Current measurement - Dynamic range (looking for suggestions)

Thread Starter

Dyslexicbloke

Joined Sep 4, 2010
566
HI folks,
My PV and battery system is 12V, not a great choice I grant you but it just sort of evolved and changing it involve essentially new everything, well apart from a few PV's that match and could be reconfigured in series.
Basically it is what it is for now, 12V and occasionally silly amps.

That said the problem would still be the same if the voltage were higher and the current lower as in the potential range would be big.

What I need ...
Accurate measurement with a resolution of 1% or 0.1A whichever is larger, IE. If I am drawing/charging anything between 0.1A and 10A I want a resolution of 0.1A however above that 1% is good enough IE at 200A a resolution of 2A will be just fine. (Better would be nice but not necessary)

Brief surge currents of circa 1000A are possible, starting motors and the like, and as I don't want to drop more than 0.5V across a shunt at any time that would demand a shunt at 500uV / A which would make measuring 1A a little tricky to say the least.

What I have ...
The inverter is a cheap model and is supposed to be capable of 20kW surge but I have always assumed this was just silly and that 10 was probably pushing it. My biggest loads to date is a chop saw at 1600W and the microwave which is about 1800W, which it has absolutely no problem with.
I havent actually measured the surge current but the saw will have a starting load of at least 5000W which is likely to be circa 500A when you take the dip in terminal voltage into account. Having just done that calc I think I could get away with a shunt at 1mV/A but no more.

The batteries are 450Ah@C20 and hooked up with 50mm, 2 X 25mm in parallel, which is good for 200A, going by general specs, but in practice could be pushed well beyond that without suffering any thermal issues.

So...
Can I reliably measure down to 1uV or less with a relatively easily constructed amplifier that will stay linear?
I appreciate that I would have to have some sort of gain control to change the range and resolution as the resolution of the DAC would be an issue otherwise.

I also appreciate that a, probably some, hall effect devices would be good but I am amusing I would need several so \i could ignore the saturated ones as the current went up. The problem here is cost... I want to measure at at least 2 points, preferably 3 although the last on is fused at 100A and will never see more than about 50.

I had wondered about an active system ... Essentially using a hall sensor but adding biasing coils to the core.
The thought process was to have perhaps 200 turns on the same core as the battery cable passes through. I wouldn't worry about having a linear output from the hall sensor but would always maintain it at 0 and whatever current it took to do that would be 1/200th of the current in the target cable.

I would appreciate any and all thoughts or ideas about this subject in general as well as comments with respect to my musings.

Thanks for looking,
Al
 

Papabravo

Joined Feb 24, 2006
21,094
Outside of using a laboratory instrument, I don't see how measuring to a 1 μV will be possible, especially if this voltage represents a similar sized current. Think about the following:
If you have an A/D converter with a reference at say 2.048 Volts, then 1 μV would require 21 or 22 bits of resolution. Designing such circuitry is not for the faint of heart, and requires careful design, layout, and calibration. The test equipment required to verify such a design would put a large dent in most wallets. You might want to start with more modest and practical goals.

You might find this datasheet interesting:
http://www.ti.com/lit/ds/symlink/ads1226.pdf
 
Last edited:

nsaspook

Joined Aug 27, 2009
12,997
I have a similar system that I play with. Open loop Hall sensors were my choice mainly because of galvanic isolation for the controller and the simplicity of the volts to amps interface with a controllers onboard ADC. I have a 50A sensor for PV input and a 300A sensor for battery current from the neg terminal to ground so I can easily measure the balance of current from PV to load and to/from the battery bank.

PV 50A input current sensor


Battery 300A in/out current sensor


Two strings of GC-2 batteries for about 420Ah at 12vdc.


AMP200 sensor specs
http://www.ampsense.com/KEY100 AMP200 AMP300.pdf

At +5vdc you're looking at 9.5 mV/A from the null offset Vs/2 (2.5vdc) for a 1.9 volt delta from 0 to max +- 200 amps,. With a 5+ ref 2^12 ADC that's about 1.22 mV per count over that 1.9 volt/200A current delta @ ~0.129A per count. With a small amount of oversampling and signal processing it should be possible to get under 0.2A resolution with pretty good accuracy with a high precision 5vdc voltage ref chip for your ADC conversions and sensor supply. An off-board Delta Sigma ADC (MCP355X at low sample rates but that's OK for a battery current sensor) could easily get you into 2^16 range for better possible resolution but high current hall sensor errors might make those extra bits mainly noise.

http://www.eetimes.com/document.asp?doc_id=1276974
http://www.microchip.com/ParamChart...2&mid=11&lang=en&pageId=79&redirects=deltasig
 
Last edited:

nsaspook

Joined Aug 27, 2009
12,997
My system tracks voltage/current in 0.1 units so in this ~ 2 day chart of data saved every 30 seconds to the systems SD card.
Column:
AC PV current
AD Battery current from the neg terminal to power ground so negative currents are battery recharges.
AE Calculated load current.

X Controller supply voltage
Y PV voltage
Z Charge Controller voltage
AA String 1 voltage
AB String 2 voltage

C source code of the complete data format and variables.
 

Attachments

Thread Starter

Dyslexicbloke

Joined Sep 4, 2010
566
Guys that is great information thanks....
I dont know much about ADC, other than conceptually and the basics so having somewhere to start is great.

With respect to getting a handle on an actual working example the insight is invaluable, despite the fact that I haven't had chance to fully investigate yet I think you have given me enough to focus on for now and gone a good way to demonstrating what not to pursue.

I sound silly but I hadn't considered relative measurement, in practical terms, and I now realize, silly me, that placing several sensors in key positions on my system effectively removes the need to achieve one solution that is capable of both high current and high resolution.

It also occurs to me that sensors in parallel effective doubles the resolution because I could use 2 on board 10 bit ADC's but have each look at aproximatly half the current. I absolutely take the point RE external DAC and when I have tested the Amtel chip's performance I may well find that I need to go there but it is looking doable without extreme measures.

I have to be realistic ... I dont have time to play with this for month whilst I learn the finer points of building the precise circuitry that a differential amp would require and even if I did I expect trying to do that on strip-board would be somewhere between hard and impossible.

However building a relatively low gain amp to scale and offset a relativly large signal is a much more realistic goal.
In fact now that I think about it if I measure charge current, which cant exceed about 90A, inverter current which could be big but only needs modest resolution when it is and 12V aux bus, which is my 24 Hr loads, and is never going to exceed 50A it all looks much more realistic.

Am I correct in assuming that a hall sensor will not be damaged by over-current?
I appreciate that the reading will be useless but a 100A sensor, for resolution at low load and a 400A sensor to quantify high load, when the 100A is swamped could work well, assuming it will work at all.

Again thanks both, I look forward to your comments whilst I am doing some more parts research.

Cheers
Al
 

nsaspook

Joined Aug 27, 2009
12,997
It's unlikely the 100A sensor will be harmed by the overload but I don't know what the recover time will be for accurate measurements again. I would use a controller with a 12bit ADC if possible to reduce electrical measurement absolute errors so you can see battery aging trends easier of you collect long-term data like I do. Most of my parts came from the work junk bin so it's a total mix-mash kludge on top of kludge. One tip is to partition the system into at least two parts, a measurement front-end that collects raw electrical data and processes the data into some sort of standard easily human readable format to a communications port for later processing with a more powerful 32 bit system for easy web or remote access.


The application uses the realtime data collected from the battery monitor front-end.

 
Last edited:
Top