OPA855: How to decide RF, CF

Hi Yoshihito,

you need to perform a phase stability analysis, like shown here, for instance:

https://e2e.ti.com/support/amplifiers-group/amplifiers/f/amplifiers-forum/1064558/opa855-output-oscillates

Kai

The simple answer is that the feedback pole can be used to tune the Q of the 2nd order response. This is approx., but the 2nd order Fo is set by the 1/(RfCs) zero in the noise gan and the device GBP - their geometric mean is the Fo, then, for instance, if you set the feedback pole at 0.707 of Fo you will get a Butterworth closed loop - approx. 

Kai san

thank you for your reply.
I confirm the content of your contact.
By the way, can the same content be referred to the following materials?

www.tij.co.jp/.../jaja098.pdf

yoshihito

Michael san

thank you for your reply.

I still don't understand because of lack of study.

As related materials,

Can the following be used?

www.tij.co.jp/.../jaja098.pdf

yoshihito

Hi Yoshihito,

yes, these formulas can be used as a starting point. But for the fine tuning I carry out a TINA-TI simulation.

Can you tell more about your application? What is your detector capacitance? What is your transimpedance (gain)? What is your -3dB bandwidth?

Kai

Kai san

thank you for your reply.

> yes, these formulas can be used as a starting point. But for the fine tuning I carry out a TINA-TI simulation.

⇒ I have installed TINA-TI.

> Can you tell more about your application? What is your detector capacitance? What is your transimpedance (gain)? What is your -3dB bandwidth?

⇒
About the application:
・ Data receiver
・ APD is used for the light receiving element
・ APD capacitance is 2pF
-The data to be received is DDR data of about 300 to 625 MHz.
-The ideal gain is the maximum that can meet the above bandwidth.

Due to specifications,
-The input current from the APD changes.
･ The received frequency changes.

It should be amplified with a bandwidth of up to 625MHz.
At this time, I recognize that the maximum amplification that can be obtained is about 12 times.
Correct?

takahashi

Hi Yoshihito,

have you thought about using an optical networking IC?

https://www.ti.com/interface/optical-networking-ics/products.html#p1241=Transimpedance%20and%20limiting%20amplifier

Kai

Kai san

thank you for your reply.

Of course there is.
I investigated ICs for receiving optical fibers, including those made by other companies.

The most ideal IC is below.

www.ti.com/.../ONET8551T

But I don't have the package I want.
The optical fiber receiving circuit
Due to miniaturization, it is likely that it will be mounted on a wafer on an SFP module or APD body.

However, the products we develop are niche.

Therefore, I decided to build a combination of OPA855 and ONET4201PA.

Regarding OPA855, the recommended application is different, but I confirmed with another person in this forum that it can also support the application I want.

That is how we chose this IC.

Takahashi

well with that source C, and targeting a little over 1GHz closed loop, 316ohm is the Rf I found, 

And then putting in a 625MHz +/-0.316mA input, I get this - kind of sine wavy as this is  a pretty fast signal - if you need more gain follow this stage with another OPA855 as a gain block, 

OPA855 1.8GHz BW Zt.TSC

Michael san

thank you for your reply.
> well with that source C, and targeting a little over 1GHz closed loop, 316ohm is the Rf I found,

⇒The tool that carried out this simulation is
Is it TINA-TI?

The circuit of the prototype I prepared is as follows.

The connection is different from the simulation circuit,
Do you have to worry about finding Rf and Cf?

As a way to determine 316Ω
Rf that can maintain a gain of about 50 dB in the band of about 1 GHz,
Did you get it by inputting while checking the simulation waveform?

About Cf = 670fF
How did you calculate this value?
It's a little different from my calculated value.

If Rf = 316Ω,
From the formula in the following data sheet p2

jaja098.pdf (tij.co.jp)

Cf = √ 2pF / π × 8GHz × 316Ω = 502 fF

> And then putting in a 625MHz +/- 0.316mA input, I get this --kind of sine wavy as this is a pretty fast signal --if you need more gain follow this stage with another OPA855 as a gain block,

⇒ If there is an input of 0.316mA or more, this quality can be maintained.
Can you recognize it?

The output signal of OPA855 is input to ONET4201PA as a differential CML.
It has a circuit configuration to output.

Hello Takahashi-san,

  Thanks for looking for source material on TIAs (transimpedance amplifiers)!

  This quick step-by-step blog is very helpful in these calculations:

1. Part 1: What op amp bandwidth do I need (Transimpedance Amplifiers)
2. Part 2: What op amp bandwidth do I need (Transimpedance Amplifiers)
3. Part 3: What op amp bandwidth do I need (Transimpedance Amplifiers)

  Also, we have a TIA calculator available located at this thread that will calculate the feedback capacitance and bandwidth or gain bandwidth of the amplifier at a Butterworth response Q= 0.7 which will give you a maximally flat passband in the frequency response with a phase margin of 65 degrees. 

  There are different estimations around for Cf, but the calculator above should guarantee stability of the amplifier. If more bandwidth is needed, then you can decrease Cf which will cause some peaking in the frequency response, which will decrease phase margin, but increases closed-loop bandwidth. 

  Michael used Tina-TI above. Using Michael's circuit, you can increase the gain up to the bandwidth you would need of 625MHz is that correct? But, yes you are correct that you would need a second gain stage. What is the input current range and output voltage range the design would need? 

Thank you,
Sima 

Takahashi-san,

  Thank you for the update! 

Best Regards,

Sima