OPA2141 Problem with current senseing, differential output

You must have a DC path to ground on each input of the op amp to properly bias the front end of the op amps.  That path can be through a resistor tied to ground or supply on each +input node of each input op amp.    This is true for any op amp.  

Hi Tim,

Thanks, but ....

The OP-amps generate a bias current Ib, which is +-5nA over full temperature range for this type.

This current generates a voltage across the two added resistors. When I use small values like 100k, this drop is about 500uV. On the other Hand, 100k is very low, cause my greatest shunt is 2k (+- 50µA full scale) so 200k introduce a nice error ... Higher values, like 10M give good accuracy, but the drop by Ib is large now.

As Long as the two voltages were more or less the same, this will be no Problem, but whats going on, if one amp have Ib = 5nA and the other Ib = -5nA ??

Or other question: How can I build a amp with high Input resistance, having a DC path for Ib and didn't get large Errors?

Yes, I know the OP amps with very low Ib (some fA) but the noise Level of this amps is large ..

With best regards

Gerhard

Give us a few more system level details as to what you are trying to accomplish and maybe we can recommend a solution. This current you are monitoring differentially - where does it come from and how does it flow relative tot he return of your +/-14V power supplies? The more system details and what you  are trying to achieve from top level down to circuit level will help us help you.

Hi Tim,

I Need to measure a current, which variys widly, but max. is 1A. The current is generated by a high voltage source, driving current through a material probe. Voltage can be 400V max. To prevent danages caused by short cuts and other effects, the current monitor should be isolated. Unfortunately, sampling the voltage and the current cross the material probe should be simultaniously, so simply use two ADCs and use digital Isolation doesnt work.

Analog Isolation is hard, only the ISO124 is there, and this part has a poor noise figure, related to other low noise op amps..

So I want to amplify the current pickup signal to a high level, pass this level though the iso barriere and use a ADC 1274 to do the sampling job.

My current is between 1A and some µA, very high dynamic range and to get values of high accuraty, I need to switch between different ranges. At the Moment, I think about 1A, 50mA, 1mA and 50µA range, full scale.

First idea was to have 4 resistors in series, three of them can be shorted out by a switch. A differential analog mux routes the ends of the correct resistor to a differential amp. This way, the drops of the switches doesnt matter. One resistor, the smallest remains always active to prevent a short cut as the controller do some junk with his relays/analog Switches.

The current sensing circuit is powered by an isolated power supply, fly back converter, dual output with LDO post segulator, giving nice supply.

The System is modularized, and the current sensor is one module. He gets the raw isolated and non isolated supply from the mainboard and creates the clean analog supply on board. The isolated ground and the ground coming from the mainboard. This was no problem. as long fully differential pickup is used but maybe taken into account using a single ended solution.

Again, the current is bidirectional DC and should be measured with high accuracy, think about this sensor as an good desktop amp meter.

Thanks a lot for helping.

With best regards

Gerhard

Hi Tim,

here one of my single ended ideas, but has the disadvantage of a direct path through for the current if Switches were in a wrong position. Other Version, with ground fix connected at the low end of the resistor chain has the disadvantage of getting effects of the Ron of the ranges switches (currently bidirectional solid state Switches, leakage 300pA)...

Thanks again for helping

With best regards

Gerhard

I will post 3 separate posts with TINA schematics.  Attached is an architecture idea for your consideration. This IMON Concept I think might yield better results than your original concept.

Attached is a concept for high current bidirectional measure on one isolated supply.  Most ADC inputs are single supply.  You might look at a simultaneous sampling ADC to read voltage and current at the same time. Some parts my colleagues have recommended for investigation are:

 ADS8353./ADS8354, ADS8363, ADS8556/ADS8566.  

With same single isolated supply here is a concept to monitor the battery voltage.

Hi Tim,

Thanks, but ..

Your IMon concept is that from what I started and has disadvantages. There is the Need of four relays, able to Switch 400V and >1A. As the System is used far off road (mountain Peaks, >3000m ...) and has to carried there by man power, it should be small and light-weight.

Using SSRs to short out the resistors never opens the circuit, so I can use parts with a footprint of a 4 pin SOIC, carrying 2A continous, using very low energy to Switch. The whole sensor is guarded by two anti parallel diodes, so this switches never see more than 2V. I found diodes (SiC) which offers a leakage current of some fA under normal conditions, so this works fine.

You told me, that there should be a DC path to ground from each OP-amp input, but in your concept, this path is missing??

I plan to use the AD1274, a Sigma-Delta 4 channel, simultanously sampling ADC. I need higher bandwiith, cause there is planned to measure time domain effects, so I have to observe the potential at certain points after switching off the current.

I also should try to have very low flicker noise (DC measurement), but I can't find an Isolation amp with good noise figure. So I have to pass very high Level signals through the iso-barrier to maintain the best SN possible.

I will check out the other two ideas observing batteries, maybe there is something which can help me.

With best regards

Gerhard

Hi Tim,

as I thought, I have to add a DC path to ground to get correct values with the Instrumentation amp. I made a file, I have to use the INA128E model here, cause there is no INA141 model available. But it seems, there is no difference execp the resistor Network used internally.

Problem is, 5nA Ib, when I use 1M I end up wit 5mV .. as Long this voltage drop is more or less identical for both Inputs, the measurement is correct, but if not ....

Noise Performance is really good, and using a single ended Output I have +-10V instead of +-20V, so less dynamic range cross the ISO124, but now I only Need one ISO124, hope the noise added by the ISO124 is tolerabke.

The converters you list above were 'only' 16 bit, we use 16 bit now but want to extend, cause we see some dynamic Problems. This is also the reason, why I looking so Close to noise figure (flicker!!) and Ib effects.

With best regards

Gerhard

I think you model of an ideal current source is not properly modeling the real world for your application.  If I understand your system correctly it is a 400V battery with a current sense resistor in series to a load that will draw current. If that is true you do not need the 1Meg resistors.  The battery monitor circuits I sent illustrate this concept.  If my assumption is not true help me understand a bit clearer what you are trying to achieve so if I have any ideas I can send them your way.

Hi Tim,

yes, use an ideal source is maybe a problem.

What do we have:

There is a switching power supply, working as a constant current source, delivering a current between some 100µA to 0.75A, max. output voltage is 400V. This source is isolated. Sometimes the material probe is very high resistive, so 400V giving only 20µA or so. The current flows to our material probe and generates some voltage, this potential is picked up at certain points and measured together with the current. Is  more or less a 4 wire resistance measurement. To get out effects caused by the electrodes used and other effects, we change the direction of the current half the way.

Now I have to measure the current precicly and the potental, but sometimes the potential is taken nera the one end or the other end. So Isolation is a good idea otherwise I will end up with very high common mode voltages, not so easy to deal with over full temp range.. As there were more potential channels, I think, isolating the current channel is the easier way.

As there is a dynamic range of some µA to 0.75A I Need a range selection to get high Resolution results. As we want to use 24 bit ADC (16 bit uptil now) the analog frontend should be low noise and well done to get good results.

First try was a pickup amp with two LME49990, fully diff Output and two ISO124, which gives a nice noise figure, but Ib of the LME49990 is very high over the full temp. range. Switching between ranges should also avoid large relays, cause the Equipment has to be carried at Points like tops of mountains ..

Hope you get a better idea now, what I have to do here.

With best regards

Gerhard

You have quite some detailed challenges ahead of you. Attached TINA-TI SPICE schematic may be of interest.  Forget about the polarity reversal of the voltage source for now.  Also ignore that the op amps I chose may not be the final for lowest offset. You need to accurately measure both current through the probe and voltage across the probe.  The attached circuit does this by robbing some current from your constant current excitation, as shown when you do a DC simulation.  It does not matter since the actual current through the probe is measured  accurately.  I think it is easier to see what concepts would work on one polarity of the supply and then add in later reversing the supply to minimize electrode issues. Hope attached helps you with some ideas for a final solution.