ipq5018/qcn6122: add BDFs for TP-Link Archer AX55 v1 - #143
Conversation
|
Force-pushed an important correction: the original files carried the generic QCA reference-board data, which left the 2.4 GHz receiver ~40 dB deaf on this hardware (uplink stuck at HE-MCS 0). The |
|
Force-pushed a second correction. Both blobs came out of the stock-firmware Verified on hardware, median of 30 RSSI samples from a stationary client:
md5: For context on priority: openwrt/openwrt#24197 (Archer AX55 v1 support) needs |
|
Why are you including BDF-s with no variant string as well? |
|
You are right - those were leftovers from the stock-firmware extraction: the reference-board entries (qmi-board-id 32/35/36 on the IPQ5018 side, 80/96/144/176 on the QCN6122 side) plus a variantless duplicate of the calibrated entry (byte-identical to it). On this board the firmware reports qmi-board-id=255 and both radios set the calibration variant in the DTS, so none of those extra entries is ever selected. Both containers are now trimmed to the single calibrated entry, matching the rest of the feed - each file drops from ~656 KB to 128 KB. The board data itself is unchanged (same MD5 as before). Force-pushed. |
Board data from the stock firmware of the TP-Link Archer AX55 v1 (IPQ5018 2.4 GHz radio). The qmi-board-id=255 entry carries the calibration the vendor firmware actually selects for this board (bdwlan.b24, qcom,board_id 0x24) - the generic reference-board data leaves the 2.4 GHz receiver ~40 dB deaf on this hardware. The target-power tables came out of the extraction with 33 regions (1136 B) zeroed, which made the firmware fall back to 13 dBm instead of the 20 dBm the regulatory domain allows; those regions are filled from the vendor entries of the same file. Measured on hardware as the median of 30 RSSI samples from a stationary client: -48.0 dBm at 20 dBm against -56.0 dBm with the output forced to 13 dBm. The container carries just this single calibrated entry (qmi-board-id 255 plus the calibration variant), matching the other files in this feed; the variantless duplicate and the unused reference-board entries that came along from the stock extraction are dropped. Signed-off-by: Stanislaw Pal <kuncy7@gmail.com> Link: openwrt#143 Signed-off-by: Robert Marko <robimarko@gmail.com>
Board data from the stock firmware of the TP-Link Archer AX55 v1 (QCN6122 5 GHz radio). The qmi-board-id=255 entry carries the calibration the vendor firmware actually selects for this board (bdwlan.b60, qcom,board_id 0x60). As with the IPQ5018 blob, the target-power tables came out of the extraction with 39 regions (1472 B) zeroed, capping the band at 13 dBm instead of 23; filled from the vendor entries of the same file. Measured on hardware as the median of 30 RSSI samples from a stationary client: -61.0 dBm at 23 dBm against -69.0 dBm with the output forced to 13 dBm. The container carries just this single calibrated entry (qmi-board-id 255 plus the calibration variant), matching the other files in this feed; the variantless duplicate and the unused reference-board entries that came along from the stock extraction are dropped. Signed-off-by: Stanislaw Pal <kuncy7@gmail.com> Link: openwrt#143 Signed-off-by: Robert Marko <robimarko@gmail.com>
3 participants
Board data files for the TP-Link Archer AX55 v1 (IPQ5018 2.4 GHz + QCN6122 5 GHz), extracted from the stock firmware.
Requested in review of the device support PR: openwrt/openwrt#24197 — the BDFs were originally carried in
package/firmware/ipq-wifi/files/there and have been removed from that PR in favour of this submission.Both radios verified working on hardware with these board files (ath11k, factory MACs from caldata).