AnsweredAssumed Answered

HMC799 TIA oscillation (again!)

Question asked by KenCrocker on Apr 9, 2018
Latest reply on Apr 12, 2018 by salemdar

I'm re-posting this from HMC799 TIA oscillation upon request from ADI support so they can response directly. I'm still waiting for a response from 'salemdar'. We have collected a lot more data in the meantime, but the issue still remains.

Thanks for your consideration!

--Ken Crocker


I have a PCA that has three photodiode amplifiers using HMC799's. All three outputs oscillate at ~900 MHz with no input. The PCA is in a 6-sided grounded metal box to shield if from stray RF signals.

Schematic of Photodiode Transimpedance Amplifier

PCB Layout for Photodiode TIA

I'm using five vias to connect the central pad of the HMC799 to GND. They use 10 mil holes. The next layer in the PCB is a solid GND plane. To prevent solder wicking, the PCB was fabricated using VIPPO, so the holes have been filled with non-conductive epoxy, planarized, and then plated with copper. I can make the oscillations come and go -- they are apparently related to the capacitance seen on the input to the HMC799. Here are some experiments:

  1. With 15nF and 1.00K installed, PD (photodiode TO-66 socket D3) open => oscillation
  2. Making top cut (see JPG) isolating trace from 1.00K to D3.3 => no oscillation
  3. Making bottom cut (see JPG), then connecting 1.00K to D3.3 with flying AWG #30 lead => oscillation
  4. Disconnecting flying lead from D3.3 => no oscillation
  5. Connecting flying 1N4148 diode between D3.4 (cathode) and 1.00K (anode) => oscillation
  6. Disconnecting flying 1N4148 from D3.4 => no oscillation

The same circuit built with the evaluation PCB does not exhibit this issue. We've been able to go up to about 50 pF without oscillation using the evaluation PCB. At this point, I suspect that there is a difference between the production lot of HMC799's and the one on the evaluation PCB. I'd guess that, somehow, this lot of HMC799's has a much lower input capacitance tolerance. BTW, the larger capacitor packages are required because this design is based on a flight qualified QPL. I replaced the 10 uF tantalum with a 1210 ceramic without effect.


Thoughts? Ideas? Thank you for your consideration!!