Op-amp Input Impedance

Hi Mathew,

In the electrical characteristics section of THS4531 Fully Differential Amplifier, the specified 200 kΩ differential input impedance is the impedance  between the VIN+ and VIN- pins of THS4531 and 200 kΩ commonmode input impedance is the impedance between VIN+ / VIN- pins and ground under spesified test condition mentioned on the top of electrical characteristic table. For your application, since you are having a sensor with source impedance which can vary between 1k and 5k, you can use a buffer stage and then use THS4531. If you can give some information on which configuration you are planning to use THS4531, we will be glad to help you to choose the right amplifier for your application.

Thanks and regards,

Eldho

Hi Matthew,
1. What kind of sensor are you using?
2. Is it current or voltage output?
3. Does the sensor have a single-ended or differential output?

The THS4531 is a good choice of a fully differential amplifier, however if you want high input impedance, the THS4120/THS4121 may suit your needs.

-Samir

Samir and Eldho,

Thanks for your responses. I am using a novel sensor which was developed to measure heat flux. It outputs a differential voltage in the +- 40 mV range.
Thus, I am looking for a fully differential amplifier to interface with an ADC with differential inputs (the ADS8354 - www.ti.com/.../description).

Are there any good solutions for a differential buffer, or would I need to buffer each end of the sensor's output separately?

Matthew,
We have a few options for dual channel opamps that you can use. My suggestion (if you need very high-input impedance) would be to use a dual channel CMOS amplifier in a non-inverting configuration with some gain. You can follow this up with a fully-differential amplifier to change the common-mode voltage to match the ADC requirements.

1. What is the bandwidth of your application?
2. Also, you mentioned that the signal has a differential +/-40mV output. On what common mode voltage is this differential signal riding?

Before we go down that path however can you please tell me if the output of your sensor is well matched between the differential outputs? In other words if one side of the sensor has 1kohm of output impedance will the other side be 1kOhm +/- 0.1%? or can the other side be way off, like 5kohm. If the two sides are well matched then I think the error will be negligible and you can continue to use the THS4531A in the standard fully differential configuration. The reason I don't think there will be an issue is because at the input pins of the fully differential amplifier, the signal swing will be the differential output swing of the amplifier, Vout divided by the amplifiers open loop gain, Aol. In other words, Vin_diff = Vout/Aol. The error current due to the finite input impedance will then be Vin_diff/Input resistance (say 200kohm).

So say the Aol @ 100kHz is 40dB and the output signal is swinging 1Vpp, then at the input the signal is only swinging 1Vpp/100 = 10mVpp. The error current is then 10mVpp / 200kohm = 50nA. If you assume that your sensor output is swinging 100mVpp (close to the stated +/-40mV) and you assume that the output impedance of your sensor is 1kOhm, then 100mVpp/1kohm = 100uA, so your signal current is 2000x your error. If you want still lower error you can consider using a higher bandwidth amplifier like the THS4521. This will have even higher Aol which translates to smaller error current. Does this make sense? You can try simulating your circuit in TINA spice to see this effect.

-Samir

Samir,

1. The bandwidth is very low, around 20 Hz.

2. From my understanding the sensor acts purely differential. It's basically a 1 kOhm resistor that induces the voltage from within the resistor. So if you were to measure the impedance you would just measure it across the entire sensor. With that said, I'm not entirely sure what you mean by well matched since the impedance would be across the entire sensor and not just one side.

Also, since the voltage is induced within the sensor, the output is differential across the sensor with a common mode voltage of 0.0 Volts.

My one question about what you are saying is that the THS4521 has only half the input impedance (100k vs 200k of the THS4531), which suggests to me that more current would tend to enter the THS4521. This plays out in my simulation, as well, with the THS4521 drawing 2x the current vs the THS4531. I would have thought this would increase error.

Thanks,

Matt

Hello Matthew,
At such low frequencies you can just use the THS4531 which will have lots of Aol so your error is going to be low. I am now confused by how you plan to design your circuit with the sensor and the amplifier. Do you have a schematic that you can draw it out on? How much load can that sensor drive? If the sensor cannot drive much load then you will have to use two opamps in a buffer configuration and an FDA (fully differential amplifier) may not work for you.
-Samir

Samir,

As far as load requirements, do you mean how much current does it output? I'm guessing the current would be the voltage output/resistance - something in the range of 15 microAmps. I wasn't aware that I needed to worry about driving a load.

This is a picture of the circuit in TINA. It's just set up as a simple op-amp circuit with gain (1k) and feedback resistors (64.9k) and a low-pass RC filter on the output. The source is modeled as a 40 mV source with 1k impedance. The problem that I am seeing is that even though I give a 40mV signal, it only measures as 26.6mV because of the internal resistance.

Samir,

Thank you very much for your help. We are still using the THS4531 to provide a gain to the signals and provide more resolution by utilizing the full range of the ADC. A buffer amplifier is needed in front to decouple the gain resistors from the sensor's impedance since it is so variable from sensor to sensor. The OPA333 that you suggested has worked well for us, although we had to put a pull-down resistor at its input due to very low currents from the sensor causing fluctuations, similar to what is described here: electronics.stackexchange.com/.../pulldown-resistor-on-a-unity-gain-buffer