1
\$\begingroup\$

I'm toying with the idea of building a >50MHz Bandwidth Photodiode Amplifier, built around a 1GHz AD8009ARZ Current Feedback Amplifier. I got two photodiodes that have a bandwidth exceeding 1GHz.

The problem is that the AD8009 (the only CFA available to me with a big enough Bandwidth) has gigantic input currents, on the order of tens to hundreds of uA. To circumvent this issue I plan on using a buffer with 190MHz bandwidth (made out of a Voltage Feedback Amplifier) to buffer the output of the resistive readout of the photodiode .

The final circuit I have in mind looks something like this: enter image description here

My questions are:

  1. Would biasing the non-inverting input of a CFA as shown above be a viable option, or is there a need for a different type of biasing for the high current draw CFA?

  2. Is the configuration used for the non-inverting amplifier viable for Current Feedback Amplifiers?

\$\endgroup\$
5
  • \$\begingroup\$ If you're cap-coupling it anyway, and the source impedance is low, what difference does it make? -- have you compared to the unbuffered case? In other words: you say your concern is DC bias current, but you haven't asserted proof that the stated concern is in fact a problem. \$\endgroup\$ Commented Jun 26 at 2:04
  • \$\begingroup\$ Where are the results of simulating your circuit. Surely this is the sensible next step? \$\endgroup\$
    – Andy aka
    Commented Jun 26 at 7:53
  • \$\begingroup\$ @Andyaka I would simulate it if I had the correct spice model. I've only found a pspice .cir model and I have no idea how to convert that type of file format. \$\endgroup\$ Commented Jun 26 at 8:26
  • \$\begingroup\$ @TimWilliams I am perhaps a bit confused. What I wanted was to have a DC path for the input of the CFA (hence the input voltage divider) and also have only a small "error" current. \$\endgroup\$ Commented Jun 26 at 8:40
  • \$\begingroup\$ Error from, or with respect to, what? The voltage divider is a free variable; you can set the values almost arbitrarily, and C5 ensures that variation won't be compounded by voltage gain. Perhaps another angle will do: what are you using VOUT for? AC signal analysis (e.g. AGC, filtering, mixing, etc.)? Fixed (DC) thresholds / bit slicing? Adaptive thresholds? I would imagine most end uses are immune to DC offset, at least until clipping occurs; and this is true whether dealing with it in the analog or digital (post-ADC) domain. \$\endgroup\$ Commented Jun 26 at 11:26

1 Answer 1

1
\$\begingroup\$

The problem is that the AD8009 (the only CFA available to me with a big enough Bandwidth) has gigantic input currents, on the order of tens to hundreds of uA

Suppose you DC-servo the bottom end of the 50Ω source resistor to keep the average value of the amplified signal where you want it, e.g. mid-supply. That will allow you to DC-couple the thing.

With DC coupling, the bias currents flow into/out of the 50Ω source resistance. Their drift, from all sources, is much slower than signals of interest, and thus immaterial.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.