RF Photodiode Amplifier

[Revky27]

Hi,

I have a fast photodiode and trying to amplify it with a simple gain block circuit in order to have wide bandwidth up to 5GHz. (Think ERA-1 type gain block). I am using reverse bias and there is enough signal from a 50Ohm termination on the amplifier.

The issue is the amplifier is oscillating when I am using the photodiode as reverse biased input and there are some reflections/ringing giving me a very unclean output.

Does anyone have experience with similar high speed photodiode amplifier? Did you manage to match the photodiode to the amplifier and are there any other tricks?

Thanks

[RFDx]

Diode specs? Schematic?

[TimFox]

If the connection/cable from the diode to the amplifier's input termination is too long, it may be necessary to double-terminate, i.e. add a second termination resistor close to the diode.
This would reduce the signal level by 1/2.

[Revky27]

The diode specs are as this:

https://www.fermionics.com/files/FD50series.pdf

The schematic is simple. I reverse bias the diode via a 1k series resistor on the cathode and the anode is tied to a 50Ohm resistor at the input of the ERA-1+ amplifier. The connection is a 50Z 20mil CPWG.

But I notice some ringing after the initial pulse and reflections.

[Revky27]

The distance from diode to amplifier is <20 mm of 50z CPWG

[EggertEnjoyer123]

The ERA-1+ is unconditionally stable. This means it's stable over all input/output impedances, and reflections from the source/load can't cause it to oscillate.

Do you have a capacitor at the cathode to short any RF to ground? If you don't then you basically have a voltage divider with the 1k and 50 ohm, and most of the voltage swing will be across the 1k ohm. Shorting the cathode to ground at RF using a capacitor would fix this issue, while allowing you to still bias the diode.

[Revky27]

The ERA-1+ is unconditionally stable. This means it's stable over all input/output impedances, and reflections from the source/load can't cause it to oscillate.

Do you have a capacitor at the cathode to short any RF to ground? If you don't then you basically have a voltage divider with the 1k and 50 ohm, and most of the voltage swing will be across the 1k ohm. Shorting the cathode to ground at RF using a capacitor would fix this issue, while allowing you to still bias the diode.

Thanks, yes, I do have capacitor at cathode, so it forms a LPF reverse biasing the diode.

[mike449]

Isn't the input impedance of the amplifier already close to 50 Ohm? Then the extra termination 50 Ohm resistor is bringing the total down to 25 Ohm. Shouldn't cause the oscillation, but it reduces the gain for sure and causes some reflections.

[Kleinstein]

The ERA-1+ is unconditionally stable. This means it's stable over all input/output impedances, and reflections from the source/load can't cause it to oscillate.

The amplifier shpould not oscillate, but a poorly terminated source can still make it ring.

With the rather short distance between the diode and amplifier reflections are at really high frequencies (e.g. 5 GHz range). If there are really reflections with a significant delay, they may be on the output side, e.g. from probing the signal. What is the frequency in the ringing part ?

The extra 50 ohms to ground should not be necessary, but they still bring the amplifier closer to a 50 ohm source impedance, at least for most of the frequency range. One could still consider a little more than 50 ohms as at the high frequencies the small diode capacitance also adds a little. Even though the distance is small, the resistor should be more to the photo-diode side, not at the amplifier input.

At the very high frequencies the ground layout can have quite some effect. A schematics may not show all and if possible a picture of the actual circuit / layout would  help more.

[iMo]

Those mmics are sensitive on the pcb design (like the layout, grounding vias position and their number) and the components used at the output (usually a choke in series with resistor, some capacitors around).
So a detailed schematics and a picture of the pcb would help..

[LM21]

PHILIP C. D. HOBBS has written about photodiode front ends. Like https://electrooptical.net/static/oldsite/talks/GettingPDRight11.pdf  or https://zeus.phys.uconn.edu/halld/hpd-3-2005/frontends.pdf