• State TI Thinks Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 4 answers
  • Subscribers 22 subscribers
  • Views 807 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.

LMX2595: VCO calibration: Start-up sequence and CE

Christian Meyer
Prodigy 50 points
Part Number: LMX2595

Hi, 

in the datasheet of the LMX2595 you recommend the following power-up sequence:

1. Apply power to device.

2. Program RESET = 1 to reset registers.

3. Program RESET = 0 to remove reset.

4. Program registers as shown in the register map in REVERSE order from highest to lowest.

5. Wait 10 ms.

6. Program register R0 one additional time with FCAL_EN = 1 to ensure that the VCO calibration runs from a stable state.

I have a few questions regarding that procedure: 

Does the 10ms waiting time relate to point 1 or point 4 in the power-up sequence and whats the reason for this waiting time?

Is it possible to start the VCO calibration when the CE pin is low?

Kind regards

Christian

  • 0
    TI__Guru** 108550 points

    Hi Christian,

    The 10ms is to allow the internal LDOs to settle down before doing VCO calibration again (step 6). Following this programming sequence will ensure a successful VCO calibration. 

    When CE is low, the device is in power down mode except that the SPI interface and related logic is still awake. So, we can program the device when it is in power down mode. However, VCO calibration will not happen as all the related functional blocks are sleeping. 

  • 0 in reply to Noel Fung
    Prodigy 50 points

    Hi Noel,

    thanks for your answer.

    Originally I wanted to bypass the VCO calibration:

    I built up several PCBs with the LMX2595 and saw that the tuning voltages differ from the measured one (1.25V) when I read back the parameters for full assist calibration. Some of the PCBs start with an output frequency (not a pure CW, there are more frequencies close to each other in a band of a few MHz) that is approx. 20-50 MHz shifted above the desired frequency when I activate the CE. Within 2 seconds after CE the frequency is regulated the the wanted CW tone. I tracked the tuning voltage that starts from the upper value (2.5V) and lowers to 2.0 V. Other PCBs with the same programming don't show this effect, they start with an tuning voltage of 1.1 - 1.3V and the desired frequency is almost immediatly there. When I start the LMX2595 at cold temperatures I can see similar effects.

    Is it normal that the tuning voltages varie that strongly in full assist mode for different synthesizers? I expected that also in full assist mode the frequency will be stable over the whole operating temperature range of the synthesizer. I already noticed that the CAPCODE has a major influence on the tuning voltage, could it be helpful to change something there or do you have any other hints related to that problem?  

  • 0 in reply to Christian Meyer
    TI__Genius 12560 points

    Hi Christian,

    For operating the LMX2595 in full assist mode, each part may have slight different VCO settings (CAPCODE, etc..) and there are possible chances where each part have different tuning voltage if set with the common VCO settings. Please confirm, if you are using the common VCO settings in full assist mode for all the devices and seeing the different tuning voltage?

    If yes, please have individual VCO settings for each part and then lock the device and should get the required tuning voltage. Also over the temperature, there can be variation in tuning voltage in expectable range and get the PLL locked with correct frequency.

    Regarding the 2 second time for the frequency correction, if the VCO settings (digital) are not correct, it can show the frequency shift but the CE high the analog loop filter and charge pump current settings, it will correct the required frequency in sometime.

    Thanks!

    Regards,
    Ajeet Pal

  • 0 in reply to Ajeet Pal
    Prodigy 50 points

    Hi Ajeet,

    yes, you're right, all devices use the same VCO settings (VCO_CORE, VCO_DACISET, VCO_CAPCODE) in full assist mode. These settings I found with the evaluation board (first VCO calibration, then read back the VCO settings to TICS Pro) with an already modified loop filter (same loop filter as on the PCBs).

    The tuning voltages of the devices differ by almost 1V at ambient temperature. For me that seems to be quite a lot if it's based on the manufacturing tolerances of the device.

    So, do you think the reason for the 2 seconds has something to do with the VCO settings or maybe could there also be an issue with the loop filter?

    Kind regards

    Christian

  • 0 in reply to Christian Meyer
    TI__Guru** 108550 points

    Hi Christian,

    Is below your operation procedure?

    - full assist mode

    - all devices use the same VCO full assist VCO settings

    1. Vcc power power (CE pin = Low)

    2. Program all the registers

    3. CE pin = High

     --> some devices output signal with correct frequency and Vtune immediately

    --> some devices output signal with incorrect frequency and Vtune. After 2 seconds, frequency and Vtune resumes to normal

    I think the 2 seconds delay is normal because VCO_DACISET and VCO_CAPCODE depends on the internal VCO and varactor bias voltage, they are determined by the internal LDOs. When CE pin is pull-high, the LDOs take some time to settle to the programmed values. As a result, the output frequency may be off immediately after CE pin is pull high. During this time, PLL tries to track the VCO, therefore, Vtune may be off as well. 

    2 seconds delay looks a bit long to me. One of the reasons could be the pull-high signal slew rate is poor. 

    Due to part to part variation as well as the LDO bypass capacitors variation, the LDOs response time will vary. In addition, the VCO settings will vary, too. My suggestion is do not use the same VCO settings for all devices. Every device should be calibrated on-board. This would also ensure that the calibrated VCO settings works for all temperature.

  • 0 in reply to Noel Fung
    Prodigy 50 points

    Hi Noel,

    yes, the procedure you described is exactly the way I operate the LMX2595.

    I played around with the programming of the synthesizers. When I change the VCO_CAPCODE a few numbers, so that I get a tuning voltage of approx. 0.9 - 1.2 V at ambient temperature. When enabling CE at ambient temperatures, low and high temperatures I don't see such effects anymore. When the tuning voltage is above 1.4 V or below 0.7V at ambient temperature there are locking issues at high temperature (for 0.7V) and low temperature (for 1.4V). Then The output frequency isn't a stable CW (similar effect as described in my second post).

    Is it normal that the voltage range of VTUNE (0.7 - 1.4V) is such small when I want to ensure that the synthesizer works at the entire operating temperature range of the device? For the desired output frequency I have a tuning gain around 70 MHz/V. Could that be a reason for this small voltage range of VTUNE?

    Kind regards

    Christian 

  • 0 in reply to Christian Meyer
    TI__Guru** 108550 points

    Hi Christian,

    VCO calibration algorithm will try to get approx. 1.3V Vtune at the current temperature. This is to ensure that the PLL remains lock over the entire operation temperature range without a re-calibration. That is, after calibration, the three VCO setting values remain valid for any temperature in range. Vtune, however, will change accordingly with temperature. This is obvious as the VCO free running frequency will change over temperature. Another purpose of the 1.3V Vtune after calibration is to ensure Vtune also has sufficient room to swing on both direction in response to the temperature change. Our charge pump is 2.5V operated, valid Vtune voltage is approx. 0.4V to 2.1V. If you have 0.7V Vtune at room temperature, you may have problem at high temperature. 

    Please also note that Kvco is not linear over the Vtune range. It is highest at 1.3V and decrease at other Vtune voltages.