Well, the fact that the ‘corruption’ was later in the packet did suggest a LDRO issue, but maybe not.
What I would want to do is have another receiver, next to the one that is reporting the errors, but using the RadioLib or another LoRa library direct, as in no Meshtastic involvement, to see if the same errors are reported.
I’ve seen these kind of errors when the radio gets overloaded. The signal seems good enough to detect the preamble, but data decoding fails. Lowering the transmit power or moving them further apart may “fix” it.
Daily messaging restriction is over, i’m back)
Just for fun - my node with disabled CRC check re-transmit corrupted messages to other nodes 
And here is my problematic nodes
Using the LoRa settings of SF11, BW250Khz, Power 2dBm, I ran up a TX and RX test with a pair of SX1262 sending packets of length 230bytes. TX and RX were next to each other on the bench, received packets reporting RSSI -30dBm, SNR +8.
4,000 packets sent, 1 receive CRC error. Seems reliable enough.
Yup, my other nodes works like a charm with LF preset.
Just found what lowering tx power to 2 mainly solves the problem.
I plan to
If you had two receivers, both in the same location, you should be able to tell if the problem is a TX or RX one.
If its a TX issue, expect both RXs to report and error for tha same packet and possibly have the same corruption for the same packet.
If its an RX issue expect the RXs to report errors for different packets.
Nice idea.
But everything points to a problem with TX (3 my receivers got CRC error on incoming message) and they are in 1-2 meters from each other.
But i’ll check if they will catch the same corrupted message later (anyway need to update they fw to latest 
)
So are you are saying that 3 receivers, at the same close distance from the transmitter, report CRC errors on the same packet ???
Yes, and no one send back ACK, so transmitter tries to send 3 times and mark message as undelivered in app
So
At 20 tx power still low percentage of success delivery with any antennas combinations
And I see better result when I put my hand on pcb antenna
My current opinion is that the antenna is poorly designed and causes overload of the lora module…
For reference - Here GPSFan modified Chatter 2.0 for use with RA-62
I dont find much else where people used this module on the web
There are some users in "DIY LoRa modules & NRF boards" thread in discord
I think problem solved and reason is - terrible antenna ( but not fully understand why this behavior exist only on few presets).
Antenna removed, used ipex pigtail and small sma antenna - now long messages delivers near in 100% attempts
Later will try to fix second problematic node…
Thanks for all for your support
If the antenna was ‘bad’ you might expect to see differances in RSSI\SNR.
In 2-5 meters - I don’t see any diff in RSSI/SNR
In 400m (no LOS) - node was able to receive, but can’t transmit. And i didn’t made comparison with other node at such distance
