RadioLib 5.4.0 fails initialization for CubeCell

[MrSniffer]

Code runs OK with 5.3.0. SX1262 radio = new Module(RADIOLIB_BUILTIN_MODULE); ..... state = radio.begin(923.3,500.0,9,6,0x12,17,PREAMBLE_LENGTH,1.6,false);

Fails with 5.4.0; state = -707 (RADIOLIB_ERR_SPI_CMD_FAILED)

[rjmcdougall]

also fails with -707 on lillygo t-echo sx1262

[rjmcdougall]

06:04:57 9 Starting meshradio init... 06:04:57 9 Set radio: region=US, name=testchan, config=0, ch=55, power=30 06:04:57 9 Radio myRegion->freqStart / myRegion->freqEnd: 902.000000 -> 928.000000 (26.000000 mhz) 06:04:57 9 Radio myRegion->numChannels: 104 06:04:57 9 Radio channel_num: 55 06:04:57 9 Radio frequency: 915.875000 06:04:57 9 Slot time: 42 msec 06:04:57 9 Set radio: final power level=22 M SX126x M SX126x reset CMD 80 DATW 0 AA

CMD 80 DATW 0 A2

CMD C0 DATR 0 22 0 22

M SX126x standby CMD 80 DATW 0 A2

CMD C0 DATR 0 22 0 22

M SX126x config M SX126x setBufferBaseAddress CMD 8F DATW 0 A2 0 A2

CMD C0 DATR 0 22 0 22

M SX126x modem CMD 8A DATW 1 A2

CMD C0 DATR 0 22 0 22

M SX126x fallback CMD 93 DATW 20 A2

CMD C0 DATR 0 22 0 22

M SX126x cad CMD 88 DATW 3 A2 16 A2 A A2 0 A2 0 A2 0 A2 0 A2

CMD C0 DATR 0 22 0 22

M SX126x irq CMD 2 DATW 43 A2 FF A2

CMD C0 DATR 0 22 0 22

CMD 8 DATW 0 A2 0 A2 0 A2 0 A2 0 A2 0 A2 0 A2 0 A2

CMD C0 DATR 0 22 0 22

M SX126x calibrate CMD 89 DATW 7F A2

CMD C0 DATR 0 2A 0 0

06:04:57 9 SX126x init result -707

[jgromes]

@MrSniffer could you please post the debug output as well so that we know if this is the same issue?

@rjmcdougall it looks like the failure is occuring after the calibration command - it could be caused by the newly added SPI state verification (#575).

[MrSniffer]

Here you go. From the CubeCell:

Copyright @2019-2020 Heltec Automation.All rights reserved. [SX1262] Initializing ... M SX126x CMD 80 DATW 0 AA

CMD 80 DATW 0 A2

CMD C0 DATR 0 22 0 22

CMD 80 DATW 0 A2

CMD C0 DATR 0 22 0 22

CMD 8F DATW 0 A2 0 A2

CMD C0 DATR 0 22 0 22

CMD 8A DATW 1 A2

CMD C0 DATR 0 22 0 22

CMD 93 DATW 20 A2

CMD C0 DATR 0 22 0 22

CMD 88 DATW 3 A2 16 A2 A A2 0 A2 0 A2 0 A2 0 A2

CMD C0 DATR 0 22 0 22

CMD 2 DATW 43 A2 FF A2

CMD C0 DATR 0 22 0 22

CMD 8 DATW 0 A2 0 A2 0 A2 0 A2 0 A2 0 A2 0 A2 0 A2

CMD C0 DATR 0 22 0 22

CMD 89 DATW 7F A2

CMD C0 DATR 0 2A 0 0

failed, code -707

[jgromes]

@MrSniffer thank you, it fails in the exact same spot.

The fact that it does not fail on Arduino Uno is strange, but I guess than can be exaplin by Uno being a comparatively slow platform.

It looks like it's caused by performing command verification during an ongoing calibration. In the latest commit (170ce9752bf7ccf78713d60fb1a0b93f9ba3e3d2), I changed this to skip verification in this one case.

@MrSniffer @rjmcdougall could you try this hotfix on your boards?

[rjmcdougall]

Oh interesting -- yes I will try this

On Tue, Oct 4, 2022 at 9:20 AM Jan Gromeš @.***> wrote:

@MrSniffer https://github.com/MrSniffer thank you, it fails in the exact same spot.

The fact that it does not fail on Arduino Uno is strange, but I guess than can be exaplin by Uno being a comparatively slow platform.

It looks like it's caused by performing command verification during an ongoing calibration. In the latest commit (170ce97 https://github.com/jgromes/RadioLib/commit/170ce9752bf7ccf78713d60fb1a0b93f9ba3e3d2), I changed this to skip verification in this one case.

@MrSniffer https://github.com/MrSniffer @rjmcdougall https://github.com/rjmcdougall could you try this hotfix on your boards?

— Reply to this email directly, view it on GitHub https://github.com/jgromes/RadioLib/issues/583#issuecomment-1267253028, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAATJQ37USIL3UOR57W6MP3WBRKOHANCNFSM6AAAAAAQ36M5DY . You are receiving this because you were mentioned.Message ID: @.***>

[MrSniffer]

In a bit I will try this with a CubeCell and also a LilyGo TBeam V1.1, both w SX1262s.

[rjmcdougall]

yes that works!

On Tue, Oct 4, 2022 at 9:20 AM Jan Gromeš @.***> wrote:

@MrSniffer https://github.com/MrSniffer thank you, it fails in the exact same spot.

The fact that it does not fail on Arduino Uno is strange, but I guess than can be exaplin by Uno being a comparatively slow platform.

It looks like it's caused by performing command verification during an ongoing calibration. In the latest commit (170ce97 https://github.com/jgromes/RadioLib/commit/170ce9752bf7ccf78713d60fb1a0b93f9ba3e3d2), I changed this to skip verification in this one case.

@MrSniffer https://github.com/MrSniffer @rjmcdougall https://github.com/rjmcdougall could you try this hotfix on your boards?

— Reply to this email directly, view it on GitHub https://github.com/jgromes/RadioLib/issues/583#issuecomment-1267253028, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAATJQ37USIL3UOR57W6MP3WBRKOHANCNFSM6AAAAAAQ36M5DY . You are receiving this because you were mentioned.Message ID: @.***>

[jgromes]

Thank you for the confirmation @rjmcdougall, I will publish a hotfix release ASAP

[MrSniffer]

Not quite there, I think. Instead of failing 100%, it is maybe 10% now. Here is the debug output: Copyright @2019-2020 Heltec Automation.All rights reserved. [SX1262] Initializing ... M SX126x CMD 80 DATW 0 AA

CMD 80 DATW 0 A2

CMD C0 DATR 0 A2 0 22

CMD 80 DATW 0 A2

CMD C0 DATR 0 A2 0 22

CMD 8F DATW 0 A2 0 A2

CMD C0 DATR 0 A2 0 22

CMD 8A DATW 1 A2

CMD C0 DATR 0 A2 0 22

CMD 93 DATW 20 A2

CMD C0 DATR 0 A2 0 22

CMD 88 DATW 3 A2 16 A2 A A2 0 A2 0 A2 0 A2 0 A2

CMD C0 DATR 0 A2 0 22

CMD 2 DATW 43 A2 FF A2

CMD C0 DATR 0 A2 0 22

CMD 8 DATW 0 A2 0 A2 0 A2 0 A2 0 A2 0 A2 0 A2 0 A2

CMD C0 DATR 0 A2 0 22

CMD 89 DATW 7F A2

CMD 80 DATW 0 AA

CMD 17 DATR 0 AA 0 0 0 0

CMD 97 DATW 0 AA 0 3F 1 0 40 20

failed, code -707

[MrSniffer]

P.S. I am still getting occasional initialization errors with code -20 w 5.4.0 + this hotfix.

[MrSniffer]

I just tested multiple times on the LilyGo TBeam V1.1 SX1262 and no -707 errors found. Still having CubeCell failures. I can try to work around by re-initializing in the interim, as I do for the -20 error.

[MrSniffer]

T-Beam now exhibiting random failures, e.g. Bad startReceive(). Err=-705 RADIOLIB_ERR_SPI_CMD_TIMEOUT Still lurking problems, this one not at startup hence rather hard to get debug log.

[jgromes]

Reopened until I have more time to test with ESP32.

EDIT: In the meantime, patch 5.4.1 was released. Also the macro RADIOLIB_SPI_PARANOID can be disabled to completely stop the SPI status checks.

[jgromes]

@MrSniffer I was not able to replicate your issues directly, however, in two cases I was able to get my SX126x/ESP32 into an unstable state.

The first case was by introducing a slight error in the configuration - for example, attempting to use TCXO on a module with XTAL, which consisntely yielded -707 error on startup.

The second case was by using insufficient power source. For example, by powering an SX1268 module from a 3.3V pin of NodeMCU 32S board connected only to a USB port, I was able to get semi-regular -2/-20 error codes. When using a dedicated LDO, there were no issues transmitting or receiving on ESP32. I would suggest to examine the board(s) you are using, and what is the power source for the SX1268.

[MrSniffer]

FYI: I am using the LilyGo T-Beam V1.1 1262 model. It surely has a TCXO. I am using a voltage of 1.6 (default) but I have no clue what it ought to be. Where would I find that spec?

[jgromes]

@MrSniffer you will have to contact the manufacturer of that module.

[jgromes]

@MrSniffer I fixed an issue in the receive method that caused calls to receive() to end with error -705 when there was a timeout event. Not sure if this is one of the issues you had, but wanted to let you know.

[MrSniffer]

I have queried LilyGo to find out what TCXO voltage they designed for. The Semtech SX1262 data sheet says it should be a minimum of 1.6V, typical 1.7V, up to a max of 3.3V (using more power!)