LT5581_detector error

Dear team,

Our customer is using the LT5581 to detect 3.5GHz TDD RF signal.

But at different modulation of the input signal, there is much difference in the detector value(about 5dB error).

She tried to tune for improvemnet but could not.

##test condition.

 - Frequency : 3.6GHz

-  Input signal modulation : LTE20M TM1.1 and LTE20M TM2.0

- Input Power : -10dBm.

I am looking forward to your advice for inproving performance.

Note)She is also using the LTC5582 and this part has not difference in detector values depending on the different modulation of input signal.

Thank you.

Best regards,

Josh

  • 0
    •  Analog Employees 
    on Dec 17, 2020 8:05 PM 2 months ago

    Hi Josh,

    I'm not familiar with TM1.1 vs TM2.0, but remember a TDD waveform requires special treatment to measure, because of the waveform duty cycle. To measure TDD power, normally it is only during the signal active time, not the off time of the waveform. If your customer uses a simple DVM, the DVM will average across the off times along with the on times of the waveform. There will be huge error.  The solution is to use an oscilloscope to only measure during the correct time portion of the waveform. Hope that's clear.   -Bruce H.

  • Hi Bruce.

     

    Customer is using the TDD SINC signal for 3.5G TDD control.

    SINC signal is receiving from Test equipment.

    So there is no problem with measured method.

    And LTC5582 is no difference int detected value even if the modulation for the input signal is changed.

     

    The response time of LT5581 is slow. Is this related?

    • Fast Response Time: 1μs Rise, 8μs Fall

     

    Can LT5581 be used for TDD application(sub 6GHz_5G)?

    Is there a way to reduce the detector error when we change the modulation of input signal?

     

    I am looking forward to your advice.

    Regards,

    Josh

  • 0
    •  Analog Employees 
    on Dec 23, 2020 1:54 AM 2 months ago in reply to doolpung

    Hi Josh,

    Could you have your customer use an oscilloscope on the detector output and check the waveform? That should illustrate the problem. Because its TDD, the same frequency is timeshared for Tx and Rx. It means the signal generator is going to generate a complex output waveform for a certain period of time that might be milliseconds, then there will be no output for another period of time that might also be milliseconds. A simple DVM on the output cannot capture this, but an oscilloscope will clearly illustrate the TDD nature.

    The simple solution to power measurement here is to synchronize the measurement with the TDD rate. In other words, customer uses an ADC on the detector output, and the ADC must be programmed to only take the power sample after the PA or Rx carrier has been on for at least a few microseconds. Any power measurement during the off time of the TDD waveform will be meaningless and invalid.    -Bruce H.

  • Hi Bruce,

    I sent the tested data of LT5581 and LTC5582 at TDD application. The Oscilloscope was used.

    When the signal mode was changed, LT5581 had a large difference in detection voltage, but LTC5582 was similar.

    1. Is there a way to reduce the detection voltage difference of LT5581? In fact, This part was tuned aroued but did not improve.

    2. The LTC5582 is fine, What is the difference between the two parts?

    3. Please advise if there are any additional methods to solve this problem of LT5581.

    I am looking forward to your advice.

    Thank you.

    Regards,

    Josh

  • 0
    •  Analog Employees 
    on Jan 5, 2021 8:39 PM 1 month ago in reply to doolpung

    Greetings Josh,

    From the scope pictures, we see the LT5581 has much faster rise & fall time, compared to the LTC5582. It's the faster response time that causes the higher variations in output power measurement vs time, while the RF carrier is on. In essence, it appears the LT5581 detector is too fast for the application, tracking the instantaneous power through each of the symbol transitions. The simple fix is to install larger capacitor on the LT5581 CSQ pin, which is provided just for this purpose. It may require a little experimentation to find the best tradeoff of output noise vs response time for any given input waveform.

    LT5581 design uses a square cell input stage as the core detector.

    LTC5582 also has a similar rms detector stage, but also includes a logarithmic VGA at the front end, and uses internal closed-loop servo control of the VGA to deliver extra dynamic range performance.    -Bruce.