About tx failed

[JenkinTu]

SX1302 has certain propability less than 10% failed to tx packets. As the following log, the packet should be transmitted when 9:18:11, but it had been replaced by the packet should be transmitted when 9:18:22. This packet had been written into SX1302 registers ahead of 38ms before transmission. Why hadn't this packet been transmitted?

---

4-29 9:18:10 [/usrdata/log/pktfwd] INF thread_down:3219 JSON down: {"txpk":{"imme":false,"rfch":0,"powe":17,"ant":0,"brd":0,"tmst":3624665456,"freq":475.5,"modu":"LORA","datr":"SF7BW125","codr":"4/5","ipol":true,"size":25,"data":"YDQAACSA+U8CytoOJOBBIHsN/Eyouzmo8w=="}}
4-29 9:18:11 [/usrdata/log/pktfwd] INF thread_jit:3764 lgw_send done on rf_chain 0: count_us=3624665456, contractor_time= 3624627141
...

4-29 9:18:22 [/usrdata/log/pktfwd] INF thread_down:3219 JSON down: {"txpk":{"imme":false,"rfch":0,"powe":17,"ant":0,"brd":0,"tmst":3636165503,"freq":476.7,"modu":"LORA","datr":"SF7BW125","codr":"4/5","ipol":true,"size":25,"data":"YCAAACSAGFUC9QccIlsmFO2gTAba26fXSQ=="}}
4-29 9:18:22 [/usrdata/log/pktfwd] INF thread_jit:3737 WARNING: a downlink was already scheduled on rf_chain 0, overwritting it...
4-29 9:18:22 [/usrdata/log/pktfwd] INF print_tx_status:3674 INFO: [jit] lgw_status returned TX_SCHEDULED
4-29 9:18:22 [/usrdata/log/pktfwd] INF thread_jit:3764 lgw_send done on rf_chain 0: count_us=3636165503, contractor_time= 3636126593
......

[mcoracin]

If the packet has not been transmitted, it is probably because it has not been programmed in the sx1302 tx buffer fast enough.
I'm surprised that you dodn't get an error from the JIT mechanism of the packet forwarder, because the minimum TX delay is set to 40ms.

sx1302_hal/packet_forwarder/src/jitqueue.c

Line 36 in 4b42025

#define TX_JIT_DELAY 40000 /* Pre-delay to program packet for TX in microseconds */

Anyway with 38ms, you are very close from the limit we have set.

Which LoRaWAN Network Server are you using ? The TX request should arrive sooner than this.

[JenkinTu]

We are using ChirpStack. The TX packet had already been sent from ChirpStack to gateway, and properly been enqueued into jit queue.
Ahead of 38ms the TX packet had been written into SX1302 tx buffer before transmission. Should we use TX_JIT_DELAY some bigger value than 40ms?

sx1302_hal/packet_forwarder/src/jitqueue.c

Lines 422 to 428 in 4b42025

	if ((queue->nodes[idx_highest_priority].pkt.count_us - time_us) < TX_JIT_DELAY) {
	*pkt_idx = idx_highest_priority;
	MSG_DEBUG(DEBUG_JIT, "peek packet with count_us=%u at index %d\n",
	queue->nodes[idx_highest_priority].pkt.count_us, idx_highest_priority);
	} else {
	*pkt_idx = -1;
	}

[mcoracin]

Yes, you can try to increase the TX_JIT_DELAY to better match with your system latencies.