ADA4530-1 TIA How to use it as a Current sensor inline with a larger load

[Cad]

Hi Peeps

So iv been trying to design an opAMP that could be used to measure the voltage coming of a shunt resistor to read a current range of 1uA to 200mA ( without changing settings ) After a lot of reading the suggestion seems to be to use an OpAmp as a transimpedance Amplifier as this will provide the best accuracy vs Bandwidth. I need a minimum of 10Khz.

I have come across a purpose build TIA ADA4530-1 which seems to be good for this type of application but my problem is I have no idea how to send/use it as a burden/shunt resistor voltage pickup( attach is schematic. I know in this configuration the input expected is current and not voltage. so the question is how do I connect it to measure current across a shunt resistor.

See images for design details

[Kleinstein]

1 µA to 200 mA is quite difficult range, as with 200 mA the power loss at the resistor can become quite significant. So one can not use a very high voltage. Chances are the resistor would be limited to some 10 Ohms or so, so 2 V at 200 mA and this only 10 µV at 1 µA.

The ADA4530 is a nice OP for fA currents, but not so much for the µV range. So it would be the wrong OP. Here I would more consider something like an ADA4522 - so an AZ OP with low noise. With a 1 µA lower limit even 1 nA of bias current does not matter. The OP would need an additional buffer to drive the current. This would be a good idea even for smaller currents from some 1 mA on.

The current is no over a shunt, but through a shunt / resistor. With an TIA one would have the precision resistor / shunt as part of the TIA, not a TIA to sense the voltage over the shunt.
In principle one could use an 10 Ohms shunt and amplify the voltage. However this would have a high burden voltage that often is not acceptable.

[Cad]

Hi @Kleinstein
Thank you for your input that was helpful. So I thought about this and I want to use the high precision stuff and i will to get the best accuracy over a larger range rather then a tiny range and not usefull for me.

So my idea before i read about the TIA options was to use ADA4528-1 or even LTC2054/LTC2055 as a shunt burden resistor voltage amplifier but with a difference, I will be creating the 5v supply to feed the system and the feedback pin will both be tied to the VRail after the shunt resistor aswell as monitored in software and output voltage controlled via software. this way i can dynamically compensate for the burden voltage. The output will be fed into a 24bit ADC and read a minimum of 10Ksps. What I am struggling with is to figure out how accurate this device would be over the entire range. is 1uA accurately possible? i still want to deploy all the air wiring technics, metal boxes, etc to minimize noise.

EDIT
Will always range between be a 0.9V to 4.2V system supply DUT

[Kleinstein]

The TIA is kind of doing the compensation with the OP. Doing a similar thing with a shunt + amplifier and software control may be slow. So there would be still some change in voltage when the current changes fast. The TIA way is usually faster and thus little overshoot in the voltage.

With a usually 5 V powered SD ADC one would normally choose the FB resistor to match the ADC full scale range (usually some 2.5 - 5 V). The TIA output is single sided, while the high resolution SD ADCs are usually differential. So one would either need some extra circuit to produce a differential signal or accept the slightly higher error when using the ADC ground referenced.  So the part is similar to a 6 digit amp meter for a single range.
1 µA resolution should be possible, though it would not be at a high speed. Because of mains hum it would likely need averaging over 20 ms or a multiple to get to the 1 µA level.  Noise wise it may be slightly faster, but not much.
Due to heating of the shunt accuracy in the 1 µA range would be difficult  when the current is high (e.g. > 50 mA). Much depends on the money spend on the resistor. Another point would the calibration and ref. stability. This gets tricky once you aim for better than 100 ppm or even 10 ppm. ADC linearity can also become an issue - the linearity specs are usually for differential drive.

[Cad]

I see your point about the device is slow to compensate. I don't have a problem spending the money on the resistors so tolerance will be low, Thermally stable no an issue,
I just don't know how to integrate a TIA in the manner of which you speak. Iv been reading the PicoAMP thread and see the way they do it but don't understand why what is where. ( very limited Analog design knowledge ) especially to modify the system to measure my required range,  If i can get the TIA setup correct then a simple 16 bit ADC in MCU should be good enough as single-ended input.

[Marco]

The concept of a TIA is that you connect the load to a virtual reference point and let the amplifier pull the current through the sense resistor to maintain that virtual reference. So more like the following. The opamp will probably need an extra class AB stage to pull 200 mA. Analog has a circuit to do single ended to differential for feeding your  LTC2380-24 ADC.

PS. forget about using a microcontroller ADC, the dynamic range and resolution you want requires a state of the art 24 bit ADC. If you just require the dynamic range but could make do with somewhat lower resolution at higher current ranges you could maybe do something like the second image. Note that zeners below 6V aren't really Zeners and don't have a steep transition ... but you don't strictly speaking need that, any current not going through the high current range shunt will still be measured by the low range shunt and vice versa.

[Cad]

Hi Marco

Thank you for the input.
I looked at that ADC and not sure I follow as to how this would operate. I have attached a crude image of the setup to use for clarity.

Question
1) Where does the current flow to? I see that ADC has 10ma max sink current and does it flow in the direction drawn on the image?
2) Is that how you would more or less connect the different stages or is the image you provided in place of the OP1177 ?
3) Is this only capable of low side sensing ( doesn't matter just asking)

My sense resistor would need to be 25R for 1uA to 200mA to provide the scale,

Iv attached another picture, of a possible setup, not sure if this is what you mean? So the burden voltage will be sinked through the 1R that will produce 25uA at 1uA Load and 5mA and 200mA load?
This could be 2R to get closer to the limit.

Apologies for all the questions,
 

[Marco]

1) Where does the current flow to?

Through the resistor then through the opamp output to the negative voltage rail (not to the ground at the positive input, in normal circumstances essentially no current goes into the inputs of the opamp ... it's essential to operation of the TIA). The opamp generates whatever negative voltage at the output which is required to make the bottom of the load and the negative input be at ground potential. The opamp does need to be able to sink 200 mA, which rules out most precision opamps without some kind of booster stage.

2) Is that how you would more or less connect the different stages or is the image you provided in place of the OP1177 ?

You need the circuit from 3a, the output from the TIA will be a negative voltage and only the 3a circuit will convert that to a positive differential signal for the ADC (I think at least, this isn't quite the way they mean it to be used ... it's meant to work with a positive voltage, but I think it still works with a negative one).

3) Is this only capable of low side sensing ( doesn't matter just asking)

This specific configuration yes, but the concept can also work high side.

Iv attached another picture, of a possible setup, not sure if this is what you mean? So the burden voltage will be sinked through the 1R that will produce 25uA at 1uA Load and 5mA and 200mA load?

You still don't get what a TIA does, what I called the sense resistor is actually usually called a feedback resistor. Probably better to call it that.

That 25 Ohm resistor you drew in shouldn't exist. The feedback resistor needs to be 25 Ohm. The bottom of the load is directly connected to the negative input of the opamp ... just like I drew it.

[Cad]

Hi Marco

Thank you for the info,
I understand what you said, and can see the approach. Its just the first time I see that I see this setup and that a TIA will be able to sink up to 200mah through it to ground

Im okay with the device is being high side sensing, as I will be providing the power to the DUT from the same board. What would you say is the smallest that Sense/Feedback resistor can be and still get good accuracy on the low end? 

[Kleinstein]

One can reasonably resolve some 1 µV. Much below that it gets tricky with thermal EMF and offsets.
So to resolve 1 µA it would be about 1 Ohms as the lower limit. It is a balance between offsets at low currents and self heating of the resistor.

For more than some 1 mA there should be an extra buffer after the OP in the TIA to boost the current capability.

[Marco]

Im okay with the device is being high side sensing, as I will be providing the power to the DUT from the same board.

High side makes things a bit more difficult, because the extra level shifting step to low side introduces yet more errors.

That said, given the dynamic range you want together with DC accuracy you probably already can't get away without chopping.