Post Go back to editing

HMC369 and HMC561 Phase noise issue

Dear Forum

For a reference frequency multiplier from 5 GHz to 10 GHz, I planned on using the HMC561 first. After measuring the phase noise before and after the doubler, I noticed a very pronounced increase in phase noise between 200 kHz and 20 MHz due to the device. So after a quick search in your frequency multiplier database, I came up with the HMC369, whose limited frequency range and easier use is more suitable for our purpose. But, unfortunately, the same phase noise pronunciation is observable with this device, which leads me to the question: Am I doing something wrong or is this to be expected from these devices (despite the datasheet information)?

Both frequency doublers were evaluated using their respective evaluation board and the feeding 5 GHz is absolutely OK. The input amplitudes have been checked multiple times, as was the gate voltage and sequencing in case of the HMC561.

In the attached picture, the light blue curve is the input signal phase noise (@5 GHz), the green curve is the HMC561 output phase noise, purple is the HMC369 phase noise and the red curve is the output phase noise of our Anapico APSIN20G signal generator at 10 GHz (just for comparison!). Encircled is the problematic phase noise region which in an ideal world should follow the light blue curve with an offset of +6 dBs.

Any suggestions?

Thanks in advance and best regards,


  • Hi Fabian, 

    What you are seeing are the effects of the residual or additive phase noise of the multiplier but admittedly the degradation is much worse than what I'd expect.  Residual phase noise is shown in the datasheet spec table and is -139 dBc/Hz for the HMC561 and -142 dBc / Hz for the HMC369. This noise adds linearly as opposed to 20 * Log(N) due to the multiplication. I would have expected the plots for the HMC561 & HMC369 to be reversed and to be roughly 10 dB better than shown as there will be a 6dB increase in the absolute phase noise due to multiplication then a db or two due to the additive noise. Is the input signal modulated? Perhaps there is noise folding back?

    Best regards,


  • Hello Fabian,

    I would say not only green and purple curves are issued but also the blue curve. I don't know the PLL architecture of 5GHz signal source, but the theory for VCO says, that Phase Noise roll off factor is 20dB/decade, which is not fulfilled in the range 20kHz...200kHz. I assume here, that PLL BW<<20kHz. 

    Nevertheless it is not an answer to your question. You want to know strange behavior of spectral mask multiplied signal - green, purple. I have seen similar phenomenon in my designs. The root cause were noise from power supply (in your case power supply for x2 multiplier) OR high content of harmonics/subharmonics in the measured RF spectrum OR strong external interference near RF signal. The second note can be measurement device specific, I used FSUP, so the sensitivity for higher harmonics will be different over measurement instruments. The third note applies mainly to VCO and it is not exactly your case.

    Regards, Viktor

  • Hi Marty

    Thanks for your answer and sorry for my late response. I've been busy with another project and I had to set the phase noise issue aside for this time.

    Exactly, I was prepared to see the phase noise degrading due to the additive phase noise term of the doubler, but not to the extent showed in the measurements. The input signal to the doubler is purely CW, no modulation applied. What do you mean by noise folding back? Folding back from where?

    By the way: the 5 GHz input signal to the HMC561/369 itself is generated from an HMC1096 active doubler. Despite it having an additive phase noise term as well, it is working as per datasheet. Both doublers are powered from the same power supply, isolated through heavy filtering and decoupling to prevent crosstalk through supply lines.

    Thanks and best regards,

  • Thanks Viktor

    As I stated in my answer to Marty, the active multiplier generating the 5 GHz from a 2.5 GHz phase-locked VCSO was run off the same supply voltage as the doubler from 5 to 10 GHz. This happened without the observed phase noise degradation on the 5 GHz signal for the first doubler, so I think I can rule out supply issues. What I can NOT rule out though, is the phase noise analyzer. I'll see if I can get an analyzer from a different manufacturer and repeat the measurements.