How to connect TIA with PIN diode in 10GHz communication ICs
I'm designing a TIA used in 10GHz optical communication system. Without connection with PIN diode the BW of TIA can reach 15GHz. However, I'm looking for some solution to connect TIA with PIN diode. Because of large capcitanece from PIN, direct coupling will degrade the BW stongly. Is there someone have idea to resolve this problem? Best show with a circuit topology. Thanks in advance!
You need a photodiode capable of operating at those speeds which means it won't have high capacitance.
The input impedance of the TIA needs to be very low so you get a high frequency response with the photodiode connected.
I don't know if the technique is used for high speed communications TIAs but you can bootstrap out the photodiode capacitance with specific TIA topologies.
Keith
It's assumed that the photodiode is available, which has a capacitance of 500f. My TIA has a input resistance of 100ohm. What do you mean by bootstrap out the photodiode? Could you please explain that?
By the way, it is realized in Bipolar technology with ft=100GHz.
That seems a very high input impedance - are you sure it is not the transimpedance? I would expect an input impedance of a few ohms.
Bootstrapping requires that you have an output from the transimpedance amplifier which is a buffered version of the input pin. Then, instead of connecting your photodiode cathode directly to the positive supply (for example) you connect it through a resistor and capacitively couple the cathode to your buffered output. It helps to cancel out the photodiode capacitance effect. However, I think your problem is getting the input impedance sorted first.
Keith.
The first stage of the TIA is a cascoded amplifier with a source follower and a feedback resistor of 600ohm. It followed by two stages differential amplifier as the post amplifier to get a reasonable tramsimpedance gain. Of cource, with the reduce of the feedback resistor, it is possible to enhance the BW. However the input refer noise increases at the same time. That's the dilemma.
I'm thinking about if it is possible to use a commen gate stage between the photodiode and the TIA to decouple the capacitance. But I'm afraid that the current gain will be reduced by this stage. Have you some better idea?
With a feedback resistor of 600 ohms I would expect considerably less than 100 ohms input resistance. If that is correct then it sounds like your amplifier gain is too low.
Keith.
That's true! But in order to get a BW over 10GHz, it's impossible to have a high gain. However, with the post amplifier after the first stage 75dB ohm transimpedance gain could be reach.
I am talking about the open loop gain of the amplifier, not the transimpedance with the feedback resistor. If the open loop voltage gain is low, the input impedance will be high.
Keith.