OPA657: Used the OPA657 as a transimpedance amplifier, but the TIA output is unstable and oscillates severely.

Hi Dat,

I am looking into it.

Can you please let me know what is the expected photodiode capacitance?

Regards,
Aditya Gosavi

Hello Aditya,

Thank you very much for your response.

Currently, I am using a Keithley 2450 Source Meter to supply current through a BNC connector. Keithley does not specify the output capacitance, but I estimate it to be less than 10000 pF.

With the OPA657 having a GBP of 1.6 GHz, and assuming Cin < 10000 pF, I selected Cf = 0.25 pF and Cpcb = 0.1 pF. The bandwidth is 30 kHz, which is still below the maximum allowed bandwidth of 40 kHz.

I have attached an image of the output voltage when the input current is 100 nA. Although the output voltage is unstable and oscillates severely but the mean voltage remains accurate.

I don’t know the exact cause of the oscillation. I have tried gradually increasing Cf, and it reduces the oscillation amplitude, but at the same time, it also lowers the bandwidth, which does not meet my requirements.

Best regards,
Dat

Hi Dat,

Thank you for providing all the details, will look into it and get back to you.

Please try our TIA circuit design tool: Analog circuit design studio where you can provide all your inputs like: Cpd, TIA gain, op-amp selection, etc. The tool also has "suggest TIA" mode. The tool will provide recommended Cf, TIA bandwidth, noise and SNR information, etc.

Regards,
Aditya Gosavi

Hi Aditya,

I have already used that tool, but it does not support the OPA657. However, I tried calculating with the OPA818 ( Performance upgrade to OPA657), and I found that the tool gives results similar to when I calculated using the transimpedance calculator file provided by TI.

Previously, I used the AD4817-1 from ADI, which has a GBP of 410 MHz. When measuring at the TIA output, I observed oscillations, so I suspected that the issue was related to the GBP. I expected that replacing it with the OPA657, which has a GBP of 1.6 GHz, would solve the problem. However, after the replacement, the TIA output voltage oscillated with an even higher amplitude than before.

This is the GBP of AD4817-1:





I look forward to your feedback on this issue.

Regards,
Dat

Hey Dat,

To accurately assess the transimpedance amplifier (TIA) stability, I recommend performing a check without the source measure unit (SMU) connected. The SMU's output capacitance may significantly impact the TIA's bandwidth. With a 10nF cap at input, the TINA TI simulation results in a TIA bandwidth of 4-5kHz for OPA657, and 17kHz for OPA817, which is very less as per your needs.

Please connect the photodiode (or an equivalent capacitor representing the photodiode's junction capacitance) directly to the TIA input and apply the necessary bias. Observing the TIA output for oscillations under this configuration will confirm stability, independent of an input signal.

Furthermore, a 10nF input capacitance presents a significant challenge to achieving a bandwidth exceeding 40kHz. To provide tailored recommendations for circuit modifications, please provide the expected photodiode capacitance. This parameter is critical for determining both the TIA's bandwidth and stability.

Regards,
Aditya Gosavi

Hi Aditya,

Thanks for getting back!

Could you send me the TINA-TI simulation files for the OPA657 and OPA817? I’m surprised because the OPA657 has a higher GBP than the OPA817, but the OPA817 gives a larger TIA bandwidth. Did I misunderstand something? Can you explain it more clearly?

Best regards,
Dat

Hi Dat,

The observed disparity in TIA bandwidth between the OPA657 and OPA817, despite the OPA657's higher gain-bandwidth product, can be attributed to the interaction of the substantial 10nF input capacitance with the feedback network. This large input capacitance introduces an early zero in the inverse feedback factor (1/β), also known as the noise gain (1 + Zf/Zin). Consequently, the noise gain curve intersects the op-amp's open-loop gain (Aol) curve at relatively low frequencies.

Specifically, in the case of the OPA657, this intersection occurs around 5 kHz, whereas for the OPA817, it occurs around 17 kHz. This difference arises because the closed-loop bandwidth of a TIA, is defined by the frequency at which the noise gain curve intersects the Aol curve (For butterworth response or Q-factor of 0.707).

Furthermore, the OPA817 exhibits a significantly higher open-loop DC gain (approximately 90 dB) compared to the OPA657 (approximately 70 dB). This higher DC gain in the OPA817 effectively shifts the intersection point of the noise gain and Aol curves to a higher frequency, resulting in a wider closed-loop bandwidth.

This phenomenon is readily observable in the provided TINA-TI simulation schematics.

For bandwidth: opa817/opa657 tia.TSC

For Aol and one-by-beta analysis: opa817/opa657 tia only stability.TSC

Regards,
Aditya Gosavi

opa657 tia.TSC

opa817 tia.TSC

opa657 tia only stability.TSC

opa817 tia only stability.TSC