High side current measurement

high-side current measurement

Hello ,

Can anyone suggest how I can measure the current using a shunt resistor ( 1E,300mA with a potential of 30VDC).
A 3 OpAmp differential circuit using OP07 was what I had designed, but it needs the one of the inputs connected to ground which will obviously not work in this case. I need to amplify by a gain of 10 and feed it to an ADC.
As an alternative if someone can suggest some current sensing ICs commonly available in India,it would be helpful.

current mirror high side measure

The question is not clear. Why can't you use a current to voltage converter circuit using a single op amp?

lm358 high side current sensing schematic

Take a look on this topic.

**broken link removed**

PDF
**broken link removed**

Hope it help

Bye
Powermos

maxim ic+high side current measurement

If special IC shall be avoided, a standard OP differential amplifier circuit (4-resistor circuit) can be used, if the common mode voltage is pulled down to the allowed range by two additional resistors. The resistors should have a good tracking(in pairs) to achieve a sufficient common mode rejection.

транзистор bcv62 zetex

Hello ,

Thanks for the replies and links.

FvM : Is the circuit you mentioned as shown in fig 4. of TekUT's link :
**broken link removed**
I will be making that circuit and putting an additional OpAMP to get a gain of 10.

Are you suggesting that Rs should be pulled to ground on both sides with additional resistors ? Can you oblige me with some link about your idea.

TekUT :
Your links were helpful in understanding a bit more,though I would probably not be able to source those ICs here.

NitroJacob :
Is what you are suggesting the same as in the link ?Actually

Another query :
Is a 25ppm resistor same as 0.025% tolerance ?

Regards to all

shunt current measurement circuit 2904

If you don't want to spend the money for a dedicated HS sensing IC,or they are not readily available for you,the LM358 should do the job. These are widely available and dirt cheap even the A grade is only 22 cents in small quantity. Other versions of the 358 are the 258,158 and 2904 different temp and offset.

If you do use a 358 you have to ensure the input CM range is not exceeded. This is done like FvM suggested resistively dividing the input down; in this case by about 1.5 or 2V check the data sheet on the 358 for sure.

Rather then explain it all here is a note where its used for current limiting from NXP.

shunt resistor for ina139

Some of the older JFET op-amps are also very cheap, but have the advantage that the positive common-mode voltage can exceed the supply rail; at least enough for this application.
The LF411 is one such device, but there are many others.

current measurement high side

MAX0412,

That was extremely useful. The appnote though speaks of 25% accuracy.I need about 1% accuracy. Most of the errors though seem to offsets which can be calibrated out in the software,I guess.
I will try this.
Is there any commonly available single chip solution which I can try for.
Thanks again.

Added after 4 hours 5 minutes:

Hello again ,

Regarding the appnote :
**broken link removed**.
It is working as a comparator I think.
I have modified it so that it works as a unity gain amp which feeds a second amp to get a gain of 10.
Any comments would be welcome.

high side current measurement

Hello Phaedrus,
It would be difficult to achieve 1% with this design unless you are willing to use high precision resistors, and even then...
Perhaps I can recommend the method used by Zetex in their devices.
The basic idea is shown in the datasheet;
**broken link removed**

As you can see, this circuit can be readily made from discrete components, but it does need an op-amp who's common-mode range is at least that of +V.
However, it is possible to bias the inputs using resistors, as in your example.
In both cases, the dividers result in a loss of sensitivity and offset. If software compensation is available, then it makes sense to capitalise upon it, and use the simplest hardware.

transistor current mirror high side measurement

Hello Humber,

Thanks for replying.
Yes, I would be using precision resistors. Over phone I was told I can procure resistors for 25ppm by the trader.This would pertain to temperature , I guess going by googled results. I am still not sure what's the tolerance though.
The output is fed to an 12 bit ADC and captured by a microcontroller. So I could do a bit of offset tweaking in software I suppose. With this what accuracy and precision I could end up with ?
Can you suggest any ICs I could digikey as a last resort ? The Zetex is for 20V , my app would probably have about 32V.

Cheers.

You can use zetex for above 20V,they explain it in detail in this appnote,as well as other considerations.


Maybe the AD211 Digikey.


I didn't think you could order from Digikey,just search current sense you'll get about four pages of results narrow it down to your Vcc. See also if you can find a RR opamp with uV offset in the Vcc range you need (this likely wont be cheap) then select the cheapest solution .

Heres is another appNote from analog,how to minimize errors.

Phaedrus,
It is difficult to answer these general-yet-specific questions and not mislead, but here we go.

Resistors are available in almost any tolerance, but price becomes an issue. 1% tolerance is now virtually standard, and therefore inexpensive. However, even using 1%, the divider bridge alone will consume your error budget, so software compensation or manual calibration will still be necessary. The accuracy that you achieve will depend upon many factors, such as the signal level ( so that the dynamic range of the ADC is usefully employed) and the method itself. Can you calibrate the system at 2 points, say zero and full scale ? What is the temperature range?

I was not suggesting that you use the Zetex device, but copy the topology using componants available to you. However, the ZXCT1081 is a 40V device, stocked by Digikey. This device has a general accuracy of 3%, so calibration will still be necessary.

Digikey has the LF411, and it seems they are available for less than 50 cents, but I would think that you could do better than this locally, and equivalents are more likely to be available.


If the sensing resistor is 1 ohm you suggested, then make R2 100 ohms and R3 1K, giving 3v out. If all other considerations are the same, this will be more accurate than the divider method.

**broken link removed**, **broken link removed**,... lot of manufacturers produce "single ship solution" . Usually offset errors will be main source of error, which is easy to compensate for. Nonlinearity errors are fairly small. Just go on web sites of major manufacturers and look for high side current monitoring or measurement IC.

Thanks to everyone for their time and replies.
Humber , very kind of you to draw the schematic for me. I will try this out and let you know how it goes.
I am using a 12bit serial ADC which has a 2.5V reference (LM336 ). I should be able to calibrate at 2 points in my software, after tweaking the R values to try and utilize the full scale.
As far as the LF411 equivalent is concerned, the parameter I should be looking out for would be : Differential voltage : +/- 30V ?
The datasheet sheet shows the supply voltage as +/-18V. Would it be ok to connect the IC supply to 30V and ground ?
Thanks for the support.

I've did a check on the circuit proposed, here a working schema, just take care of the common mode voltage on the input pin of the op-amp, to get all working the power supply must be correctly sized. Into the example I suppose to work with a 12 V power supply, voltage on the input pin of the operational (expecially the non-inverting pin) rise the main voltage then you need a power supply for the operational amplifier that is greater than this voltage, a 15 V or 24 V supply voltage is ok.

If you like to use a voltage of 5 V or similar you've to use a differential amplifier as into the reported schematic. To get all working match the resistor into the diferential amplifier, if you use same value for all resistors the gain is set to 1.

Bye
Powermos

Hi PowerMOS,

Thanks for pitching in to help.
This schematic may not work for me i think.
For me V1=Vcc=30V.
Well I have managed to locate a LF412 which is a dual version of LF411 suggested by Humber. The input offset voltage is higher,but probably can be calibrated out.

Hi,
you can try with this schematic, opamp supply voltage is a dual 5 V (then +5V and -5V), for the negative supply voltage you can try to use a simple switched capacitor voltage inverter, there is many type around, one that I've used and work well is the MAX828).
Take in mind that using 30 V require input attenuation, output voltage is equal to the Rsense current divided by 10 (look at the simulation waveform, signal VOUT2.

Hope it help
Powermos

Thanks to Powermos for the simulations.
I don't think it is a matter of whether the designs will actually work, but the relative cost and ease of achieving the same result.

Dividing the common-mode voltage with resistors will work of course, but the sensed voltage is also divided by the same amount, ( making it smaller relative to the op-amp's input offset) and there are now 4 resistor tolarances to consider. This method allows lower input-offset bi-polar op-amps to be used, but the advantage is largely erroded by the above errors. Accuracy is not readily defined by a single number. For example, it depends upon the temperature and the range of measurement, so perhaps the errors I have just mentioned may not be significant if you need 1% accuracy only at the 300mA leve.

The supply voltage of the LF412 is the sum of both rails, so grounding the -ve supply will allow positive voltage of 36V. Connect the +ve rail to the supply side of the sensing resistor. The LF412's input offset voltage is nominally 3mV (1% of the maximum sense voltage). The advantage of this JFET device, is the the input can exceed the +supply. It's not a lot, but enough. I think this offers the lowest cost option, which can be brought to 1% by calibration.

With calibration, the absolute tolerance of the components is not as important as their stability, so there would be little advantage in say, using 0.1% resistors over cheaper 1% resistors. The LF412 is by no means the only choice. There are many modern rail-to-rail opamps that could be used, but I have been working on assumption that you would rather avoid special or hard to get components.

With a 12-bit converter and 2.5V dynamice range, there is more than enough resolution to achieve a reliable 1% measurement. Be careful though, as the accuracy will depend upon the load current - errors will become more significant at lower lower levels.

Cheers.

Hi Humber ,
Is there any generic name for a circuit of this type, I mean with the transistor ?
I am waiting for the components,will surely post back when I have something tangible.

phaedrus said:

Is there any generic name for a circuit of this type, I mean with the transistor ?

Click to expand...

I think your talking about matched transistors "current mirror".

See this from zetex.


These aren't cheap 4 Bucks in this case.

You could try these. BCV62

I use them small and pretty cheap. They dont have the NPN complement in stock.


If you wanted to go cheap and are using SW to minimize errors you could try it with with single transistors.

*Edit the BVC62 is only rated for 30V.