• State Resolved
  • Locked Locked
  • Replies 8 replies
  • Subscribers 55 subscribers
  • Views 317 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.

CC1310: CC1310 Frequency Offset

Maaz Awan
Prodigy 155 points
Part Number: CC1310
Other Parts Discussed in Thread: CC1101

Greetings forum!

We are experiencing a peculiar behavior that we are not able to corelate with theoretical or expected outcome on CC1310. We are testing the ability of the device to withstand crystal drift by streching to its limits. The test setup and radio settings are elaborated below:

1.The radio configuration is legacy long range mode with DSSS=1 at 70kbps symbol rate and 35KHz deviation and Fc at 868MHz.

2. The crystal in use has an accuracy of +-10PPM on both Tx and Rx,as per Design Note DN005, titled CC11xx Sensitivity Versus Frequency Offset and Crystal Accuracy, the compensation for crystal inaccuracies is given by the 4*XTAL ppm i-e, 4*8.68Khz,34.72KHz.So the required receiver filter bandwidth becomes=2*70KHz+34.72KHz=174.72KHz. This effective signal bandwidth is easily catered by Rx filter bandwidth of 196KHz at the receiver.

3. With radio link established and working fine, we begin to change the receiver Fc from 868 MHz in steps of 1KHz. It is at 868.030 MHz, a crystal drift of 30KHz, when the link is compromised and everything goes blank at the  receiver. At this point.

4. If our mathematics is correct, there is now a need for a Rx filter bandwidth of 174.72+30=204.72Khz to receive the complete channel bandwidh, so we proceeded to change the Rx filter bandwith from 196KHz to 255KHz only to find out that there was no link availability. But as soon as we increase the Rx filter bandwidth to 311KHz, the next programmable step for Rx filter bandwidth, the link is established again smoothly.

We are unable to fathom that why an effective signal bandwidth of 204.72KHz is not completely received within the 255KHz BW of the receive filter?

Please guide me through this query as we are not able to explain this behaviour.

Best Regards

Maaz Ali

  • 0
    TI__Expert 3960 points

    Hi Maaz,

    if you crystal is +- 10 PPM, the about +- 34 PPM frequency error tolerance you are getting and 196 KHz bandwidth should be enough for your application. Is there a reason you would like to support a higher frequency error?

  • 0 in reply to Diego P
    Mastermind 6160 points

    Diego, have you checked if the combination of deviation and RX BW is possible?

  • 0 in reply to Diego P
    Prodigy 155 points

    Greetings Sir!

    It is more than enough to cater the signal within the 196KHz Bandwidth...But actually there are a couple of issues in our production line including pick and place process of PCB stuffing. Baking the PCB at temperatures above 200°C in the oven for melting the solder paste led to a significantly higher drift in the crystal PPM values.Moreover, the aging per year also adds a toll of 1PPM/year.

    In light of these issues, we are moving to CC1312 with TCXO support but for now we want to test the limits of the device with present constrains. Please let me know the theoretical aspect of the phenomenon we are experiencing.

    Best Regards 

  • 0 in reply to TheGhostOf
    Prodigy 155 points

    Greetings sir!

    Please elaborate the point you are alluding to.

    As per my understanding, when the offset gets to the same order as the deviation, this requirement of 311KHz arises, but as per my understanding of the 2-GFSK waveform, it does not make sense with modulation index of 1.

    Best Regards 

  • +1 in reply to Maaz Awan
    Mastermind 6160 points

    Some comments/ questions:

    - PHY: Why have you selected legacy long range mode with a fairly high datarate and with DSSS=1? In case it's for some reason related to the phy, have you tested with a "normal" phy with the same symbolrate/ deviation?

    - For the test you have done, what is the initial frequency error? When you program the device to use 868 MHz, what is the offset?

    - It's a few ways to do calibration on the production line of the frequency offset. Have you considered this? 

    - One of the reasons the legacy long range mode was replaced with a different mode due to poor frequency offset tolerance. 

  • +1 in reply to TheGhostOf
    Prodigy 155 points

    Greetings sir!

    We used the Legacy long range mode because as per my knowledge, there is no DSSS=1 on the simplelink long range mode and the 2-GFSK mode does not offer FEC. Since we wanted to benefit from FEC, the only fallback was legacy long range mode. WB-DSSS uses FEC but the constraint length K is 4 which is much less than K=7 in both long range modes.We wanted better error correction at highest possible data rate.

    Let me get back to you on the unmodulated CW test results soon.

    Thank you so much sir for your time.Pray

    Best Regards

  • 0 in reply to TheGhostOf
    Prodigy 155 points

    This is an epiphany... great source of information. However, I believe WB-DSSS with spreading=1 and 2-GFSK with CC1101 FEC both have K=4 convolutional channel coding so i should expect similar performance.

    Thank you so much for the help.

    Best regards