This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

OPA2353: AC coupled multi stage amplifier for current sensing: how to achieve higher ac gain

Part Number: OPA2353
Other Parts Discussed in Thread: REF2030, LMP7721

Hi, I'm designing a current sensor for the photodiode.

The sensor input is a maximum of 20fA.

The target gain of the circuit is 260dB and the target bandwidth is 3kHz.

Because the application uses a 3.7V LiPo battery, I used a rail-to-rail OP amp(OPA2353).

I used a trans-impedance amplifier with a gain of 140dB and a three-stage AC coupled voltage amplifier with a gain of 60dB, 40dB, and 20dB

The schematic is shown below. 

When I performed a transient simulation, the result showed a very noisy signal. 

I changed the last stage gain to 10dB, then the result showed a very clean signal.

Here are my questions.

1. Why did this happen?

2. How can I fix it?

3. Can I replace the voltage amplifier stages with a better circuit design?

4. Can I use a voltage reference (like REF2030) to drive the four operational amplifiers? (The four amplifier need 1.5V bias at + input)

Thank you.

  • Hello 

    1) This also occurred in TI-Tina so I suspect something odd in the device model. 

    2) I could not make the result consistent 

    3) 20fA is really small, the 1st stage should have very low IIB and input current noise. OPA2353 is not the best here. The second stage should have very low input voltage noise. 

    The noisy version is more believable. I would expect a lot of noise on the final output .

  • I have been working on this and the LMP7721 as the first stage should be lower noise. However the total noise sim for OPA2353 came out less.  Be sure to add a capacitor across R8 to limit the bandwidth as that will lower noise. 

  • Hi Yunsik,

    I would do some things differently Relaxed

    1. The OPA2353 is a 600R OPAmp. It's no good idea to short-circuit the output with a 100R load.

    2. There's no need to concetrate most of the gain onto only one OPAmp. It's better to evenly distribute the total gain onto the three OPAmps.

    3. With such a high gain and without any bandwidth limiting in each OPAmp stage your setup will oscillate like hell. Been there, done that Relaxed

    I heavily recommend to add low pass filtering in each stage.

    4. The simulator's "analysis parameters" need to be optimized for such a high gain. Set the "TR maximum time step" to 1µ.

    With these modifications everything is fine:

    yunsik_opa2353.TSC

    Kai

  •  Hi, kai

    I have few questions about the circuit,

    1. Should I change the amplifier in the first stage to a low-noise amplifier for noise reduction?

    2. Since it is applied to an application that uses batteries, I would like to make a 1.5 volt bias using a voltage reference.

    Can the voltage reference drive all 4 amplifiers? I would use a REF2030.

    3. For 6 kHz sampling, is the performance of delta sigma adc better than sar?

    When simulating with SAR adc, I realized that the kickback current in the circuit could cause problems in the circuit.

    Thank you.

  • Hi Yunsik,

    Ron is right, the OPA2353 is not very low noise at the lowest frequencies. The LMP7721 would give a noise improvement of about 2dB.

    Keep in mind that U2 has a noise gain of up to 300p/4p7=64. Any noise given to the +input of U2 would be amplified by this noise gain. The noise of REF2030 would be about 10µV. This about half the noise of the U2 circuit produces itself. Multiplied with the noise gain would result in a desaster Scream

    So, in any case, low pass filtering of REF2030 output voltage would be needed.

    Or you take a 10k+10k voltage divider with a 47µF filter cap? Would be cheaper and provide much less noise.

    Referring your question on the ADC, it depends on the ADC. There is no simple answer. But the charge kick back of the sample stage of SAR-ADC should be easy to handle.

    Kai

  • This is the noise gain of first stage:

    Before you ask Relaxed

    Kai

  • Yunsik,

    Very important to remember is the expected input signal of just 20fA AC. This is very small signal so I suggest reading the three part article starting here. https://www.edn.com/design-femtoampere-circuits-with-low-leakage-part-one/

    Because the signal is AC and AC coupled, the actual op amp input current isn't very important. The op amp input current noise can be a concern. The input current noise from the op amp or extra forces gets the same gain as the desired signal.

    Due to the very high system gain, layout will matter to prevent any unwanted feedback path. Consider adding more supply voltage filtering to the first gain stage (TIA). 

    LMP7721 has a unique pinout to help with low input current designs. The LMP voltage noise is less than OPA2353. The LMP current noise 0.01pA looks higher than the OPA's 4fA, but I consider that just a rounding error. It seems counter initiative that a much lower IIB device would have a higher current noise.

  • Hi Yunsik,

    does it work now?

    If the sensor current is actually 20fA maximum, your signal might be buried in noise:

    20fA x 10M = 200nV

    compared to the noise of 10M resistor

    SQRT(4 x k x T x R x B) = SQRT(4 x 1.38 x 10^-23 x 300 x 10M x 3.4k) x V = 23.7µVrms

    Could you use a lock-in amplifier with synchonous demodulation?

    Kai