• State TI Thinks Resolved
  • Locked Locked
  • Replies 1 reply
  • Answers 1 answer
  • Subscribers 20 subscribers
  • Views 1206 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.

LMX2594: lock problem

CEC KcMeter
Prodigy 80 points
Part Number: LMX2594

Hi, I have a few questions about the "LMX2594 vco calibration lock".

1. Lock time is too slowly

Test method

MCU will pull up one GPIO when the R0(FCAL_EN) is been written.Oscilloscope's one channel capture the GPIO rising edge，

and the another channel capture the LMX2594's MUXOUT pin rising edge.

Calculate the difference value between two rising edge, so this is the lock time.

Calculate result

The calibration lock time is 2ms when i use the No assist calibration.

What I have tried

1. use "Partial assist"

This is the step I have used according to the manual page 19:

1.1 Calculate the VCO core frequency according to the table "VCO Core Ranges".So I can determine the value of VCO_SEL.

1.2 Calculate the VCO CapCode and AMCal value according to the equation.So I can determine the value of VCO_CAPCTRL_STRT and VCO_AMPCAL.

1.3 Writen the value VCO_SEL to R20 and  VCO_CAPCTRL_STRT to R78 and VCO_AMPCAL to R17 before R0.

Final the calibration time is 1ms, but this is too slowly, i saw the manual indicate that the fast calibration algorithm allows changing frequencies faster than 20 us.

2. use "Full assist"

The manual didn't indicate the detail about "Full assist", so this is my procedure:

2.1 Set LMX2594 sweep at the range of 7.5GHz ~ 15GHz, and the sweep step frequency is 10MHz.

2.2 Readback the value about VCO_CAPCTRL and VCO_DACISET, save all this value to memory calibration list.

2.3 When user set a frequency, calculate the VCO_SEL, and readback the calibration list.

2.4 Then I use linear interpolation between the two calibration list.So I can calculate the value of VCO_CAPCTRL and VCO_DACISET.

2.5 Writeback this value to register...

Question

1. What are the factors about lock time? Am i do something wrong?

2. When I use Full assist, the readback value about VCO_CAPCTRL is always 55, and the VCO_DACISET is always 128, I have no idea about that...(And I'm sure I have set the MUXOUT's mode is readbak).

2. What is the relationship between output power and calibration lock?

The manual have indicated that the output power can be set the range of 0 ~ 63.But when I set the value larger than 30, the LMX2594 couldn't lock all the time! I don't know what happened about the chip.

I would be appreciated for your help...

  • 0
    TI__Mastermind 27581 points
    Test Method:
    The preferred method is to set the lock detect to "VCO Calibration Status" then also ensure that LD_DLY=0.
    The width of the negative pulse is the VCO calibration time. As for your method, if you were to trigger on the rising edge of the last pulse to the CSB pin, you should get a similar result. Can you confirm that this is what you are doing? 2 ms seems pretty slow and it would be good to be sure that you are not adding the time to program the device in your measurement.

    Use Partial Assist:
    Parital assist is closer to 35 us. 20 us is closer to close frequency calibration and 5 us is closer to full assist. 1 ms seems suspiciously long to me. Note that TICSPro has an automatic tool to calculate the values for VCO_SEL, VCO_CAPCTRL_STRT, VCO_DACISET_STRT. Also, be sure that you are not mixing up VCO_CAPCTRL_STRT with VCO_CAPCTRL and the same for VCO_DACISET_STRT.

    Full Assist:
    This is discussed more in the datasheet on the table in page 19. Basically the idea is you read back the values and then you force them in so there isn't really a calibration at all. We have measured 2 us in the lab. The thing with full assist is that it is dominated by the write speed, so you need something that can write fast. From your other results, it seems that you could be having some other setup problems, so you might want to debug them before grabbing a word generator for full assist.

    Comment:
    For debugging purposes, you might want to try just enabling the "QUICK_RECAL_EN" This should give you about 20 us calibration time assuming that you just toggle the FCAL_EN bit. This is good for debugging in that there are much less things that could go wrong with the setup.

    Questions:
    1. What are the factors influencing lock time?
    For the LMX2594, it is the OSCin frequency and phase detector frequecy that have the largest impact. The TI PLLatinum Sim tool from TI (www.ti.com/.../pllatinumsim-sw) simulates calibration time. Also, in the next few weeks, I am planning an upgrade for this for improved accuracy on calibration time, but the one on the web is reasonably good. For more detailed discussions, my 5th edition book at (www.ti.com/.../pll_book) has a chapter just on VCO calibration time.

    2. In regards to the output power, I am not sure what is happening. However, I can tell you that between 31 and 32, the power drops and then it goes back up again. I think at 48 you get the same power as 31, and therefore codes 48-61 might be higher, but not by much. At higher frequencies, it seems keeping this value at 31 or less is better for noise floor. I have certainly tested and characterized the part over all the power codes and don't know why you have the issue, but I am just saying that going above 31 might not give that much benefit anyways, and it increase the current consumption.

    Regards,
    Dean