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.

Connection lost with good RSSI but POOR Correlation value on CC2530.

Other Parts Discussed in Thread: CC2531, CC2530, Z-STACK, CC2591

Hello!

In our application the Zigbee network is composed of one coordinator and several end devices. The poll rate of the end device is 10 seconds for low power consideration as well as the indirect transfer timeout of the coordinator is 18 seconds. The coordinator maintains the connection status by sending "Heart Beat" packet to each end device and wait for its acknowledgment.  The coordinator will do this one by one, which means it will not poll the next end device until the acknowledge packet of current end device is received or the indirect transfer timeout occurs.  If no acknowledge is received within the past 15 minutes the connection between the coordinator and the end device will be treated as "Lost".  

We setup this network and observe the stability of the communication between the coordinator and the end devices. It works fine most of the time however sometimes the loss of the connection is detected. We logged the RSSI and correlation value from the data packet (with correct FCS result only) to trace the signal quality and found that during the lost period and a short period before the lost, the RSSI value is stable while the correlation value become worse. For example, the RSSI value could be about -75dbm all the time, but the correlation value could be changed slowly from 108 to 59, and then the connection is lost. After a while (several minutes or half an hour or even longer) the end device is re-joined the correlation value could be changed from a poor value back to 108 slowly. In another example when the connection is lost, the last RSSI is -58dbm while the last correlation value is 85. 

The following description is from TI "CC253x System-on-Chip Solution for 2.4-Ghz IEEE 802.15.4 and Zigbee Applications User's Guide" Section 23.10.4

“The RSSI value can be used by the MAC software to calculate the LQI value. This approach has the
disadvantage that, e.g., a narrowband interferer inside the channel bandwidth can increase the RSSI and
thus the LQI value, although the true link quality actually decreases. The radio therefore also provides an
average correlation value for each incoming frame, based on the first eight symbols following the SFD.
This unsigned 7-bit value can be looked on as a measurement of the chip error rate, although the radio
does not do chip decision.”


As far as I can see, if the correlation value is decreased by a narrowband interferer, the RSSI should be increased. But in our cases the RSSI value keeps as a constant value, sometimes even 2-3 dB lower, it does not look like a narrowband interferer right?  So would you please give me some advice about the possible reason and, which is more expected, guide me some way to find out the actual reason and to improve the communication's stability, since it is very important in our application.

By the way, our application is based on ZStack-CC2530-2.5.1a Znp mode.

Regards.

Zhou Liang

 

  • I would suggest you to use Ubiqua Packet Analyzer to check this issue first.
  • Hi YiKai Chen,

    Nice to meet you and thanks for your suggestion.

    We use SmartRF Packet Sniffer together with CC2531 USB dongle to observe the Zigbee network traffic. Under your suggestion I download the Ubiqua Packet Analyzer from its web site and request a trial license with fully functionality. It looks like similar as SmartRF  Packet Sniffer, may be more powerful in protocol analysis, but I didn't find any information about the correlation value which is calculated by the physical layer based on the preamble symbols before the SFD. 

    My interest is why the correlation value is decreased slowly while the RSSI is hold as a constant value and not increased? In our application almost every time when the end device lost its parent,  we can find that:

    1. Before the lost moment, the correlation value, both from this end device and from its parent, have been decreased slowly from a good value (such as 107/108) to a poor value (such as 60/70).

    2. After some time the correlation value is back to the good value and the communication between this end device and its parent resumes stable. This Growth is also very slowly. 

    3. The RSSI measured by this end device keeps almost unchanged during the whole period, and so be its parent.

    4. The communication between the parent and the other children keeps fine with good correlation value and unchanged RSSI during the whole period

    So the correlation value's decrease looks like an indication to the signal quality's worsen. We are trying to find out the clue that cause the decrease. Could you please show me some possible factors that could lead to this and, is there some way to make a DEEP debug/observing on the signal processing on CC2530?

    Thanks.

  • I am not sure what you mean correlation value. By the way, Z-Stack 2.5.1a is obsolete and I know there are lots bug fixing in latest Z-Stack Home 1.2.2a. So, I would suggest you to use latest Z-Stack Home 1.2.2a to test this again.
  • Dear YK,

    Thanks for your suggestion. Z-Stack Home 1.2.2a supports CC2592 and CC2590 as descripted in TI's web while we are using CC2591 currently. Could we still use Z-Stack Home 1.2.2a directly or is there any modification need to adapt CC2591?

    And for the correlation value,  you can refer to TI "CC253x System-on-Chip Solution for 2.4-GHz IEEE 802.15.4 and Zigbee Applications User's Guide" Chapter 23.9.7 (SWRU191C) 

    The correlation value is the average correlation value over the first eight symbols following the SFD.

    Thanks.

  • Z-Stack Home 1.2.2a supports CC2591. You only need to define HAL_PA_LNA which is the same in Z-Stack 2.5.1a.
  • Dear YK,

    Thank you and I'll have a try.
  • You are welcome.