Current sensor

[rc-pilot-freak]

I am looking for a current sensor for my Onboard data analizer project. I have on-order some free samples from Allegro (0-50A), but now I realized that they have poor resolution (high level of noise).

I would like to measure current in range from 0 to 5A with at least 10mA resolution on a 10-bit A/D on my PIC16F819.

Which company's sensors do you recommend?

[Trevor_G]

I think you are pushing your luck with the Allegro Hall Sensors that I have looked at. I use a UGN3503 on my datalogger. This is the bare device i.e. no copper conductor to carry the current to be measured. I use a ferrite toriod with a slot cut in it for the UGN3503. Works well and gives me full scale ~+/- 50A like the devices you are talking about. Two problems poor linearity and hysteresis in the magnetic component causing zero to wander.

You might do better to look at the LEM range of sensors. They use a comparison current to null out the current being measured. The Hall Effect device just detects the null point. The result better accuracy and linearity. Downside bigger, heavier and more costly www.lem.com (http://www.lem.com/) .

Trevor

[thanhTran]

Check this thread out:

http://www.rcgroups.com/forums/showthread.php?t=490031

I use two sensors mentioned in there and they work quite well

Thanh

[rc-pilot-freak]

Thanks Thanh, i've seen you were experimenting with Allegro's sensors. Are they any good? Did you make the output stable enough and filter the noise? How good resolution did you get?

[mmormota]

What about just measuring voltage on a small resistor (several cm wire)? No drift, simple, accurate.

[thanhTran]

Thanks Thanh, i've seen you were experimenting with Allegro's sensors. Are they any good? Did you make the output stable enough and filter the noise? How good resolution did you get?

I put a low pass filter (a resistor and a cap) based on the suggestion from Allegro's app. note and use software to averate the reading - will slow down the data. I got 1A resolution (I only use 8 bits out of 10) pretty stable on a 50A range. It seems working okay :)

Hope that helps

Thanh

[rc-pilot-freak]

I put a low pass filter (a resistor and a cap) based on the suggestion from Allegro's app. note and use software to averate the reading - will slow down the data. I got 1A resolution (I only use 8 bits out of 10) pretty stable on a 50A range. It seems working okay :)

Hope that helps

Thanh

Isn't a resolution on 8 bit 0.2A resolution? Can you post a current sensor's filter schematics?

[rc-pilot-freak]

What about just measuring voltage on a small resistor (several cm wire)? No drift, simple, accurate.

I had that in mind, but i would like better to use a single chip sensor, rather then the wire and the op-amp and resistors and caps (space saving).

But if I couldn't get the right sensor, i would have to use that.

[Hell-e-Guy]

Hi,
Allegro has 15A hall effect current sensors in SO-8 package, so you will have 3 times the resolution then with the 50A part.
As for filtering , if you don't want to add analog components, you can use the PIC to implement a second or third order IIR digital LPF. Since you are measuring a DC line the PIC should be up for this task (such as a 10hz LPF).

Hell-e

[Trevor_G]

I spoke about accuracy which is precisely what I mean. You keep talking about resolution on a Hall transducer :eek: an analogue device. On an analogue device resolution has no real meaning. On a digital device resolution does have a meaning it is usually +/- 1 bit i.e. for 8 bit 1/256 for 10 bit it is 1/1024.

In terms of these measurements you have three parameters:

1) Resolution
2) Repeatability
3) Accuracy

They are not the same. It is important be clear what we are talking about and despite what has been said I am sure it is not resolution.

Another parameter you will need to consider is linearity. It is also worth check ing that your ADC is monotonic.

To be clear about what I was saying earlier you are asking for an accuracy of 10mA in 5A that is 0.2%. I don't believe the Allegro devices will come close. LEM quote 0.15% although I suspect this would be difficult to achieve in practise. You will also need to check your 10-bit ADC spec. the accuracy figure may give you a nasty surprise.

[AndyOne]

What about just measuring voltage on a small resistor (several cm wire)? No drift, simple, accurate.

Unless you use the right kind of wire there will be drift and non-linearity. The best type wire is Nichrome but some sorts of Nichrome are better than others. A better solution than that is to use a low temperature coefficient resistor, one from the Welwyn range of power resistors is a good choice. Make sure it's rated to at least 10 times the power it will normally dissipate and it will hardly get warm and therefore the drift will be minuscule.

Andy.

[AndyOne]

It is usually the rule with A to D converters that you throw the LSB away when making a rough determination of it's accuracy, therefore from a PIC's 10 bit converter, the best you can realistically rely on is 9 bits.
As for the filtering the easiest way of doing this would to be to take several samples over a predetermined time and average them together.

Andy.

[mmprestine]

The Allegro sensors work fine in my project. Everything is here take a look.

eFlight Motor Analyzer (http://www.rcgroups.com/forums/showthread.php?t=623055)

Matt

[Trevor_G]

Matt,
I have just looked at the spec sheet for the Allegro ACS750 and they quote a typical error of +/-2%. Although they give no maximum error reading between the lines it could be around +/-5%.

These are the sorts of figures I would expect with the device I used for an onboard datalogger. Personally I am fine with this level of accuracy. Indeed I am curious why more is needed rather than nice to have.

[ghoti]

If there is a fuse in the circuit you could measure the voltage drop across it.
Bill

[mem]

If there is a fuse in the circuit you could measure the voltage drop across it.
Billbear in mind that most fuses will have significant resistance variation with temperature.

BTW, Bill is it ghoti as in fish?

[ghoti]

bear in mind that most fuses will have significant resistance variation with temperature.

BTW, Bill is it ghoti as in fish?

Yes, my last name is Sturgeon

[JohnMuchow]

Matt,
I have just looked at the spec sheet for the Allegro ACS750 and they quote a typical error of +/-2%. Although they give no maximum error reading between the lines it could be around +/-5%.My only concern (when considering using it for a project) was that the +/-2% figure was only for a temp. of 25-degrees C. Over the full temp range of the sensor, the Total Accuracy is +/-13% (-20 degrees-C to 85-degrees C).

Of course, any data logger will never see this temperature range but there's no data showing the Total Accuracy numbers vs. temperature (just offsets and zero-amp accuracy) to help gauge the actual accuracy for "typical" temperatures the sensor might experience.

Anyone have some numbers for, let's say, 0-60 degrees-C?