Hi,
I've designed an amplifier based on op-amps, a while ago. Overall, it works well. That I find it strange, is that the input impedance starts to deviate right away, and I've seen -15dB reflected at 200 MHz, which I find it a tad much, despite it being a broadband amplifier that is able to go up to 96 MHz.
If you look at the Smith chart above, where the input is being measured, you can see that the impedance already deviated considerable at 200 MHz. I've seen people having much better results with 0805 resistors as terminations. So, probably it is an issue with the PCB design, but I can't get why. I have controlled impedance up to the 49.9 Ω resistor. Attached, along with the schematic, you can see the calculation for that trace. I'm using the 4-layer stackup from OSH Park, on a FR408 material from Isola.
I've seen several ways to tackle this problem:
- The input termination resistor in R2, which is a 49.9 Ω resistor, gets replaced by two 100 Ω resistors in parallel. I've seen many experiments on that subject, and it was found that two resistors in parallel, besides being more accurate, have a much lower inductance than a single 0805 resistor.
- The input trace goes to the bottom of the board, so the termination can be done much closer to the concerning input of IC3. The termination resistors would be on the bottom plane, and then a via would connect to that pin of IC3.
- At the time I've used a grounded coplanar waveguide design. Maybe change that to microstrip. I'm guessing that the increased distance from the ground plane on the top layer would improve capacitance? Despite the characteristic impedance being the same with a 13.6 mils trace (recalculated for microstrip)?
What do you think it is causing this?
Kind regards, Samuel Lourenço
I've designed an amplifier based on op-amps, a while ago. Overall, it works well. That I find it strange, is that the input impedance starts to deviate right away, and I've seen -15dB reflected at 200 MHz, which I find it a tad much, despite it being a broadband amplifier that is able to go up to 96 MHz.
If you look at the Smith chart above, where the input is being measured, you can see that the impedance already deviated considerable at 200 MHz. I've seen people having much better results with 0805 resistors as terminations. So, probably it is an issue with the PCB design, but I can't get why. I have controlled impedance up to the 49.9 Ω resistor. Attached, along with the schematic, you can see the calculation for that trace. I'm using the 4-layer stackup from OSH Park, on a FR408 material from Isola.
I've seen several ways to tackle this problem:
- The input termination resistor in R2, which is a 49.9 Ω resistor, gets replaced by two 100 Ω resistors in parallel. I've seen many experiments on that subject, and it was found that two resistors in parallel, besides being more accurate, have a much lower inductance than a single 0805 resistor.
- The input trace goes to the bottom of the board, so the termination can be done much closer to the concerning input of IC3. The termination resistors would be on the bottom plane, and then a via would connect to that pin of IC3.
- At the time I've used a grounded coplanar waveguide design. Maybe change that to microstrip. I'm guessing that the increased distance from the ground plane on the top layer would improve capacitance? Despite the characteristic impedance being the same with a 13.6 mils trace (recalculated for microstrip)?
What do you think it is causing this?
Kind regards, Samuel Lourenço
Last edited:
