Low Current at high voltage Reading.

Hi All.

Can you suggest a good design that will measure current of around .1mA to 100mA at 110-230 Volts? as normally greater than that I can use shunt resistor, also with a good accuracy on reading that current.

Or any commonly use scheme for this type of design?




Thank you

The most common and easiest way is the shunt resistor method.

Another popular method is using a Hall effect sensor method. You can get those from Austria Microsystems

Here is the circuit of a high current circuit as example.

yes on shunt, but on requirement of .1mA, can i still have accurate reading with that, or what probably shunt be use at that current to achieve 1% accuracy? also do you have some app-note on hall effect sensing? about its accuracy?

Sorry loren_tm its not from Austria, its from allegro : http://www.allegromicro.com/en/Products/Categories/Sensors/currentsensor.asp

my bad.

Also I have found an excellent application note, that includes comparison of different current sensing techniques and details on Hall effect sensing : http://www.allegromicro.com/en/Products/Design/current_sensing/bsp_v1_52.pdf

Here is an indepth article on Hall Effect sensors : http://content.honeywell.com/sensing/prodinfo/solidstate/technical/chapter2.pdf

To get a the accuracy you need, everything depends on how you calibrate.

If you are using 10K resistor with -5% tolerance, .1mA would produce about 95mV. Now if you have not calibrated and you calculate using 10K, you get 5% tolerance. but if you have calibrated the current reading using an ammeter or a DMM, you will get the same tolerance as the original DMM.

Using a 250ohm resistor would mean the minimum read value would be
V=250*0.0001 = 0.025V = 25mV
At 100mA
V= 250*0.1 = 25V which is too large a voltage drop.

The range you specified seems to be too large. A 1000 fold increase. I would suggest using a Hall effect sensor. But still the lowest reading product from allegro produces 66-185mV per Amp.

You can increase this value by using an inductor but that would imply many consequences.

I have attached an application note from Sentron here.

Hope all this will help you.

Just to mention an obvious fact, a hall effect sensor, particularly a highest precision compensating type, will finally convert it's measurement by a shunt. So there should be no doubt about the precision achievable with shunt measurements...

The problem with shunt measurements is in the galvanic isolation requirement of course. A hall effect transducer can be one of many solutions for it. The said direct measuring sensors as e.g. supplied by Allegro have a rather poor signal dynamic and can't work well for a 1:1000 range. Compensating sensors, e.g. available from LEM can do, but are more expensive. By using multiple turns at the primary, the intended current range shouldn't be a problem.

by using multiple turns at the primary, the intended current range shouldn't be a problem.

Click to expand...

you mean using CT?

The problem with shunt measurements is in the galvanic isolation requirement of course.

Click to expand...

can you give me some range on current measuring when this factor would eat up my signal?


@FVM, can i have good result using shunt on a shunt resistor? maybe il divide my range like
.1mA-10mA--10mA-60mA--60mA-100mA-->16A
so ill have around 3shunt, with relay for it. is this look possible?

you mean using CT

Click to expand...

I was talking about hall effect sensors or "DC current transformers".

Using multiple shunts with multplexing relays would give highest dynamic range of course.
A 1:1000 range, as specified above, should be possible with a single shunt, I think.

FvM said:

...The problem with shunt measurements is in the galvanic isolation requirement of course....

Click to expand...

Hi,
Newly they are isolating current sensors (+/-15A & up,I think with 1mA resolution) from Isabellenhütte on the mart, but they didnt have these dynamic...
K.

any alternative on using relay? to lower switching time and increase lifespan? maybe going shunt, its simple and low cost.

Hi,
problem is with alternative techniques to relays, that a relay is practically an ideal switch!
If you have to select between small resistors (at tenths of Ohms or less, but by some Ohms too): you will have problems with the added contact resistors!
They are relevant in values(non negligible) to shunt values, plus they are instabile in lifetime, temperature...
K.

Here is one possible solution. Along with he series resistor method use an Analog MUX to select the resistors. (If the MUX can't supply 100mA, try using a shunt for one series resistor. You can use DIP switches ar another switch, but make sure you are isolated. (As it is a direct measurement, the circuit can never be isolated)

I am also posting a crappy solution. (Just to give you an idea to make this complex)

Added after 8 seconds:

I am just about to get you a crappy solution. Please so not consider this as an option. It is just here to let you know, there is a solution.

This doesn't provide any isolation and hence can be dangerous!!!

Here is an outline of what you could possibly do (only if you are desperate) Use two series low values resistors on the line. Add a 2 resistor potential divider after the first resistor and then again after the second resistor.

Phase --------/\/\/\--------PontA--/\/\/\---PointB----


Neutral ---------------------------------------------------

Place the first pot divider b/n PointA and Neutral and the second pot divider b/n PointB and Neutral
Take a the output from the pot divider and give it to a logarithmic difference amplifier to produce a logarithmic value so that you can have all the dynamic range in the world, no loading and good resolution if you select the resistors correctly.

If you can't build it all once, build a unity gain difference amplifier and then a log amp.

Thanks eXnol for your desperate advice

anyway, saw 1 unit of current meter that uses .02 ohm shunt resistor. and could measure down to .01mA, another problem would be, do you any idea where could i buy such shunt? and maybe if i have that site, it will be a good start. i saw application note on that multiplexing at atmel.

Hi,
Pls chek out Farnell: https://www.farnell.com/
Isabellen Hütte a German wendor, IRC, Vishay, Welwyn..., but search pls for "low ohm resistors or 0.01 Ohm current Shunt"...
Practically all the bigger vendors has that.
K.

It's betetr to use 4 terminal shunts. They can get you better reading when such low value resistances are used. Farnell, digikey and Moser are the major players. I do not know your country. So, can't get more specific.

Hi eXnol,
I thik, my suggestion so 10 days ago are same/very similar ...
@leoren
Anyway;
A Kelvin resistor can be relative expensive, but with a good layout you can modell it with all SMD resistors! You must guide the "sensing lines" into the middle area under tha resistor from both pads & is your "Kelvin resistor" modelled_finished!

In my Opinion their is a bigger problem to dimensioning correctly the cooper surfaces around the resistors, then they need relative much cooling! :-(

For these problem are the most datasheets not correct/needs reading in detail... :-(
I have good experiences with Isabellenhüttes 2/3W series, but would forget i.e. Panasonics 2W types; they need if I correct remember a minimal cooper surface of ca. 2x2 cm!!_I have had lot of problems with these type in PWM circuits...
K.

You have 60db dynamic range. Sinve you need isolation too,i would recommend using some miniature isolation amplifiers.Maybe some digital output parts are better.You can check ACPL-C79B.