• State
  • Locked Locked
  • Replies 5 replies
  • Subscribers 23 subscribers
  • Views 1662 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.

LMX2594EVM: Output Power and Control

Stan Zubiel
Prodigy 80 points

Part Number: LMX2594EVM
Other Parts Discussed in Thread: LMX2592

I've been experimenting with the LMX2594EVM and was unable to get the output power levels shown in the data sheet.  With the output connector tied directly to an Agilent MXA spectrum analyzer input, the measured power drops below 0 dBm at about 13.5 GHz, ultimately hitting -2.8 dBm at 15 GHz.  I also noticed that the output power at 2.0 GHz, for example, varied by 5 dB depending on whether the output divider is 4 or 6.  It also appears that when the output power control register is changed, it isn't monotonic.  Seems to reverse slightly around a setting of 30.

Do you have some guidance on using reactive components to increase the output power at the higher frequencies.  I know the data sheet recommends using  1 nH inductor instead of the 50 ohm pullup, but I wonder what the effect is at low frequencies.  Any impedance or equivalent lumped element models for the outputs?

I also noticed that there is some leakage of the lower order output dividers appearing along with the desired signal.  For example, with 15.0 GHz programmed, there is a -55 dBc component at 7.5 GHz.  Seems to improve as the output frequency decreases.

Still have some questions about the PLLatinum Sim program, but will leave that for another day

  • 0
    TI__Expert 8625 points

    Stan,

    The datasheet number of 2dBm at 15GHz is at the point of the device pin, then depending on the loss through PCB (the EVM loses about 2-3dB) it can measure less at the spectrum analyzer. Regarding output divider 4 or 6, the output power should not be dependent on what divider is used but by the output frequency you have, 12G / 4 = 3G or 12G / 6 = 2G, although I don't expect that big a change between those two frequencies. The output power control isn't monotonic you are right, it ramps up then slowly saturates, and there are two portions, so 0-31 is one ramp, then like you saw at 32 it drops down and begins the next ramp 32-63. sometimes the highest power is actually around code=50 (since after saturating it goes down a little).

    In picking the pull-up inductor use the equations below. You should expect highest output power around the frequency you designed the inductor at, but also the output power to fall when you go higher or lower frequency from that.

    For the leakage, it is sometimes related to if you don't have the internal dividers turned off, check that first. Also, remember to terminate the other side of the output "N" or "P" side that is not used, I've seen if the two sides are not matched you can get some undesired mixing or reflections.

  • 0 in reply to Brian Wang
    Prodigy 80 points
    Brian:

    Thanks for the information. Regarding the output power vs divider, I am referring to the observed change at 2 GHz using DIV = 4 or 6. You can get to 2 GHz both ways, 12 GHz/6 or 8 GHz/4.

    I don't see the equations you mentioned.

    The leakage I am referring to also appears when a divider is selected. For example, with Fvco at 12 GHz, DIV = 2, Fout = 6 GHz, I see a component at 3 GHz. Apparently the divider is a ripple counter with the additional stages active. Is it possible turn the unused dividers off?
  • 0 in reply to Stan Zubiel
    TI__Expert 8625 points

    Stan,

     

    Regarding the using different output power, the level should be dependent on the output buffer, however using different channel divider together with deriving from a different VCO can have slightly different waveform characteristics and slightly different output power measurement. I get about 1dB difference, what do you see?

     

    The equation is from LMX2592 datasheet section 8.1.6. Looks like it doesn't let me copy images here.

     

    The leakage is interesting. We didn't expect that. After looking into it, there is a bit R31[14] to set to 0 (it creates that additional div-by-2 component). Let us make sure and update that accordingly. In the meantime you can try it from the "Raw registers" tab in TICS Pro since it is not an exposed register bit.

     

     

     

     

  • 0 in reply to Brian Wang
    Prodigy 80 points

    Brian:

    Clearing that bit eliminated the discrete spurs, which previously appeared symmetrically around the desired output.  Now all I see is a broad (1 GHz) sinx/x looking spectrum at one-half the output frequency at about -80 dBc.  I need some guidance on how to use this register.

    If I set the output power register to 63 at 2000 MHz with the channel divider at 4, I measure Pout = +9.3 dBm.  With the channel divider at 6, I measure +1.5 dBm.  Repeating with output power set to 50, I get +8.7 dBm and +7.6 dBm respectively.  Much better.

  • 0 in reply to Stan Zubiel
    TI__Expert 8625 points
    Stan,

    That register you can force write to 1 for now, it is being updated and described in upcoming datasheet update.

    Yes that output power register is the control that determines the output power. Like you saw already, it goes up but at some point saturates (around 50) and then could go slightly down approaching the max value of 63.