Integrate ampere measurement over time? 2

[SolarTrap]

Hello,
I am looking for an I2C device that can help with accurate measurement of amps (+/- 50A @30VDC) flowing in 2 directions (in and out of batteries). I am currently using an Arduino to do this but accuracy is low in terms of available bits at the ADC but especially with the sampling rate I am able to do the measurements (10Hz) with. Is there a chip that is able to integrate the amps over time until I read it?

Thanks
Markus

 

[IRstuff]

Doesn't the ATmega328 have a 10-bit ADC??

TTFN
I can do absolutely anything. I'm an expert!
faq731-376 forum1529

 

[SolarTrap]

Yes it is 10bit. In the range from -50 to +50 ampere with 1bit of noise you end up with a resolution around +/-0.195A.

 

[VEBill]

10 bit = 1024 = 0.1%

It'll be tricky to even find a current sensor much better than that. Certainly doable, but not a trivial step.

You should break this up into three parts.
1) Current measurement sensor (tricky)
2) ADC with as many bits as you like
3) uC to do the management and integration

If it's for solar power, you should obviously design for minimal power consumption. No point having a solar control system that needs hundreds of watts .

 

[SolarTrap]

I am currently using a INA286 sensor. My circuit works but I want to increase accuracy.
What do you mean by "uC to do the management and integration"?

 

[IRstuff]

Like an Arduino, but just to do the bookkeeping

TTFN
I can do absolutely anything. I'm an expert!
faq731-376 forum1529

 

[SolarTrap]

That brings me back to my original question: any links to a chip ideal for this task?

 

[VEBill]

The INA286 measures the voltage across a shunt. Presumably the shunt is a length of heavy wire, or similar block, likely copper. Internet says "The Temperature Coefficient of Copper (near room temperature) is +0.393 percent per degree C." (Resistance,to be clear) So if the temperature of the copper shunt drifts by 10°C (very possible, could be more), then you've got a 4% error in the assumed resistance, and thus 4% error in everything that follows.

If you're worried about 10-bit resolution, there are perhaps other issues to address. Unless you've already addressed them. Like using different material for the shunt, or measuring its temperature and compensating, or actively controlling its temperature.

Yes, the uC would interface to the ADC, and can do the integration of the current over your selected time intervals. It's almost certainly the simplest approach to integrating the current as you'd mentioned. In theory, you can integrate with analog hardware, but it's likely a lot more effort to design and debug.

 

[IRstuff]

So, why not

http://www.ti.com/product/INA226/datasheet/specifications#SBOS7432075

16-bit resolution and I2C out?


TTFN
I can do absolutely anything. I'm an expert!
faq731-376 forum1529

 

[SolarTrap]

The INA226 looks interesting but how can I do the integration over a time interval?
I goggled "uC" but did not find anything useful.

 

[itsmoked]

uC to do the management and integration

Use a microController to do the management and integration.
There are a truckloads of pre-made units that could do this for you. You'd need to do some programming. Some would have a lot of it done already, none would have it all done though.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Skogsgurra]

I can't understand you, SolarTrap. You say you are a computer guy and you say that you googled microprocessor and didn't find anything useful. It doesnt make sense. At least one of the three statements must be false.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[SolarTrap]

I think my only disconnect here is the meaning of "uC" -- what does it mean?
I am already using a micro controller to do all of it - my problem is I can only sample the ampere with a rate of 10 measurements per second and in a very dynamic load pattern I might miss a few peaks.
VE1BLL is talking about an analog method of integrating measurements - how is this done?

 

[IRstuff]

uC = micro-computer, or controller. This term has been in use for over 40 years.

The INA226 appears to have at least 10x that bandwidth

TTFN
I can do absolutely anything. I'm an expert!
faq731-376 forum1529

 

[VEBill]

[URL unfurl="true"]https://en.wikipedia.org/wiki/Op_amp_integrator[/url]

It's likely a lot more effort to design and debug allthe associated analog circuitry, than repeatedly sampling and integrating in a uC (microcontroller).

 

[IRstuff]

I think you also have to be more concerned about drift and dynamic range in an analog integrator.

TTFN
I can do absolutely anything. I'm an expert!
faq731-376 forum1529

 

[SolarTrap]

I think I just stick with my current solution to measure the amps - the analog way might be too much effort.
Thanks for your suggestions. I will start a new thread about a more important problem in a minute.

Markus
uC = micro-computer was interesting news

 

[itsmoked]

uC also often denotes MicroController which is generally considered a fully-supported-on-a-single-die processor. Like an Atmel or PIC.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[jimkirk]

Not I2C, but this might be of interest:

http://www.linear.com/product/LTC4150

Generates a level to tell the direction of the current and a pulse with the frequency proportional to magnitude, it does the integration for you.

Designed as a coulomb counter for battery management, has pretty good accuracy (0.3% of full scale error). Range set by a shunt resistor.

 

[SolarTrap]

@jimkirk: Thanks, that is an interesting IC. It seems to be designed for small lithium batteries only so would that work with SLA around 100AH 24V at all?