Transimpedance amplifier with decompensated op amp

I am trying to tune a wideband transimpedance amplifier for maximum speed. I am experimenting with the OPA846 and OPA847 op amps which have large GBW but they are not unity gain stable with minimum noise gains of 7 and 12 respectively.

The minimum noise gain limits how much bandwidth I can get from the op amp with a typical transimpedance topology, and I am thinking about operating below the minimum noise gain with the following circuit:

simulate this circuit – Schematic created using CircuitLab

The current source with shunt resistance (~1Gohm) and capacitance (<1pF) is an APD in my case. The APD is one of the reasons I am able to explore such high frequency operation because it has very small shunt capacitance at <1pF.

Ignoring Rin, the noise gain has a zero at 1/(RF * CD) and a pole at 1/(RF * CF). The noise gain at high frequency is 1 + CD/CF and it needs to be greater than the op amp's minimum noise gain for stability.This is shown by the "Typical transimpedance application" curve in the bode plot below.

What are the pros/cons of adding a pole with RIN at 1/(RIN * CD) after the feedback pole 1/(RF * CF) and having a high frequency noise gain less than the recommended minimum? The idea is that the second pole in the noise gain would add stability when operating below the minimum noise gain where the op amp's open loop gain has a second pole. This is shown in the second noise gain curve in the bode plot.

The only problems I can see is that there would be some current division between RIN and RS, but it would be very small since RS is very large. Also, RIN would add Johnson noise.

Has anyone seen this anywhere or done this before?

Thanks

• I have a hunch that adding Rin will make it a lot slower because Cd * (Rs || Rin) comes into play when Rin > 0. May 7, 2017 at 16:08
• For speed you don't operate in photovoltaic mode, you operate in photoconductive. So add a bias supply and redo your schematic. May 7, 2017 at 16:41
• @WhatRoughBeast He does say he is using an APD so I assumed this is the AC circuit ignoring the biasing. May 7, 2017 at 16:54
• Ya I have a 60v reverse bias and other circuitry but I kept the schematic simple trying to be more general. May 8, 2017 at 6:49

Increasing Rin will reduce your bandwidth - the division in current is between Rin and Cd. You're right in that it will improve stability.

To get good performance I would recommend reading the book by Phil Hobbs. You can find out more at his web site. (http://electrooptical.net

The book by Jerald Graeme Photodiode Amplifiers is also very useful.

• why will my bandwidth be reduced if the 1/(RinCd) pole is at a higher frequency than my 1/(RfCf) pole? The impedance of Rin will be much smaller than Cd until after the transimpedance gain rolls off. May 8, 2017 at 23:40
• i wish that Jerald Graeme had a better editor for Photodiode Amplifiers. It doesn't match up to his books with Burr Brown. Dec 9, 2018 at 13:08