to go with inductive types, but would consider Hall types.

What I need is:

0-250 amps, output 0-5V, unidirectional. I'd also need a 100 amp model. Of course, I may not be able to find my ideal and am willing to consider anything.

Thanks!

Jonathan

[Rover]

Hi Jonathan,

I use Halls, Google Amploc. I've had good successs with them in my application running 4 amp100's  in parallel off power supply. May not match your specs though, bidirectional and offset is Vs/2 . So basically if using 5V in a unidirectional application output is 2.5 to 5V


Rover

[madlabs]

Rover,

Thanks for the reccomend. I was pretty sure I wasn't going to get a unidirectional anyway. The ones I am using now are bi-directional with the same Vs/2, I just would just have liked the greater resolution of a uni's. With a 12 bit adc that works out to 6.9 ADC count per amp, which is fine, but I have to admit my DIY controller is a little noisy and there is a few count bounce in my ADC board.

I had looked at those sensors and it is nice to hear that they work well. Do you have any experience with the Honeywell sensors? They are certainly more expensive.

Jonathan

[Rover]

I never triied the Honeywells ..... sorry .. did try various things with shunts... ended up back with the Halls.

When I moved my electronics to metal boxes I did notice a uptick in noise, when I had em isolated in separate box was not getting much noise... now I have em with my mppt controllers ...  it isv a little more  bouncy .. not real bouncy .. but enough I can see it (I also think the new BZ  mppt is noisy bu can't prove it as yet) . The metal boxes & BZ happened at the same time.

..just thoughts

 

[madlabs]

Rover,

Pulled the trigger and bought $120 bucks worth of the amploc sensors. Now I'm itching for them to arrive.

Do you have any experience with putting multiple wires (in my case different PV panel feeds) through one sensor? I'm wondering if that will be a problem.

Thanks again for the help!

Jonathan

[Rover]

Nope never ran multiple wires from different sources through one sensor. But I have run 2 current sources through one... basically same thing and I only did it for testing something when the other source wasn't producing anything so I knew what was producing.

My sensors all have the same type of wire dbl looped through the sensor. (Bascially makes a 100A sensor a 50 A sensor with higher resolution.) I have 2 barrier strips (1 before and 1 after the sensors). So if I wanted to run sources through a single sensor I just added it to one the in use barrier strip posts .

Here is a description and pic from previous post http://fieldlines.com/board/index.php/topic,144750.msg979190.html#msg979190

Rover

[rossw]

Do you have any experience with putting multiple wires (in my case different PV panel feeds) through one sensor? I'm wondering if that will be a problem.

Works fine. Couple of things to be aware of though - these are not "problems" and can be used to advantage.
1. Watch the polarity. Magnetic "summing" occurs. One wire with 5A one way, and one with 7A the other way will show   *2A*. Reverse either and you see 12A
This can be great if you want "nett current" for example.

2. *SOME* of the devices I've used over years don't have perfectly even sensitivity across the entire measuring aperture. Easily checked with a moderate current in a small conductor - move it around in the sensor and see if the output remains constant. If it does, great. If it doesn't, you can mitigate the problem to some extent by either twisting your current carrying cables together where they pass through the sensor, or at least use a ziptie either size to keep them bundled closely together.

[madlabs]

Ooh, wish I had thought of double looping the wires through, I would have ordered different sensor values. Oh well, it'll work and I can get some more if I really want the resolution. I'm going to do a little more work towards reducing the bounce on my ADC board and I should get decent single amp resolution. Or maybe I'll finally play with a 16 bit ADC.

Ross, great information. I'll do some testing as you suggest and secure the wires for consistant readings.

Jonathan

[rickysmartz]

Hi,
I use Magnelabs CTs and never had a problem. They have a good range and are pretty cheap here in Europe. They work like a clamp meter and easily clip onto your output line(s).
Richard

[madlabs]

Richard,

A quick look at the website revealed only AC sensors. Can these be used for DC as well?


Jonathan

[joestue]

no, CT's are ac only, and dc currents can cause them to inaccurately read low.

However, they do make CT's for dc and they are priced accordingly. they use three coils and some active circuitry.

[kevbo]

Well, I assume you mean an analog output sensor, but it reminded me of this:

In the category of "cute trick" I once saw a reed switch used as a current sensor in a high current battery charger for an electric car.  It was used to detect when to switch the charger from bulk to float mode.  The reed switch was placed at right angles to the conductor and the spacing adjusted to obtain the desired trip current. It got around all common mode and isolation issues, which were significant as the battery voltage was fairly high.  Of course you need fairly high current to detect the field of a straight conductor this way, and the reed switch has some hysteresis, which was desirable in the charger application.

[rossw]

Well, I assume you mean an analog output sensor, but it reminded me of this:

In the category of "cute trick" I once saw a reed switch used as a current sensor in a high current battery charger for an electric car.  It was used to detect when to switch the charger from bulk to float mode.  The reed switch was placed at right angles to the conductor and the spacing adjusted to obtain the desired trip current. It got around all common mode and isolation issues, which were significant as the battery voltage was fairly high.  Of course you need fairly high current to detect the field of a straight conductor this way, and the reed switch has some hysteresis, which was desirable in the charger application.

I once saw a perculiar "lamp failed" detector that worked sort-of similarly. The wire to the lamp was wound in a few turns (don't remember exactly, perhaps 4 turns or so), with a normally-closed (unusual in itself) reed switch in the middle.

When power was applied to the lamp, the inrush current in conjunction with the coil made enough field to pull the reed switch open, and the running current was enough to hold it. The reed switch was also connected to the same power source, and had a wire back to an indicator light for the operator - so if the light was supposed to be on, but either didn't come on, or blew, the operator had a "LAMP FAIL" light come on. This was old stuff when I saw it 30+ years ago.

[ghurd]

I once saw a perculiar "lamp failed" detector ...

Wow.
Amazing what people thought of back before all these fancy ICs.
G-

[madlabs]

Just a quick update. The sensors arrived and I have gotten around to installing a 100 amp sensor on the solar in and a 300 amp on the load side. The 100 amp one is reading very smooth. It does not, however, put out exactly 2.5V at 0 amps, so I had to fudge the numbers by a count of 17 to make it work out. Other than that the math is just as the data sheet suggests, so over all I am very happy. The 300 amp on the load is having some issues, but I think it's because it is too close to all of the power in line (solar, gen, wind). So I'll reconfigure the layout and see if it plays nice after that.

On to putting a 200 amp one on the generator! Thanks for the reccomend Rover, I'm very pleased.

Jonathan

[Rover]

Glad to hear it Jonathan.

Yep I calibrated/fudged mine as well. It also depends on how good your 5 volt rail is, as this is V/2 if that rail is off so is the output from the halls. Other considerations are the ADC and noise on the 5V rail, % error on your measuring eq, yada.

I use a switched power supply my monitoring system and halls use the same supply. I also have an el cheapo 12V automotive noise filter found in car stores as a prefilter before the micro's power supply.

Mine are back up and runing after hurricane Irene when my shed, where every thing is, was inundated with water and finaly crushed by a tree. The same halls are back up and working just as they were before.

[madlabs]

Rover,

I just spent some time trying to get the 300 amp load sensor to work. Even after physically seperating the load cable from the charge cable, I am getting some funky results. I substituted a 200 amp model and tried a different ADC input, and the result is the same, so I have to guess that it is in my setup, not the fault of the sensor.

What is happening is that when I turn on a load, say 40 amps, the readings will climb slowly above 40 amps, to say 50-55, then setlle down and wind up lower than 40, say 35. I do have a cap and a resistor input buffer on the ADC, like this

ADCInput  <-----/\/\/\/\--------||----- gnd
                                    |______ sensor output

The cap is a .1uF ceramic and the resistor is 330 ohms. The intention is to smooth out the output of the sensor. Do you think this cap could be accounting for the weird rise/settle of the load sensor?  The solar charge sensor has the same buffer and it is working smooth and accurate.

Jonathan

[TomW]

I just bet that the DC in on the inverter shows some spikes and relatively large variation of the current in it? This might cause peculiar readings.

Just thinking out loud.

Tom

[boB]

Rover,

I just spent some time trying to get the 300 amp load sensor to work. Even after physically seperating the load cable from the charge cable, I am getting some funky results. I substituted a 200 amp model and tried a different ADC input, and the result is the same, so I have to guess that it is in my setup, not the fault of the sensor.

What is happening is that when I turn on a load, say 40 amps, the readings will climb slowly above 40 amps, to say 50-55, then setlle down and wind up lower than 40, say 35. I do have a cap and a resistor input buffer on the ADC, like this

ADCInput  <-----/\/\/\/\--------||----- gnd
                                    |______ sensor output

The cap is a .1uF ceramic and the resistor is 330 ohms. The intention is to smooth out the output of the sensor. Do you think this cap could be accounting for the weird rise/settle of the load sensor?  The solar charge sensor has the same buffer and it is working smooth and accurate.

Jonathan

330 Ohms and 0.1 uF will start rolling off   (-3dV) at about 5 kHz.  Are you sure you want that low-pass that high of frequency ?

Maybe you might want a slower average and also get rid of any 120 Hz  (or 100 Hz)  ripple ?

I'd probably bring that response down to more like a few Hz rather than so high, depending on how fast you want it to
respond.

boB

PS...  I just noticed that you have the resistor in the wrong spot.  I think what you meant to show was to  have the sensor output and ADC input labels swapped from how they are shown here.
Probably don't want to load the sensor output with a capacitor directly AND would want to place the cap directly to ground from the ADC input which swapping the labels would fix.

boB (again)

[Rover]

I don't know what ADC you are using or the input voltage levels it will take... but I don't stick anything between my output from the HAll and my ADC (MAX186) .

Rover

[madlabs]

Bob, Rover,

The input circuit is as I drew it, it was reccomended to me a long time ago to protect the ADC input and smooth out the readings. I'm going to try removing it and see what happens. The funny thing is that it works well on the solar input, smooth and accurate. Obviously there is more fluctuation on the load line or something. I'll take out the cap and resistor and see what happens.

Rover, the adc is a MCP3208.

Thanks for the help!

Jonathan

[OperaHouse]

Do a running average of 64 or 128 samples.  I've monitored the pressure of pulsed pumps with rock solid readings.  Averaging out a little noise will almost give you another bit of accuracy.  Any ADC needs a cap on the input to lower source impedence.