ultra low power instrumentation amplifier

Hello Vaidas,

When low operating current is the most crucial factor other electrical performances must be compromised. Those most affected are bandwidth and noise. But if one's application can accept those for what they may be then a very low current instrumentation amplifier can be realized.

TI's latest precision, very low supply current Op amp is the LPV821. Its quiescent current is typically 650 nA. Therefore, three of these Op amps configured in the 3-op amp instrumentation amplifier (INA ) configuration would have a supply current of approximately 2 uA. If your application can use the 2-op amp INA configuration, the current would be even lower by a third. You can see the LPV821 datasheet here:

http://www.ti.com/lit/ds/symlink/lpv821.pdf

Below I have configured three LPV821 Op amps as a 3-op amp INA. The quiescent current is 2.1 uA, and the supply current increases with the increased output swing. The increased current occurs due to the higher current flowing through the feedback and input resistors acting as a load on the Op amp outputs.

The bandwidth is about 32.5 Hz using the LPV821 in this circuit.

There is another TI Op amp with comparable quiescent current to the LPV821, the OPA369. It isn't quite as dc precise as the LPV821, but it does provide a bit more bandwidth in the same circuit. The -3dB bandwidth is about 243 Hz for just a fraction higher current. Here's the response obtained with it:

You can find my TINA Spice simulation circuits for the LPV821 and OPA369 INAs here:

LPV821_INA_01.TSC

OPA369_INA_01.TSC

As mentioned earlier, there is also a two op amp INA configuration that might be usable and it would have less quiescent current. Here is an image from the OPA369 datasheet:

Note that with any INA that the close matching of like value resistors is imperative. If you find that one of these suggested INA circuits meets your basic requirements you will find the next hurdle is finding closely matched resistors at an acceptable cost!

Regards, Thomas

Precision Amplifiers Applications Engineering

Hi Thomas

I need to amplify only positive signal.

At our schematic Input signal has virtual ground at 1.3V (minimal virtual ground for sensor is 1V), but output signal i need  referenced to ground for MCU. So we have R6 connected to ground and we have current flow through resistors R3, R4, R5, R6.

Check out the schematic below.

Signal bandwidth i need is only 1 Hz. 

Hello Vaidis,

Okay, U1A and U1B are applied in a potentiostatic sensor circuit, followed by U2A, U2B and U3A which form the instrumentation amplifier. U3 is a straightforward non-inverting gain amplifier. If you want increased CMRR and higher gain accuracy from the instrumentation amplifier I suggest using 0.1 % tolerance resistors, instead of 1 % tolerance resistors. The cost a bit more, but the mentioned electrical performances should be improved with them. 

This all looks reasonable with the possible exception of C5, 1 nF at the output of U3B. Ultra-low current Op amps can have open-loop output impedance (Zo) that is high and complex (R+/-jX), and can vary wildly with frequency. When that is the case for the Zo adding a load capacitance at the output may result in a much degraded phase margin. That in turn can result in poor transient response, or even oscillation. You may want to check if U3B is sufficiently stable.

Regards, Thomas

Precision Amplifiers Applications Engineering 

Hi

I understand regarding U3B. i will fix it.

But the question is how big  can be R15, R12, R20 and R21 resistors (i need to decrease power consumption) to have stable operation of U2A, U2B and U3A using TLV8811 op amp?

Vaidas

Hi Vaidas,

The TLV8811 was developed by another amplifier organization within TI. Since I am not too familiar with the Op amp I needed to test the simulation model to make sure that it has correct open-loop gain (Aol) and open-loop output impedance (Zo) characteristics. These must be modeled correctly for the model in order to perform a meaningful stability analysis. I ran testing and found the Aol and Zo are close to what is shown in the datasheet. I decided to run a stability analysis with all 4 resistors set to 1 Megohm as in the INA's differential amplifier stage to see what phase margin resulted. A 10 pF capacitor was added to the output to account for stray capacitance and any small load capacitance.

The results with all the resistors set to 1 Megohm indicates 67 degrees of phase margin, which is very good. You can see the results below. We strive for 45 degrees or more to assure stability. Since there is that much phase margin you should be able to increase the resistors by several multiples and still maintain sufficient phase margin.

I think the bigger concern with the very large value resistors is the thermal noise which is:

En = √(4kTRΔf)

Where: k = Boltzman's constant 1.38e-23 J/K, T = temperature Kelvin, R = resistance in Ohms, and Δf = noise bandwidth.

With very large resistances the thermal noise can be become a dominant factor.

Regards, Thomas

Precision Amplifiers Applications Engineering

Hi

Next questions i have:

1. What could be safe maximum resistance of  R8, R7 divider ? 

2. What Riso we have to put between U3B and C5?

3. do we need to put protection diodes for OP Amp inputs?

Vaidas

Hi Vidas,

1. What could be safe maximum resistance of  R8, R7 divider ?

The R8, R7 divider establishes the common-mode voltage (Vcm) that the TLV8811 will operate with. The current values result in a Vcm of about +1.3 V which is nearly half the +3 V supply with which U1 is being powered. The TLV8811 Vcm input range is 0 V to about (V+) -0.9 V; therefore, you want to make sure you stay withing that range. Certainly the poteniostatic sensor has optimal voltage requirements and those would be key in determining the voltage that is used.

2. What Riso we have to put between U3B and C5?

The TLV8811 has a very high,complex output impedance. Driving a 1 nF load capacitance is a very large load capacitance as far as its output is concerned. When I did a stability analysis with the 1 nF directly at the output with no Riso the phase margin is negative and Op amp will oscillate. The ideal Riso approaches 1 Megohm which probably isn't acceptable from an application standpoint. I did some testing with lower value resistors to see if an acceptable phase margin could be attained. The results below were obtained with an Riso of 100 k. The phase margin is 42 degrees which should be sufficiently good to assure stability when driving the 1 nF load. If the voltage drop across the Riso can't be tolerated there is a dual feedback method that can be applied to correct for the drop.

3. do we need to put protection diodes for OP Amp inputs?

The TLV8811 has built-n ESD protection. Nothing more should be required once the Op amp is installed in the potentiostatic sensor circuit board.

Regards, Thomas

Precision Amplifiers Applications Engineering

Vidas,

The second graph in the previous Bode plot should be labeled "Phase (degrees)," not "Voltage." Not sure why TINA decided to do that.

Thomas

Hi

Thanks for answers.

first question was about divider resistance. Now it is 900k, so  current through divider is ~3uA. Can i reduce it to 1uA?  For example resistor divider would be 3M?

Is it needed to put capacitor in parallel to R7?

Hello Vaidis,

The TLV8811 has extremely low input bias current. It is listed in the datasheet with a typical of +/-100 fA. Therefore, very little current loading of the R7, R8 voltage divider would occur due to the TLV8811 input bias current. The two divider resistors could be increased in value to reduce the current consumption from the battery. Higher resistance does result in higher thermal noise, but that is normal and unavoidable. 

Adding a capacitor across R7 would provide some noise reduction, especially as the frequency of the noise increases. That may be a desirable feature and may be helpful if any RF from unexpected sources make their way into the circuit. However, adding a capacitor across R7 will set up an RC network that may have a longtime constant. So once power is first applied to the circuit the Vcm will take some time to come up to the final level. That may, or may not, be a concern for your application.

Regards, Thomas

Precision Amplifiers Applications Engineering

So we can decrease current to 1uA of R7/R8 divider.  i understand the answer  is yes?

And some small capacitor will help to lower noises.

Instrumental amplifier with TLV8812  works fine. We tested.

Now we will try to use TLV8812 for potentiostat circuit. I will inform the results. 

Vaidas

Hello Vaidis,

It is good hear the TLV8812 is working well meeting the very low operating current INA portion of the application. It should work well in the potentiostat application as well. The very low +/-100 fA input (typ) bias current is ideal for that application.

Please do let us know the results you receive from the TLV8812 for the complete application.

Regards, Thomas

Precision Amplifiers Applications Engineering

Also keep in mind if your desire is only 1Hz, you are now in the 1/f noise region, up to 500nV at 1Hz with the TLV8811. Maybe should run an output SNR for the whole thing, 

There was an earlier file using the LPV821 showing 2Vpp, that is 0.707Vrms for the SNR simulation. Might try something like this with your current circuit. 

Hi Vaidas,

If you have the information you need for now and could close this e2e inquiry it would be helpful to us who support the Precision Amplifiers forum. If you need to reopen the discussion a new e2e inquiry can be started at any time.

Thanks, Thomas

Precision Amplifiers Applications Engineering