The LoRa Developer Forum is now in read-only mode and new content will not be added.
Semtech, in its commitment to enhance user experience and streamline content, has successfully integrated the LoRa Developer Portal content into Semtech.com. As a result of this consolidation effort, the LoRa® Developer Portal Forum will be discontinued on May 1st. After this date, you will be automatically redirected to Semtech.com.
For any technical support related to LoRa, please feel free to reach out to our experts here. If you have sales inquiries, please contact us here.

FORUM

SX1262 lots of CRC errors in SF10

Semtech Transceivers SX126x family
#1

Hi

We have observed some strange behaviour of the SX1262 in SF10 LoRa mode (other SF’s work just fine).
Small packets up to ~80 bytes can be transmitted without any issues, but as soon as the packet size reaches ~90 bytes, virtually every packet arrives with a CRC error.
Interestingly, the CRC errors vanish (!) when reducing the TX power from +14dBm to e.g. +1dBm, which is counter-intuitive. We have tested on several radio frequencies and with various distances between receiver and transmitter from ~10m to 2km, always with the same result.

Our settings: bandwidth 0 (125kHz), code rate 1, preamble 10, +14dBm TX power, explicit mode, CRC enabled

Any idea what could be happening here?

#2

Hi Keepcoding,
We had a similar problem and it took us 3 month to find the reason!
At the end we found out, it was the XTAL. There is a very limited number of XTALs that are certified!
See App note: AN1200.59 Page 11!
Most of these Types are hard to find in the net. Try it with the Murata-Types. The special NDK-Type is especially made for LoRa. Don`t try a similar NDK-Type! You will have the same problem.

Good luck!

#3

Yes, this is a known phenomena which is cause by heat dissipation of SX126x.Please refer to AN1200.37- “Recommendations for Best Performance”

AN1200.37

#4

Thanks for your replies.

@ralflobenstein
Could be, but the crystal we chose (NX3225SA-32MHZ-EXS00A-CS02368) has better specs than the one suggested by Semtech. Maybe it is safer then to directly switch to a TCXO.

@hqiu
Yes, we have studied this document before we designed our board and paid special attention to the thermal design.

But I agree that this could be a possible explanation (at least partially). It doesn’t fully explain however why it only happens in SF10.