• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 2 answers
  • Subscribers 52 subscribers
  • Views 570 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.

Original question:

CC2538: CC2538 Hardware design problem

CC2538-CC2592EM-RD: Issue with CC2592?

DTeo
Intellectual 1705 points
Part Number: CC2538-CC2592EM-RD
Other Parts Discussed in Thread: CC2592, CC2538

Hi, 

I have a customer who copied the CC2538-CC2592EM-RD in their design but added 3 0ohm resister to do co-layout with on or bypass CC2592.

Now they are seeing the folloiwng issues. Measurements are done by scope and cable loss is 2dB.

  

1.       With CC2592, measure peak power is about -5dBm when target power is set 22dBm by Smart RF Studio.

 

2.       Bypass CC2592, measure peak power is about 4dBm when target power is set 7dBm.

  

3.       Set as RX mode, measure S11 from output of CC2592 to CC2538, S11 are under -10dB, it seems the RF matching is ok.

Hence customer suspect that the CC2592 is not working. Any advise on the issue?

Best Regards,

Dunman Teo

  • 0
    TI__Guru** 106705 points
    Hi Dunman,

    When using the CC2592 it is extremely important to follow the reference design to get proper performance. "added 3 0ohm resister to do co-layout with on or bypass CC2592" does sound like a source for performance issues.

    1. In this case, it sounds like the CC2592 is not running at all. Are the control signals configured correctly?
    2. Ok. The losses are probably caused by their "bypass" circuitry.
    3. It is not possible to measure the output impedance of the CC2592 like that.

    For more detailed support, please share the schematic and layout for the CC2538 and CC2592 part. They can remove everything else so you can post it here on the forum.

    Regards,
    Fredrik
  • 0 in reply to Fredrik K
    TI__Intellectual 1705 points

    Fredrik,

    For 1), After setting the TX mode, the voltage of control signal PA_EN is 2V8, LNA_EN is 3V#, HGM is 0V. is this correct?

    For 3), what is the correct way for the customer to check the output impedance of the CC2592?

    Thanks and BR,

    Dunman

  • 0 in reply to DTeo
    TI__Guru** 106705 points
    Hi Dunman,

    I have looked at the layout you sent, and as I suspected, the option to bypass the CC2592 is most likely causing issues due to impedance mismatch. The differential trace between the CC2538 and the CC2592 is too long and is made even worse by going through multiple layers.

    Additionally, the balun layout, when not using the CC2592, is not following the reference designs.

    They should still see higher output power than -5 though, so I also suspect there are issues with the control signals. The LNA_EN must be low for the device to be in TX (see the datasheet).

    CC2538 decoupling caps are not placed according to the reference designs, and CC2592 should have separate decoupling caps for the RF bias and the supply pins (as in the reference designs).

    Regards,
    Fredrik
  • 0 in reply to Fredrik K
    Prodigy 60 points
    Hi Fredrik,

    I'm the application engineer, the design problem block me too long, I need your help urgently.

    1. Set as RX mode by smartRF studio, measure S11 from output of CC2592 through CC2592 to CC2538, S11 are under -10dB, it seems the RF matching is ok. Is measure mode correct? And what's the correct way?

    2. The balun layout can not following reference design completely, because we do the co-layout.

    3. All control signals is correct.
    After setting as TX mode, the voltage of PA_EN is 2V8, LNA_EN is 0V, HGM is 0V.
    After setting as RX mode with high gain, the voltage of PA_EN is 0V, LNA_EN is 3V3, HGM is 3V3.

    4. CC2538 decoupling caps are all placed close to chip pins.
    What is the mean of RF bias? There is no decoupling caps in reference design.

    I don't think the problem above is the root cause, they maybe degrade the performance, but can not cause CC2592 not work.



    Best regards,
    Damon
  • 0 in reply to user5895909
    TI__Guru** 106705 points
    Hi Damon,

    1. The correct way is to use a load- and source-pull setup.
    2. Sure, but that will affect the performance. I believe this is why you see only 4 dBm output power for the 7 dBm setting.
    3. The logical levels look correct, but why is PA_EN only 2.8 V? The input is high impedance and should not draw that much current. This could indicate ESD damage, which again will explain why the device is not working.
    4. If you look at the CC2592 reference designs, there are 2 sets of decoupling caps. One for the VDD pins, and one for the VDD trace going to the RF output. This last part is what I refer to as RF bias.

    Regards,
    Fredrik