Shunt Electronics Help Wanted

[nicknorman]

How do I protect it from spikes >+- 32V? My knowledge of basic electronics is poor.

TransZorb.

[dmr]

TransZorb.

 

That's your shortest reply ever???

 

So, lets say the op uses a 10 ohm resistor between the shunt and the "219", with a worse case voltage spike which transorb should he use to limit voltage to 26 volts without self destructing?

(I have an interest myself here as its still on the todo list).

 

.........Dave

[dmr]

This subject has come up before so if you search the forum you might find it. I would dis-regard the shunt method altogether and get a hall effect current sensor ( see link below for examples from Farnell ). I doubt they are more expensive than the shunt ( with extra electronics required ), are more accurate, more linear, provide electrical isolation, and give a simple analogue voltage output proportional to the current. ( probably 0 - 5v or similar which could be reduced with a simple divider ).

 

http://uk.farnell.com/webapp/wcs/stores/servlet/Search?catalogId=15001&langId=44&storeId=10151&categoryId=700000005923&st=current%20tansducer&beginIndex=1&showResults=true&aa=true&pf=110151160

 

I did spend a little time looking at the specs of these things and I concluded that a shunt was still the better way to go. Can't remember the details but I think it was the zero offset that let the Hall effect devices down, though linearity was a bit iffy too. The active ones have the better performance but there is a current consumption penalty.

Its a Tall order, need to cope with the heavy charge current (150 amps?) but also accurately handle a long slow discharge. A 1% zero offset is an error of 1.5 amps.

 

.................Dave

[sjc]

Fair enough Dave, each to their own. The shunt method is more common and what people would think of first and foremost. But any solution will involve a compromise, whether people realize or not. I haven't re-checked the specs recently but seem to remember linearity <1%, and the electrical isolation appeals to me.

Also, with a shunt ( say 200A == 50mV ) you will need to use an op-amp that will run off a single supply ( 0 -12v ) AND operate right up to its' voltage rail ( not all, especially the cheaper ones do this ). Otherwise a dual rail device ( +/- 12v ) could be used. As for accuracy, to detect 50mV on a noisy electrical system is quite an achievement, as for your 1A error, that involves a 0.25mV drop across your shunt, you do not have a hope in hell's chance. People think because the display shows 3 or 4 decimal places it is accurate, often those digits are a work of fiction ! ( unless you spend a fortune on expensive circuitry ). Even 10A requires detection of 2.5mV, if you put a scope on the signal and checked out the voltage spikes you would appreciate the challenge. You will always have a value, and a figure on display, but it is not much more than a guess ! There are clever filtering techniques but you would need a Paul Daniels filter for that ! Good luck and please post back the outcome !

[dmr]

Fair enough Dave, each to their own. The shunt method is more common and what people would think of first and foremost. But any solution will involve a compromise, whether people realize or not. I haven't re-checked the specs recently but seem to remember linearity <1%, and the electrical isolation appeals to me.

Also, with a shunt ( say 200A == 50mV ) you will need to use an op-amp that will run off a single supply ( 0 -12v ) AND operate right up to its' voltage rail ( not all, especially the cheaper ones do this ). Otherwise a dual rail device ( +/- 12v ) could be used. As for accuracy, to detect 50mV on a noisy electrical system is quite an achievement, as for your 1A error, that involves a 0.25mV drop across your shunt, you do not have a hope in hell's chance. People think because the display shows 3 or 4 decimal places it is accurate, often those digits are a work of fiction ! ( unless you spend a fortune on expensive circuitry ). Even 10A requires detection of 2.5mV, if you put a scope on the signal and checked out the voltage spikes you would appreciate the challenge. You will always have a value, and a figure on display, but it is not much more than a guess ! There are clever filtering techniques but you would need a Paul Daniels filter for that ! Good luck and please post back the outcome !

 

If a Hall effect sensor can meet the required spec. then they would be the obvious choice and as you say the isolation makes everything so much simpler (and safer). I did think about using two sensors, one for higher currents and a much smaller one for the low current discharges, but even though they are electrically isolated devices it is still not possible to subject them to gross overload.

 

Don't worry about all the op-amp stuff and voltage rails, just have a look at the data sheet for something like the INA219, its all in there, the zero drift differential amplifier, variable gain, filtering, and an I2C interface. They even work out power if you don't want to do that yourself in software. One of these, an Arduino, and a little LCD display and you have all the major building blocks of a good amp-hour meter. I suspect this is exactly what is inside the Sterling device

[dmr]

and if you do find a Hall device that can measure 150 amp and do a resolution and accuracy of say 0.1 amp then please do PM me, I always think with electronics that whatever you wish will turn up if you wait long enough.

I believe some sensors work by supplying an equal and opposite current (through many turns) to balance the measured current and so the Hall sensor only needs to detect zero current. These have a better spec but then have the penalty of higher current consumption. Maybe using one of these and only turning it on to make a measurement once every second or so would be the way to go.

 

.............Dave