Post Go back to editing

Spurs in HMC394 and HMC705 dividers

Category: Hardware
Product Number: HMC394

Hi,

I have the eval boards of the HMC394 and the HMC705 for use in prototyping the development of a PLL/VCO. Overall, I have been able to build this prototype and could get the frequencies I need locked to the REF. But the spectrum of my signal has spurs and their harmonics.

Troubleshooting for this led to finding that the spurs are due to the prescalers HMC394 and HMC705. I used the HMC394 in a translation loop to divide by 6 the output of a mixer from 150-960MHz down to 25-150MHz. And I as well used the HMC705 to divide 6GHz down to 1GHz for use in some other part of my prototype.

By measuring the spectrum of the outputs of each of the HMC394 and HMC705, I can observe spurs at around +/-320 to 370kHz in each part. And these spurs are what appear on my PLL output at same offset frequency.

I ensured to use LDOs to power the eval boards and have ruled out power line noise contributions. I am ruling out noise addition from the loop filter also, since these spurs are measured in the 'isolated' eval boards of the prescalers/dividers. There were no spurs in the input signals to these dividers, as they were generated from a signal generator as a CW tone.

The images of the spectrum are as follows:

1. Spectrum of the HMC705:
Fin = 6GHz @5dBm, N = 6, meas_Fout = 1GHz

spur at +/-320kHz in 1GHz output of the HMC705

2. Spectrum of the HMC394: 
Fin = 900MHz @5dBm, N = 6, meas_Fout = 150MHz

spur at +/-320kHz in 150MHz output of the HMC394

 

3. Spectrum of the HMC394: 
Fin = 150MHz @10dBm, N = 6, meas_Fout = 25MHz

spur at +/-320kHz in 25MHz output of the HMC394

4. Image below shows the output of my PLL, with the effects of the spurs +/-320kHz offset from the 11.76GHz carrier.

spur at +/-320kHz in 11.76GHz output of the PLL

I look forward to more troubleshooting tips. And if this is a know issue with these parts, kindly clarify. If possible suggest solutions or alternatives.

Parents
  • Hi  ,

    I made a test for HMC705 to check the spurs.
    First, ı used Agilen E456A as a power supply and ı saw spurs around 255kHz, 92 dBc.
    Second, ı used Rohde Schwarz FSWP's DC output to supply the chip. I did not see any spurs up to 116 dBc.
    Because the spur frequency is different and it is gone when ı used a clean source, ı think your problem is supply-related. Power supply noise could be coupled or injected from supply to supply, air to board, or ground to board. It is difficult to isolate these noise sources while working with EVAL and you need to shut down other devices or use a battery when everything is off to be sure. but while these high-level spurs are not seen on these chips, ı think you should still be suspicious about the power line or GND. 



    Regards,
    Burhan

  • Thanks for the reply Burhan.

    Is this to also suggest that the use of LDOs may not be as effective as thought? 
    I will try the battery suggestions, and hoping that gives some resolution.

    regards

  • I made a quick test with an LDO but it did not suppress the spur as much as expected. The LDO could not suppress enough or the noise is bypassing the LDO, maybe from GND or air. While these chips are not enclosed possibilities are high. but with a clean source, you would not have trouble. LDOs can suppress the input noise depending on their PSRR vs frequency performance, but in this case problem might be different. I can suggest using ferrite beads as an additional precaution. But in some cases, ferrite beads also cause noise problems too. So you may try it with and without the bead to get the best result.

    Regards,
    Burhan

Reply
  • I made a quick test with an LDO but it did not suppress the spur as much as expected. The LDO could not suppress enough or the noise is bypassing the LDO, maybe from GND or air. While these chips are not enclosed possibilities are high. but with a clean source, you would not have trouble. LDOs can suppress the input noise depending on their PSRR vs frequency performance, but in this case problem might be different. I can suggest using ferrite beads as an additional precaution. But in some cases, ferrite beads also cause noise problems too. So you may try it with and without the bead to get the best result.

    Regards,
    Burhan

Children
No Data