• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 10 subscribers
  • Views 928 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

LMX2485e Oscillator input and Lock detect

Matt11
Prodigy 120 points
Other Parts Discussed in Thread: LMX2485E

I am currently using the LMX2485e in a design and had a number of questions. The device is configured as follows:

VCC 3v
phase detector freq 50 KHz
loop bandwidth 100 Hz
charge pump current 1.52 mA
oscillator in 450 KHz
RF out 52 MHz

Loop filter parts:

C1 = 913pF
C2 = 4.39nF
R2 = 856K

1. An in house designer said that the minimum oscillator input frequency is not a critical specification and that it is only critical that the oscillator input frequency have a fast enough rise time. Our system uses a 450 KHz oscillator input into the LMX2485e which is well below the 5MHz stated minimum. This 450 KHz signal is a square wave to provide a fast rise time on the signal edges. Is it OK to operate the device like this. If not what kind of unexpected behaviour could I expect?

2.While monitoring the charge pump output into the VCO on an oscilloscope, I seem to observe different behaviour on the line when the ftest pin is configured for RF lock detect vs. RF_N/2. Is there a known issue here?

3. Does the lock detect function operate better at higher phase detector frequencies?

4. Is there any kind of relationship between lock detect performance and oscillator input frequency?

5. In my standard set up with the oscillator input frequency set to 450 KHz I cannot get the lock detect pin (ftest) to go high unless I enable the phase detect divide by 4. Is this a known issue or could it be caused by my low input oscillator frequency?

6. Will the charge pump output have a defined state either high or low rail if it is not locked or will the output be undefined?

  • 0
    TI__Mastermind 27581 points

    Matt,

    1.    The issue with slower OSCin frequencies is the slow slew rate, so the square wave does address this.   However, if the datasheet says 5 MHz and this is viloated, then this is violated.  So what I am saying is that 450 kHz will probably work on OSCin, but it is NOT guaranteed.   If it was not to work, the issue is that you could get noise to cause the R divider to miscount, most likely making the PLL to lock to a higher frequency than expected.

    One thing to be aware of is self-oscillation of input pins.  If you remove the input completely, some devices, but not all will have the pin self-oscillate.   So I would reason that if the input frequency is too low, then at some point it would be possible.   Maybe 470 kHz is high enough to avoid self-oscilation, but we don't have much characterization data on this.   For sure the 5 MHz input spec is high enough.

    2.  This is expected as the Ftest/LD pin is configurable to many different modes of operation.  The RF N/2 is the N divider output divided by 2 and is intended for diagnostic purposes.  Going back to question 1, you might want to do an RF R/2 with your 450 kHz input and see what this says, as it is the actual counte output for the device.   Lock Detect is supposed to indicate a locked or unlocked condition.  Two different things.  

    3.  In general, digital lock detect works better at lower phase detector frequencies.   This is becuase as the phase detector period gets too small, it gets closer to the turn on time of the charge pump.  Not much of an issuein this device, but lower is better in general.  For analog lock detect, maybe higher might be better.   

    4.   The lock detect is based on the phase detector frequency, but not directly on the oscillator input frequency.

    5.   I think that this is most likely your low input frequency.  The divide by 4 is to reduce the phase detector frequency to the lock detect input, but 450 kHz is quite low already

    6.   The output could be undefined.  For instance, if you remove the oscillator input frequency completely, the input could self-oscillate and cause the charge pump to rail high.  If you remove the VCO input completely, the FinRF pin could self-oscillatoe and cause the charge pump to rail low.  If neither, you would expect to see a signal that alternates because the charge pump is oscillating between tri-state and either sink/source.

     

    Regards,

    Dean

  • 0 in reply to Dean Banerjee
    Prodigy 10 points

    Hi Dean,

    the min F  RFin is specified as 50MHz for the Rf synthe. Would it work reliably down to 45MHz ?

    Would this part be available for next 5 years?

    Erkka

  • 0 in reply to Erkka Sointula
    TI__Mastermind 27581 points

    Ekkra,

    In general, TI is very reluctant to obselete parts, so although I can't guarantee this, it is by far most likely that this part will be available 5 years from now.

    As for operation, we can see on the typical sensitivity curves that if you put enough power, a typical device can easily operate at 45MHz.   At these lower frequencies, the issue is slew rate, so if you use a square wave or higher input power, it should be able to go much lower.   However, TI takes electrical specifications very seriously so I can NOT GUARANTEE this operation.   To do this requires justification from marketing, special screening, stamp-off, and that sort of thing.   If the volume of th opportunity is very high, then this makes sense.  However, if the volume is not as high, then getting a formal guarantee is not as easy to do.  So in summary, although it proably will work down to 45 MHz, I can NOT  GUARANTEE this operation.

    Regards,

    Dean Banerjee

     

    “Nothing is certain but death and taxes.” -- Benjamin Franklin