• TI Thinks Resolved

TICSPRO-SW: TICSPRO-SW & LMX2571

Part Number: TICSPRO-SW

Dear colleagues

In my opinion there are several defects in software "TICSPRO-SW" (http://www.ti.com/tool/ticspro-sw at his use with LMX2571:

1. "User Control" tab:
- there are no fields "PLL_N_F1/2" (integer portion of the N-divider).

2. "PLL" tab:
- The field "CP_IDN" is absent. It is possible to make the common field CP_I=CP_IDN=CP_IUP.
- It is impossible to set Fden=2^24=16777216.
- Values of fields don't correspond to values of registers.
- Switching of F1/F2 doesn't update the field "N Divider", "Fden", "FRAC_ORDER", "CDIV", Tx/Rx Power/Ena...
- Prevention at input of N<22 is highlighted at FRAC_ORDER_F1/2 = 3 or 4 - in DataSheet there is no such restriction.
- The fields PFD_DELAY_F1/2 are absent.

3. It is desirable to have the recalibration button = records in R0.

Yours faithfully,
Alexander

  • Hi Alex,

    Thanks for your suggestions, we will improve TICS Pro as much as we can.
    As for Fden, you can put 16777215, this is almost same as 16777216.
    Prevention of N<22 is correct, in fact, the datasheet has to be revised to include this information.
  • In reply to Noel Fung:

    Noel Fung
    Hi Alex,

    Prevention of N<22 is correct, in fact, the datasheet has to be revised to include this information.

    Hi Noel

    Age of DataSheet for today more than two years - I believe that the probability of a release of his new version is no hight in the nearest future.
    Could you write about these restrictions for N (integer portion of the N-divider)?

    About Fden=16777215 and 16777216=2^24 - will be spurs identical in both cases?

    Yours faithfully,
    Alexander

  • In reply to Tolchinsky Alexander:

    Alexander,

    For the fraction, if you reduce it to lowest terms, this gives an idea if it makes a difference. If the reduced fractional denominator is more than 1000000, then it probably doesn't.

    For instance,
    8388608/16777216 = 1/2 and this will look like 1/2
    8388608/16777215 ~ 1/2. This will be like 1/2, but have large amounts of dithering.

    In PLLatinum Sim tool, try simulating a simple fraction like 1/2 and then see the difference between 0% and 100% dithering and this should give you some idea.

    On the other hand,

    35927/16777216 and 35927/16777215 will have very similar spectrum as the fraction does not reduce.

    I'm planning on updating TICSPro this week for the release on two other products. I don't have much bandwidth to work on the LMX2571, but I'll see if there are any quick updates I can do for the LMX2571

    REgards,
    Dean
  • In reply to Dean Banerjee:

     Alexander,

    Let me respond to some of your TICSPro comments.

    1. "User Control" tab:

    - there are no fields "PLL_N_F1/2" (integer portion of the N-divider).

    Yes there are.  In the diagram in the bottom, first item under feedback divide is PLL_N.  In the picture, PLL_N_F1=21, PLL_N_F2=27

    2. "PLL" tab:

    - The field "CP_IDN" is absent. It is possible to make the common field CP_I=CP_IDN=CP_IUP.

    I had the impression that CP_IUP set both and the only way for CP_IDN to matter was to set some test bit.  It's that way for the LMX2594 and then we just hide CP_IDN and the test bit.  But it's not clear from the datasheet.  So if really the LMX2571 GUI has been out for this long and you are the first to mention, I am sort of shocked.  But maybe it's because the LMX2571 was around when we transitioned.  I need to look into this.

    - It is impossible to set Fden=2^24=16777216.

    Noel responded to this.  I think TICSPro is right and datasheet is being corrected.

    - Values of fields don't correspond to values of registers.

    - Switching of F1/F2 doesn't update the field "N Divider", "Fden", "FRAC_ORDER", "CDIV", Tx/Rx Power/Ena...

    It does on mine for both pin and Bit.  Is TICSPro giving some error message in the status bar below?

    - Prevention at input of N<22 is highlighted at FRAC_ORDER_F1/2 = 3 or 4 - in DataSheet there is no such restriction.

    I know that there is a restriction on N.  I think that the datasheet is being updated.  I'll make sure it is included. 

    - The fields PFD_DELAY_F1/2 are absent.

    OK.  On other devices we call it PFD_DLY, so we missed this one.  It is in the user controls tab, but would be nicer if it was on the front page.  I'm not sure if I have bandwidth to fix this now though.

    3. It is desirable to have the recalibration button = records in R0.

    We have the FCAL_EN already there and when you toggle it, it mentions that R0 is programmed and gives the value it is programmed to in the status bar on the bottom.

     

    So in summary, with a quick look, the biggest thing is the CP_IDN, which I think is OK, but not sure.

    Nice to add PFD_DELAY, but I have no bandwidth right now to do this week, but hopefully I get some time later.

    Thanks for the feedback.

     

    Regards,

    Dean

  • In reply to Dean Banerjee:

    Dean, thank You

    About Fden=2^24=16777216: if I use FSK, I must set the field PLL_DEN_F*=0 - it will mean that Fden=2^24=16777216

    About Switching of F1/F2. "Is TICSPro giving some error message in the status bar below?" - Yes: "Attemted to divide by zero"

    Regards,
    Alexander

  • In reply to Tolchinsky Alexander:

    1072.LMX2571.zipAlexander,

    I verified in the lab that if CP_IDN<>CP_IUP then performance was degraded.

    Frankly, it was sort of hard to believe this was necessary TICSPro has been out for so long and this is the first I have heard of it.  Also, in our newer parts, LMX2582 and later, we hide CP_IDN and it is forced internally in our chip to CP_IUP.

    Anyways, this seemed like a pretty big issue, so I added this to the TICSPRO GUI.  Also, I updated to show PFD_DELAY_F1/F2 on the front page

    As for the update problem, I'm not sure why you see it and I don't, but the divide by 0 error stops other code from running, so it's the divide by 0 error.

    Attached is the updated LMX2571.  I'll release to the web this week, but if you want to try it first, here it is.  To use this, unzip to a folder and do the "Add User Device" in TICSPro.  Or you can replace the LMX2571 with these files in the zip.  Just be aware that program files director hides the actual data in another program data directory.  I think that the included startup file *.tcb in the zip will hopefully make the divide by 0 error go away.

    Regards,

    Dean

  • In reply to Dean Banerjee:

    Hi, Dean

    Thank You for updating - I tried it (only it was necessary to rename "LMX2571.ini" in "1072.LMX2571.ini").

    "Attemted to divide by zero" - it is caused by Fden=0.
    Value Fden=2^24=16777216 are a permissible, it is used in FSK-mode and it has to be entered (on the tab "PLL") and to be used for calculating frequency. When Fden=2^24=16777216, field PLL_DEN_F *=0 (see R24.10 and Section 8.2.1.2 in DataSheet).

    It would be good to have on tab "User Control" all fields of registers, since this tab works always, and the tab "PLL" is not... - it about fields "PLL_N_F1/2" and button "records in R0" :) (constantly it is necessary to switch between tabs "User Control" - "Raw regesters" for record in R0 for calibration).

    Regards,
    Alexander

  • In reply to Tolchinsky Alexander:

    Alexander,

    Thanks for the feedback.

    I'll try to figure out a way for Fden=0, or at least stop it from crashing.

    I'll see about the PLL_N value.

    Regards,
    Dean