• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 1 answer
  • Subscribers 21 subscribers
  • Views 687 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LMX2592: Divide by Three - Harmonic content

Sid Balasubramanian62
Prodigy 30 points

Part Number: LMX2592

Folks,

What is the best way to get a good quality 2fvco and 2fvco/3 from the LMX2592?

I am seeing plenty of mixing spurs which is a strong function of whether fvco is integral or fractional, implying one of the frequency outputs is getting compromised!

Thanks!

  • 0
    TI__Mastermind 28031 points
    Sid,

    When you divide by 3, you do NOT get 50% duty cycle, I think it's closer to 1/3rd. That's the way it is.
    This leads to higher 2nd harmonic and this needs to be filtered.

    Regards,
    Dean
  • 0 in reply to Dean Banerjee
    Prodigy 30 points
    My concern is however not on the 2nd harmonic. I notice plenty of spurs in the spectrum, all spaced at an interval of (2fvco/3)/4. Can you explain that? When the frequency is fractionalized or integerized, this behavior varies greatly.
  • 0 in reply to Sid Balasubramanian62
    TI__Mastermind 28031 points
    Sid,

    When you divide by 3, you get frequency fvco/3. Then the 2nd harmonic of this is 2fvco/3/

    The output divider has several segments that can impact spurs. Also, if it's related to the fraction, then controls like MASH_ORDER might impact this and make it clear if this is due to the output divider or MASH engine, so we can figure out how to mitigate.

    Also, if you have different frequencies from both outputs, there will be crosstalk and this causes spurs. Power supply decoupling may help some, but not eliminate this spur.

    Regards,
    Dean
  • 0 in reply to Dean Banerjee
    Prodigy 30 points

    >> When you divide by 3, you get frequency fvco/3. Then the 2nd harmonic of this is 2fvco/3/

    Owing to the frequency relationships we need, we are using both doubler and channel divider and that results in the "generation" of 2fvco/3 directly. That is what I meant in the beginning when I asked what would be the best way to generate 2fvco and 2fvco/3 on the two PLL outputs, since the amount of mixing spurs are intense.

    >> MASH_ORDER might impact this and make it clear if this is due to the output divider or MASH engine, so we can figure out how to mitigate. 

    It does not appear that the MASH engine is causing the problem. The spurs are being caused by the absolute fvco and the fractional nature of the channel divider outputs; and regardless of whether fvco's relationship to the fref is maintained integral or fractional.

    >> Also, if you have different frequencies from both outputs, there will be crosstalk and this causes spurs. Power supply decoupling may help some, but not eliminate this spur.

    The spurs that we are implying are in addition to the cross-talk spurs and the 1st degree harmonic mixing spurs of the PLL outputs. I am seeing sub-harmonics at fvco/12 or fvco/6 in this particular situation. This is clearly stemming from within the PLL's internal circuitry. The non-50% duty on the /3 only exacerbates this problem. At this point, it appears that might be challenging to achieve with the LMX2592.

    We would appreciate if the problem could be replicated by TI and advise if we need to explore alternate solutions.

  • 0 in reply to Sid Balasubramanian62
    TI__Mastermind 28031 points
    Sid,

    To replicate this I need this at the minimum:
    1. VCO frequency
    2. Doubler On/Off
    3. Output Frequency

    From what you are saying, it seems like you are using the doubler and divider at the same time. Although the datasheet might not advise against this, I can tell you that we never characterized the device this way or intended it to be used this way. There is always an alternative way to express the frequency without using the doubler and divider at the same time. For instance, OSCin=100M, Fvco=4500 MHz,Douber=On, Fout = 3000 MHz, then you would get better performance and lower current using Fvco=6 GHz and dividing by 2. For the divide by 3, I know that it was designed to work up to 7.1 GHz input, but if you put the doubler before this, it might be pushing the limits for what this was designed for.

    So in summary, if it turns out that the issue happens when the VCO doubler and 1/3 divider are used at the same time, but it goes away when one just changes the VCO frequency and uses the 1/2 divider with regular VCO, then then the "fix" would be to do this and perhaps us update our datasheet and TICSPro software to give better feedback on this.

    On the other hand, if you are seeing this issue when you are NOT using the doubler and divider together, then we need something more specific as to the setup. If you show what you are doing on our TICSPro GUI or give a TICSPro saved file, this is best to replicate, but we can also use registers.

    One other thing. Suppose that you are using the VCO doubler for RFoutA and not looking at RFoutB at all, but OUTB_MUX is set to channel divider or CHDIV_EN=1 with SEG1 = DIV3, then I could see how this could be causing the issue.

    Regards,
    Dean