HMC799 photodetector oscillation at 1.6GHz?

With the attached schematic and PCB I get an expected photodiode-response from around 0 to 400 MHz. However with the detector dark (or illuminated) the HMC799 seems to self-oscillate at around 1.6 GHz with a measured power of -27 dBm (attenuated perhaps 3-6 dB by the BUF602 I am using).

Any suggestions that I could try with this PCB, or design changes that I could make to the next version of this PCB?

  Trace A is the SA floor. Trace B is the detector powered but not illuminated. Trace C is the optical response from a modulated VCSEL light source, showing 400 MHz optical bandwidth. The photodiode is either FDS015 or FDS025 from Throlabs. Changing the photodiode changes the shape of the optical response slightly but not the 1.6 GHz oscillation.

 This is the schematic I am using. Are there better ways of applying a DC reverse bias (to reduce the photodiode capacitance) while AC-coupling the signal to the HMC799 input?

  2-sided PCB top side. There are 9 vias on the EP of the HMC799. 

 PCB bottom side.

Top Replies

Parents
  • 0
    •  Analog Employees 
    on Jul 7, 2020 12:37 PM

    Hi,

    what is the photodiode capacitance you are using? How much does it change with respect to PD's reverse bias?

    Also, could you please add a minor loss(shunt or series resistor) at the input or output to see if the oscillation is suppressed?

    Could you measure the supply current?

    Your reverse bias circuit seems allright, maybe changing the bias resistor(10k) to introduce more loss to the system to kill the oscillation could be tried.

    best regards

    Sinan

  • Hello, thanks for your suggestions.

    The FDS015 photodiode is specified as 0.65 pF at 5V reverse bias.

    The other photodiode I tried is FDS025 specified as 0.94 pF at 5V reverse bias.

    I now tried resistor in series (on the photodiode side) with the 22 nF input capacitor C112.

    This removes the 1.5-1.6GHz oscillation but instead results in a very strong -2 dBm oscillation (and harmonics) around 330 MHz. I tried both a 49R and a 499R resistor, 0402 sized. With the 499R resistor the frequency is slightly higher around 340 MHz.

    The first image is with  49R + 22nF at the RF-input, the second image is with 499R+22nF at the input. 

    I did not yet try to modify the output of the HMC799 (with e.g. a series resisotor) - would you suggest that this is a useful test also?

    The supply current in these tests was around 80mA @ 8V from the powersupply, with the LDO producing 5V on the PCB. This includes the BUF602 I have driving the output cable.

  • 0
    •  Analog Employees 
    on Jul 16, 2020 4:52 AM in reply to awallin

    Hi,

    Looking closely at the layout, I noticed the ground path for the supply capacitors are very long. The ground via at that node is very far from the exposed paddle. This extra inductance is likely to cause the oscillation. Ideally, there should be a via under the GND path of the capacitor or very close to it.

    Best regards

    Sinan

Reply
  • 0
    •  Analog Employees 
    on Jul 16, 2020 4:52 AM in reply to awallin

    Hi,

    Looking closely at the layout, I noticed the ground path for the supply capacitors are very long. The ground via at that node is very far from the exposed paddle. This extra inductance is likely to cause the oscillation. Ideally, there should be a via under the GND path of the capacitor or very close to it.

    Best regards

    Sinan

Children
No Data