Successful MG21 PowerTag reproduction with your firmware + findings and suggestions

[hocine69]

Hi,

First, thank you for publishing powertagd.

We successfully reproduced a full working Schneider PowerTag flow on an MG21 / Dongle-E class device using your firmware and tools:
• powertagctl pair successfully commissioned 0xe207b15b
• the PowerTag LED turned green
• live electrical measurements were received correctly

So your architecture works for us on Silicon Labs.

A few observations that may be useful to you:
1. The working mechanism seems broader than just a “special ACK”.
From the successful run, the important pieces look like:
• EZSP-alive NCP
• xNCP
• multi-RAIL
• raw GP interception
• scheduled secure response
• host-side PowerTag logic
2. On the successful radio trace, we consistently observed:
• 8c40
• 8c40
• 8c80
• 8c18
• 8c58
• 8c98
• then stable periodic traffic
That sequence was very useful for us as a reference.
3. We also noticed that normal Zigbee join does not work out of the box with the current host tools, not because EZSP is dead, but because the host flow appears PowerTag-specific and does not open standard permit-join.

We are not suggesting major changes here — your approach is the first clean Silicon Labs path we found that actually works with PowerTag.

If helpful, we can share:
• the successful pairing log
• the successful over-the-air pcap
• a short comparison with our previous failed attempts

Thanks again for making this available.