Debugging lgw_send (not sending)

[wavesoft]

Hi All!

I have ported libgw and fragments of lora_pkt_fwd in order to be able to control a RAK2247 concentrator (SX1301 + 2xSX1257) from an MCU. I can say that I have read through about 80% of the library by now and my only unknown parts are the internals of SX1301.

So far I had great success with the API abstractions and I am successfully receiving uplink packets and forwarding them to my server. However, I have some issues with my downlink that I cannot explain.

In detail:

• I receive an uplink packet with "tmst":666233252 from the concentrator
• I forward it to the server and I receive a downlink packet with "tmst":671233252
• I call lgw_send passing down the parameters received from the server and I notice a successful scheduling of the transmission:

  INFO: tx_start_delay=1495 (1495.500000) - (1497, bw_delay=1.500000, notch_delay=0.000000)
  Info: packet will be sent without CRC
  INFO: Enabling TX notch filter
  lgw_send done: count_us=671233252
  

• Checking the device, I never see an incoming downstream packet
• Moreover, in future transmissions (a few minutes later) I see the following warning, that obviously indicates that SX1301 is somehow still in "Transmission" mode

  a downlink was already scheduled, overwritting it...
  

Question

1. Why isn't the transmissions happening?
2. How can I debug why it's not leaving the concentrator?
3. What's the bit 7 on LGW_TX_STATUS? Can I have the full register description?

Debug Info

• I have configured the libloragw with the values from global_conf.json.PCB_E286.EU868.basic.
• The contents of status register before the first transmission is LGW_TX_STATUS=80
• The contents of the status register before the second transmission is LGW_TX_STATUS=91
• I noticed that the LGW_TX_STATUS resets to 80 after ~70 minutes which is suspiciously close to the time the 32-bit microsecond-timer overflows.
• The actual payload that goes to LGW_TX_DATA_BUF_DATA is:

  0x3ffee2c0 59 06 66 08 b9 d7 ec 0a 00 17 21 10 00 08 00 00 |Y.f.......!.....|
  0x3ffee2d0 20 53 b2 ab 8e 7f 2f d2 20 63 50 a7 8e aa e1 82 | S..../. cP.....|
  0x3ffee2e0 4e 55 65 38 f6 70 2f e8 bc 65 14 3e 60 66 2c eb |NUe8.p/..e.>`f,.|
  0x3ffee2f0 91 |.|
  

• I start the transmission setting LGW_TX_TRIG_DELAYED=1
• The calibration phase is giving me the following warning:

  problem in calibration of radio A for image rejection
  

• The (RAK2247 module) supply voltage drops from the nominal 3.3V down to 3.15V when the concentrator is started (indicating possible voltage drop due to resistance in the voltage path since the voltage right after the regulator is still 3.3V), but it remains stable during the operation of the module (eg. no further voltage drop after the scheduled transmission).
• The library is setting the following registers:

  LGW_TX_OFFSET_I=4
  LGW_TX_OFFSET_Q=18
  LGW_TX_GAIN=0
  LGW_TX_START_DELAY=1495
  LGW_TX_TRIG_ALL=0
  LGW_TX_DATA_BUF_ADDR=0
  LGW_TX_TRIG_DELAYED=1
  

[wavesoft]

To my utter surprise, it ends up to be an SPI issue!!!
Happy new year everyone!

[lichuanhe]

hello，the same problem comes 2me ,when the node send package with ack. sometimes the device cannot receie ack,then ,the the gateway printf "a downlink was already scheduled, overwritting it...".gateway useed the software simulated spi with mode 0,would u pls tell me how to solve this problem ,thinks very much!!!

[covenant1969]

Was this one solved? Sounds a bit like my problem too.