AnsweredAssumed Answered

HMC441LP3 has no nominal broadband gain over the frequency range

Question asked by Liuccc on Jul 4, 2016
Latest reply on Jul 28, 2016 by kkaya


I made an amplifier using HMC441LP3. However, over the frequency range from 6.5 GHz to 13.5 GHz, the circuit doesn't work as it is specified in the datasheet of the chip. From 6 GHz to 10 GHz, the gain is about 10 dB. Above 10 GHz, the gain declines. Around 12 GHz, the circuit has no gain but only loss.


I met this problem with the 1st version of the circuit. The PCB layout is showed in figure 1. At first, the PCB material FR4 is considered as the reason which leads to the problem. Then, the material was changed to Rogers 4350, but the problem still existed. Figure 2 is the 2nd version of the PCB.  As illustrated in figure 2, there is sudden change in the ground plane around the signal path (circled by yellow dotted line). Then the discontinuity of the impedance in the input/output signal path, which results the reflection of the signal thus causes the signal loss, is considered.


In 3rd version of the circuit, the thickness of the board is reduced, also the width of the signal path. And the ground plane on both sides of the signal path is made continuous with no sudden change, as showed in figure 3. Considering the manufacturing cost and time, also the less improvement of the problem between 1st and 2nd version, the 3rd version of the circuit is made of FR4. However, the problem even get worse in this board. From 6 GHz to 12.5 GHz, all frequency have no gain but loss.


Figure 4 and figure 5 are the frequency sweep inputs and outputs of the circuit (version 2, figure 2). As previously described, in low frequency the circuit has gain, yet in high frequency, the circuit causes loss to the signal. The S11 parameter of the input port and the S22 parameter of the output port are illustrated in figure 6 and figure 7. Both the results show the return loss of the input and output port is not very well, but figure 6 and figure 7 hardly explain the changes of the gain of the circuit.


I'm looking forward for some advice. My e-mail address is, if convenient, please contact me with e-mail for faster response. This problem has troubled me for a long time and I hope to solve this problem as soon as possible.


Thanks a lot.


7-5 update: My mistake. The earlier S11 and S22 parameter was measured without power supply to the circuit. I have added the newly measured S11 and S22 to the attachment (figure 6 and figure 7), also the S21 of circuit version 2 (figure 8), the S21 of version 1 (figure 9) and the S21 of version 3 (figure 10).