Gain peaking in TIA

Dear TUKA.

TUKA said:

This TIA has a significant gain peaking (close to 5 dB) in its frequency response characteristics, which starts around 30 GHz and is there upto 60 GHz.

I would prefer to have a flat response for the TIA using a suitable block or an enhancement technique after the TIA (and before the VGA).

I would need some insights about how to proceed further and make the TIA response flat, without affecting its bandwidth much.

Writing for myself only, if you are trying to maintain signal integrity at frequencies between 30 GHz and 60 GHz, I would prefer to better understand the source of the gain peaking and re-optimize the photodiode/amplifier parameters of your TIA to minimize the gain peak to a more reasonable value. Typically, the peaking is often a result of the feedback resistance and photodiode capacitance and both may be adjusted to minimize peaking. Adding an equalizer, which would inevitably be some type of lowpass or notch filter, will reduce bandwidth, add phase (makes stability more of an issue), and is only effective if the equalization tracks the magnitude and phase characteristics of the peaking. In other words, if your peak frequency or its peak magnitude varies with process or temperature, then your equalizer must track these variations - which is very difficult. 

I am making some assumptions about your exact topology and don't know your components nor your design constraints  - so my comments may not be very helpful.

Shawn

Dear Shawn,

Thanks for getting back to me with your helpful comments.

ShawnLogan said:

I am making some assumptions about your exact topology and don't know your components nor your design constraints

Yes, your assumption about my topology is absolutely correct. I am using a shunt feedback topology for the TIA. I am also attaching the TIA's response for better understanding.

I have a high transimpedance gain requirement (above 65 dB ohm) and a bandwidth above 60 GHz. So as you suggested, I would certainly consider tuning the amplifier parameters. Maybe reducing the feedback resistance is an option, but that would inevitably reduce the TIA gain. But eventually, I believe I can recover back this gain using a VGA, but at the expense of a reduced bandwidth (due to multi-stage cascading effect).

ShawnLogan said:

Adding an equalizer, which would inevitably be some type of lowpass or notch filter, will reduce bandwidth

Yes, you are correct. I tried adding a CTLE (Continuous time linear equalizer) after the TIA, but it reduced the bandwidth significantly.

ShawnLogan said:

if your peak frequency or its peak magnitude varies with process or temperature, then your equalizer must track these variations

Fortunately NOT. Process and temperature variations are not affecting the TIA`s frequency response in my case.

Thanks a lot for your helpful feedback once again!

Regards,

Tuka

Dear TUKA,

TUKA said:

rocess and temperature variations are not affecting the TIA`s frequency response in my case.

Well - that is good to read!

TUKA said:

Maybe reducing the feedback resistance is an option, but that would inevitably reduce the TIA gain. But eventually, I believe I can recover back this gain using a VGA, but at the expense of a reduced bandwidth (due to multi-stage cascading effect).

If you re-optimize your TIA feedback resistance to reduce some of the peaking and do need to increase the gain of your VGA (reducing its bandwidth), perhaps you might be able to include some inductive peaking in your VGA to extend its bandwidth a bit if you need to increase its gain. As an example, in reference [1], a capacitor is used to generate an active inductor that produces a resonance in the GHz range. This might prevent the need to add an additional VGA stage. However, this may be feasible for the frequency range you are targeting.

Shawn

Reference [1]:

H. D. Lee, K. A. Lee and S. Hong, "A Wideband CMOS Variable Gain Amplifier With an Exponential Gain Control," in IEEE Transactions on Microwave Theory and Techniques, vol. 55, no. 6, pp. 1363-1373, June 2007, doi: 10.1109/TMTT.2007.896787.

Dear TUKA,

TUKA said:

rocess and temperature variations are not affecting the TIA`s frequency response in my case.

Well - that is good to read!

TUKA said:

Maybe reducing the feedback resistance is an option, but that would inevitably reduce the TIA gain. But eventually, I believe I can recover back this gain using a VGA, but at the expense of a reduced bandwidth (due to multi-stage cascading effect).

If you re-optimize your TIA feedback resistance to reduce some of the peaking and do need to increase the gain of your VGA (reducing its bandwidth), perhaps you might be able to include some inductive peaking in your VGA to extend its bandwidth a bit if you need to increase its gain. As an example, in reference [1], a capacitor is used to generate an active inductor that produces a resonance in the GHz range. This might prevent the need to add an additional VGA stage. However, this may be feasible for the frequency range you are targeting.

Shawn

Reference [1]:

H. D. Lee, K. A. Lee and S. Hong, "A Wideband CMOS Variable Gain Amplifier With an Exponential Gain Control," in IEEE Transactions on Microwave Theory and Techniques, vol. 55, no. 6, pp. 1363-1373, June 2007, doi: 10.1109/TMTT.2007.896787.

Dear Shawn,

Thank you for your reply, and thanks for sharing the reference article.

Regards,

Tuka