• State Resolved
  • Locked Locked
  • Replies 24 replies
  • Answers 1 answer
  • Subscribers 56 subscribers
  • Views 2760 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.

CC2592: R+jX of the antenna port not listed in the datasheet. What is the impedance?

Jt Whissel
Prodigy 160 points

Part Number: CC2592
Other Parts Discussed in Thread: CC2530

R+jX of the antenna port not listed in the datasheet of the CC2592. What is the impedance? 

  • 0
    TI__Guru** 106705 points
    Hi Jt,

    For the same reasons as listed here: processors.wiki.ti.com/.../CC26xx_Optimal_Load_Impedance the CC2592 does not have a "optimal load impedance" value stated in the datasheet. The recommendation for best RF performance is to follow the reference designs closely.

    Regards,
    Fredrik
  • 0 in reply to Fredrik K
    Prodigy 160 points

    Can you provide the algorithms to calculate the passives for the matching circuits off your "optimized reference designs" when the stack-up / dielectric are changed? Since you do not provide the R+jX I cannot simply calculate that my self. 

  • 0 in reply to Jt Whissel
    TI__Guru** 106705 points
    I recommend that you use the same dielectric thickness between the RF layer (top) and the ground plane (layer 2). Increasing the thickness slightly is typically fine, but I would not recommend decreasing it due to parasitic capacitance.

    If you need to use thin dielectric layers due to micro vias for example, you can place the ground plane on layer 2 or 3 and simply keep the layers in between free for copper in the RF area.
  • 0 in reply to Fredrik K
    Prodigy 160 points

    This is the stack I must use. Doing any kind of antenna turning is going to be a nightmare with no reference R+jX. 

    Thickness

    		 Thickness (mm)

    Layer

    Tolerance

    		 Tolerance (mm)

    1 mil

    		 .025

    Solder Mask

    ±0.25mil

    ±0.01

    1.5 mil

    		 .04

    1 oz copper

    +0.5mil

    ±0.02

    13 mil

    		 .33

    FR4 Prepreg

     

    0.7 mil

    		 .02

    0.5 oz copper

     

    28 mil

    		 .7

    FR4 Core

     

    0.7 mil

    		 .02

    0.5 oz copper

     

    13 mil

    		 .33

    FR4 Prepreg

     

    1.5 mil

    		 .04

    1 oz copper

    +0.5mil

    ±0.02

    1 mil

    		 .025

    Solder Mask

    ±0.25mil

    ±0.01

    PCB Substrate dielectric constant of 3.95 @ 1GHz

  • 0 in reply to Jt Whissel
    TI__Guru** 106705 points
    The dielectric constant in the PCB substrate does not influence the matching significantly. We have actually stopped specifying a specific value for our kits, as we do not see any noticeable variation because of it.

    With 330 um thickness between the RF layer and the ground plane I think you should be ok. Looking at the schematic here: www.ti.com/.../tidriz6 , the DC block, C27, is your 50 ohm point. You need to match your antenna to 50 ohm here.

    Keep in mind the impedance matching and filtering out the chip is not developed through simulations and calculations alone. We have done a significant amount of testing and measurements, especially load- and source -pull, to get the performance right. And it is not only the impedance at 2.4 GHz that matters, impedance at all harmonics is also critical for good performance. Thus the recommendation to follow the reference closely.

    Cheers,
    Fredrik
  • 0 in reply to Fredrik K
    Prodigy 160 points
    It can effect it quite a bit a few ohms depending on how much you care about getting as close to that 1.0 VSWR as possible. Is it tuned with the DC block or before the DC block? If you are this confident that your reference design can be implemented on that stack up, then I will take a leap of faith and use that front end. It scares me to death designing RF blind but we will see how it goes. Since no ammount of testing can prepare you for every situation, that is why having the R+jX measured at the pin is always good to have.
  • 0 in reply to Jt Whissel
    TI__Guru** 106705 points
    It is tuned with the DC block.

    The point is having an impedance number at 2.4 GHz would not make you less blind. You need to hit the right impedance at the harmonics also. To be able to tune the IC match you need access to source- and load-pull equipment.
  • 0 in reply to Fredrik K
    Prodigy 160 points
    I have a VNA and a few spectrum analyzers. The problem is the lack of chamber, but you can get away without one. But thanks for the help. Wish me luck.
  • 0 in reply to Jt Whissel
    Prodigy 160 points
    One other question before I take a leap of faith. If there are any problems relating to RF mismatching. Does TI stand behind their design enough to give engineering support if problems arise?
  • 0 in reply to Jt Whissel
    TI__Guru** 106705 points
    If any problems arise, do let us know and we will help you resolve the problem.

    Cheers,
    Fredrik
  • 0 in reply to Fredrik K
    Prodigy 160 points

    Got our boards back from the manufacture and there is a weird problem. The boards work perfectly with Smart RF Studio, using the SmartRF05EB. I can fully control my board through that software. The interesting problem is when it comes to programming. I can erase & program but it fails on trying to verify. Do you have any guess on what could be the issue?

  • 0 in reply to Jt Whissel
    TI__Guru** 106705 points
    Which SoC are you using?

    Some of our SW Stacks write configuration settings to flash during the first boot. This can happen between the programming and verification step which will make the flash content appear to be incorrect. You can check this by reading back the flash content and compare manually.
  • 0 in reply to Fredrik K
    Prodigy 160 points

    We are using cc2530 and just trying to load up the bootloader. 2 out of our 6 boards can flash and verify perfectly. However, every single one of those boards works flawlessly with Smart RF studio. I am really puzzled what can be causing this. 

  • 0 in reply to Jt Whissel
    TI__Guru** 106705 points
    Can you try to read back the flash content and compare with your original hex file?
  • 0 in reply to Fredrik K
    Prodigy 160 points

    I inserted the hex file below (we will see if it shows up in this forum) 

    output.txt

  • 0 in reply to Jt Whissel
    TI__Guru** 106705 points
    This shows no content in the flash, so it actually fails to program.

    What does the supply voltage look like when you program?
  • 0 in reply to Fredrik K
    Prodigy 160 points

    On the ones that work, its pulled low. But on the ones that don't work it's pulled high. I have logic analyzed one of the ones that pull high as it is being plugged to the debugger. It is interesting that it looks like it does not start pulled high but pulls high after. I will attach a few screenshots of the capture.

    This is a view of the last "block" of data zoomed in a bit.

    Here is the Matlab file (I can do CVS if you have something to read it) 

    startup.zip

  • 0 in reply to Jt Whissel
    Prodigy 160 points

    Just did a capture of one working and there is only one difference between the two. I attached screenshots of the two captures. The first one is from a board that cannot be programmed (line never gets pulled back to low) and the seconds one is from one that can (line gets pulled to low)

    [Not Working Startup]

    [Working Startup]

  • 0 in reply to Jt Whissel
    TI__Guru** 106705 points
    Hmmm. This is strange. Can you also compare the reset signal? Do you have any pull-ups mounted on the debug lines? Are you able to read the info page on the failing units? What happens if you change the interface speed to slow?
  • 0 in reply to Fredrik K
    Prodigy 160 points

    The reset signals look the same, I did notice the reset lines looking like they coupled with the DD line for some reason. However, I busted out an EVM board to compare against and the coupling was there too.

    No pull-ups or the debug lines and I am able to read the page on the failing units. The same problem persists in slow mode as well.

  • 0 in reply to Jt Whissel
    TI__Genius 12900 points

    Hi, can you send a logic data with DC, DD and RESET signals in the same plot?

    Thanks

  • 0 in reply to TIABO
    Prodigy 160 points

    Sure thing. The top one is the startup of a working board connected to a cc debugger and the bottom one is from a non-working board. I also attached Matlab data for each capture. 

    Working Board Startup

    Non-Working Board Startup

    Matlab files

    matlab-data.zip

  • 0 in reply to Jt Whissel
    Prodigy 160 points

    I found the issue out and would like to report if you run VCC to the DCOUPL pin like a derp like I did then you will have these erratic behaviors. I am actually surprised some of them kinda worked 3.6v to a 1.8v pin I figured it would just be fried. Good job making it pretty high tolerances TI. 

  • 0 in reply to Jt Whissel
    TI__Guru** 106705 points
    Haha! Good to hear that you figured it out. Also good that you did not actually break the chips.