Post Go back to editing

HMC778 RF output power too low

Refer to HMC778 evaluation board, I have made a PLL board and its schematic is shown below. The RF output power of hmc778 should be 9dbm@10GHz. And I want to get 9-5+14=18dbm output power through a attenuation network and PA hmc441.

The problem is the output power of hmc778 only reaches 2dbm@10GHz. So when I use a 100pF capacitor to replace the attenuation network, total output power reaches 16dbm@10GHz. Do you know any possible reasons that could cause the decrease of output power?

The RFOUT signal is correctly locked with REFRF signal. 

  • Hi there,


    Sorry for the delay in getting back to you. I've attached our evaluation board schematic which may be helpful in understanding what we did for bypassing and various component values.


    Confirmation of expected result: The output return loss of the VCO inside the HMC778 is nominally 3 - 4dB at 10.0 GHz; even so it still generates a respectable +9dBm or so of output power at Vtune of 5.0V (~10.0GHz). The 50 ohm, 5dB attenuator should minimize any frequency pulling due to mismatch but will drop the drive level into the HMC441 to +4dBm nominally (if I'm interpreting your question correctly it sounds like you see +2 dBm).The input return loss of the HMC441 at 10.0GHz is 15dB so this isn't a concern. With +15dB of gain at 10.0GHz the +4dBm drive level is enough to push the HMC441 about 2 dB into compression so the output power should be roughly +18 - 19dBm (HMC441 DS shows a nominal +18 dBm for 1dB compression). This assumes that the HMC778 is phase locked.


    • 1)   Insufficient grounding can degrade output power at microwave frequencies. Be sure that the proper footprint was used as well as thru vias in the pad for the paddle. This improves the thermal performance as the output power will drop if the part is heating up. Also be sure all bypass caps are as close to the pins as possible and that the values are adequate.


    • 2)   Although the datasheet shows a typical value of +9 dBm or so at about 5.0V, due to process variation the output power may be higher of lower than this. I would expect ±2dB wouldn’t be unreasonable. As a side note, although the datasheet shows output power vs Vtune this is more for convenience; the internal VCO buffer amplifier only cares about match, frequency and grounding.


    • 3)   Likewise, the HMC441 datasheet is comprised of typical values as well so there will be some variation in Gain, OP1dB, etc. over process.


    • 4)   We use a 100pF, 0402, COG capacitor on our evaluation board. This results in about 0.5dB of attenuation at 10.0 GHz which is confirmed via Murata's SimSurf tool. The HMC441 DS shows that the input to the HMC441 is DC blocked so there shouldn't be a need to have an additional capacitor between the attenuator and the HMC441 but you might try this to see if it makes a difference. The ATC550L broadband caps which have less insertion loss may be a better option than the standard 100pF capacitor.


    • 5)   You also might investigate your inter-stage RF performance. Perhaps the particular RLC’s and the associated copper and vias are resulting in more than the expected insertion loss or a resonance that is impacting the power at the input of the amplifier.


    • 6)   You may be able to tweak the gate voltage and / or Vdd on the HMC441 (see DS) to improve the gain or Psat to make up for the lower drive level. Admittedly this isn’t the best solution for high volume but may help with our initial evaluation or prototypes.

    Best Regards,