• State Resolved
  • Locked Locked
  • Replies 11 replies
  • Answers 2 answers
  • Subscribers 17 subscribers
  • Views 3029 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.

Is low latency (<5ms) data streaming (appr. 200kbps) with CC2650 possible (proprietary protocoll, not BLE)?

Daniel Burgener
Prodigy 70 points
Other Parts Discussed in Thread: CC2650, CC2541

Hi, I’m looking for a data streaming solution to wirelessly transmit data over a short distance. Important points are:

  • unidirectional point-to-point transmission
  • distance transmitter-receiver <1m - Data rate: 4000 samples/s, whereas one sample consist of 6 bytes (-> Effective throughput = 190'000 bit/s)
  • latency < 5ms
  • Frequency: 2.4GHz
  • Coexistance with 4 parallel running point-to-point connections

Due to high latency (connection interval) and low data rates, BLE is not suitable. The use of audio-streaming chips is not suitable either due to the latency of >20ms. In my opinion one solution could be the use of the CC2650 in the 5Mbit/s proprietary radio mode. Unfortunately I couldn't find any timing information in the data sheets to estimate the latency. My questions are:

  • Is the effective throughput of 190’000 bit/s possible?
  • What is the expected latency?
  • 0
    TI__Mastermind 36395 points
    Daniel,

    We have made a simple point to point steaming example on our older CC254x series devices. It will actually do basically what you are asking for.

    www.ti.com/.../litabsmultiplefilelist.tsp

    Regards,
    /TA
  • 0 in reply to TA12012
    Prodigy 70 points

    Hi,

    Thank you for you answer. I already saw the AN124 you mentioned. Right know I'm evaluating an radio chip based on the latency and throughput specification. Unfurtunately in the AN124 the latency is not mentioned. I need to know the latency (time it takes for the transmitter until payload is transmit + time on air + time it takes for the receiver to receive the paket). The time on air is based on the hardware overhead and the radio playload and results in 92us (assuming 23 bytes of playload+overhead at 2Mbps). But the time it takes until the data is tramsit and received is neither mentioned in the datasheet of the CC2650 nor in the AN124. In the AN124 the expression 'TX/TX slop' is used (0.2ms, p.6 in AN124). Is this what I am looking for? If yes, that no realy helps me, as I would like to use the CC2650 for my application.

    Do you know how to calculate the expected latency for the CC2650 based on data sheet information?


    Regards,

    Daniel

  • 0 in reply to Daniel Burgener
    TI__Mastermind 36395 points
    Daniel,

    It takes approximately 650us from when you issues a command on the CC2650 to when the first bit of the preamble is transmitted. This limits the latency as the "turn around" time is set by that number.

    Regards,
    /TA
  • 0 in reply to TA12012
    Prodigy 70 points

    Hi,


    and what about the time it takes at the receiver side to receive the complete packet and, for example, to set a output pin from low to high?

    I need this information to calculate the latency for the whole path.

    Regards,

    Daniel

  • 0 in reply to Daniel Burgener
    TI__Mastermind 36395 points

    Daniel,

    On the RX side the end of packet signal is provided at the end of the last CRC bit with little to no latency. The data will be present in the memory space within a few micro seconds of the end of packet signal. 

    1. unidirectional point-to-point transmission
      1. This makes is easier to get fast communication.
    2. distance transmitter-receiver <1m - Data rate: 4000 samples/s, whereas one sample consist of 6 bytes (-> Effective throughput = 190'000 bit/s)
      1. Range is not an issue
    3. latency < 5ms
      1. As previousely state the command parsing on the TX side is likely going to be the limiting factor. However you are allowed to issues multiple commands behind each other. Getting repeat rates of 5ms is not as issue, you can likely get repeat rates all the way down to 1ms.
    4. Frequency: 2.4GHz
      1. CC2650 will work fine for this band.
    5. Coexistance with 4 parallel running point-to-point connections
      1. Say you choose 2mbps on the CC2650, each channel will occupy 3MHz. I think spacing them at 5MHz apart would be a good start. This will provide you with same 16 frequencies as Zigbee.

    Regards,
    /TA

  • 0 in reply to TA12012
    Prodigy 70 points

    Hi,

    You mentioned, that it is likely to get repeat rates all the way down to 1ms. Do I am correct that this means, that that limits the measurement rate to 1'000 samples/s? The target data rate is 4'000 samples/s (see number 2 of the list with the important Points).

    Regards,

    Daniel

  • 0 in reply to Daniel Burgener
    TI__Mastermind 36395 points

    Daniel, 

    No, what I would recommend doing is to take 8 consecutive samples at 4000sps of 6 bytes each. Put the data from those 8 samples in 1 packet, therefore sending packets at 500 packets / second. Where each packet contains 6*8 = 48 bytes of data.

    Regards,
    /TA

  • 0 in reply to TA12012
    Prodigy 70 points

    Hi

    Thanks for your input. Assuming I transmit only one sample per packet, what would be the maximum packet rate? In this specific case the sample rate would correspond to the packet rate.

    Regards,

    Daniel

  • 0 in reply to TA12012
    Prodigy 70 points

    650us is very high. 


    Surely, if  you don't change frequency or make any other adjustments, a simple Tx/Rx switch can take less time?  For example, Nordic claims 20 us for the nRF52832 if frequency remains unchanged.  I am asking for a application that will use a proprietary protocol, not BLE.

  • 0 in reply to John Malsbury
    Prodigy 70 points
    Even the cc2541 claims less - 130us - 150 us depending on BLE or prop. mode... Can someone from TI post more specifics?
  • 0 in reply to John Malsbury
    TI__Mastermind 36395 points
    Unfortunately the CC26x0 is slower than previous generation when it comes to TX/RX switch and hopping time. The number vary but are between ~200-650us depeding on if you need channel hopping or just TX/RX switchover.

    Regards,
    /TA