v5.0.1

• HAL: Reworked the way the TX start delay is calculated, taking into account
  the delay introduced by TX notch filter (when enabled) and the delay linked to
  signal bandwidth separately.

v5.0.0

• HAL: Changed GPS module to get native GPS time (monotonic, no leap second).
  WARNING: The native GPS time is not given in standard NMEA messages, so we
  get it from a proprietary message of the GPS module used on gateway reference
  design, u-blox 7. If you are not using the same GPS module, you may have to
  update the lgw_parse_ubx() function.
• HAL: Added lgw_cnt2gps() and lgw_gps2cnt() functions for SX1301<->GPS time
  conversion.
• HAL: Changed serial port configuration for GPS to properly handle binary
  messages (UBX).
• HAL: Added a lgw_gps_disable() function to restore serial configuration as
  it was before HAL initialization.
• HAL: Fixed packet time on air calculation using the actual packet preamble
  size.
• HAL: Adjusted TX_START_DELAY based on the board reference design to ensure
  that a packet is sent exactly at 1500µs after the TX trigger (TIMESTAMP or PPS),
  with a tolerance of +/- 1µs. This is mandatory to comply with LoRaWAN
  specification for Class-B beaconing precise timing.
  WARNING: This release provides tx start delay values to be used for Semtech
  reference designs AP1 and AP2. The HAL automatically detects the board version
  by detecting a FPGA or not. If you are using a different reference design or
  a different FPGA binary version than the one provided with this release, the
  value to be used for TX start delay may be different.

v4.1.3

• HAL: Reference clock frequency error improvement: The lora_gateway HAL has
  been updated (3 registers changed) to improve the performance of all gateways
  based on SX130x. The fix greatly improves the reception of packet at SF12, when
  the frequency offset of the incoming packet is large (mostly below -20ppm of
  frequency offset).

WARNING: Systems which do not have the patch will be more prone to packet loss
over time, when the crystals of the end-devices will be ageing and have more
frequency offset.

v4.1.2

• HAL: Changed configuration of IQ polarity of FPGA for TX to comply with FPGA
  version greater than v27. (Only required for AP2 Semtech reference design)
• HAL: Updated default LoRa preamble size according to LoRaWAN spec.

v4.1.1

• HAL: Fixed bug in "Listen-Before-Talk" which was preventing from configuring
  the Scan Time to 5ms.
• MISC: Added GPIO number to reset_lgw.sh command arguments.

v4.1.0

• HAL: Reworked "Listen-Before-Talk" feature to have more flexibility to define
  LBT channels frequency, and to be able to have spectral scan running in parallel
• HAL: Updated lgw_time_on_air() function for FSK packets
• HAL: Disabled GPS UART input being re-echoed as output to avoid sending wrong
  commands to GPS module
• HAL: Fixed IF frequency configuration check issue for channel bandwidths 250K
  and 500HKz.
• FPGA: Updated to v33 for new LBT and spectral scan design.
• util_spectral_scan: updated to match new spectral scan FPGA sequence
• util_lbt_test: updated to match LBT rework

Note: The provided LBT feature has been validated for Japan only, and supports
8 downlink channels maximum.

v4.0.1

• HAL: SX1301AP2: Only FPGA v27 is supported, removed (v18,v19) from the list
  of supported FPGA images.

WARNING: If you are using a Semtech SX1301AP2 ref design (GW1.5), the FPGA must
be reprogrammed with one of the images provided with this release (fpga/ dir).

v4.0.0

• HAL: Added "Listen-Before-Talk" support for Semtech SX1301AP2 Ref Design.
  A description of the feature implementation can be found in
  libloragw/readme.md.
• HAL: Updated FSK RSSI calculation for better linearization
• util_lbt_test: New utility provided for basic "Listen-Before-Talk" testing.
• util_tx_test: Extended to configure and test "LBT" through the HAL.
• Added a reset_lgw.sh script to be used with IoT Starter Kit (v1.0) to reset
  the concentrator through the HOST GPIO pin.

v3.2.1

• HAL: Fixed downlink support for SX1301AP2 reference design: soft reset of the
  FPGA was missing for proper IQ inversion configuration.
• HAL: Added support for several versions of FPGA (currently v18 and v19)
• HAL: Reduced radio TX PLL bandwidth to reduce the noise level.
• HAL: Change cs_change usage policy in SPI module to let the driver handle the chip
  select.
• util_tx_test: Added FSK support and added minimal TX gain LUT.
• util_spectral_scan: Removed FPGA soft reset, now done by the HAL.
• util_tx_continous: reworked to use HAL functions instead of 'manual' config,
  and use same SX1301 calibration firmware as the HAL.
• Updated all makefiles to handle the creation of obj directory when necessary.

v3.2.0

• Added support for SX1301AP2 reference design (with FPGA and additional SX1272). When a FPGA is detected at startup, the HAL automatically adapts SPI communication requests (using SPI header or not).
• Added util_spectral_scan diagnostic tool to scan the spectral band in background, where the LoRa gateway operates. (can only be used with SX1301AP2 or similar design). By default it uses the same SPI device as the one used by the HAL, but it can be changed depending on the hardware architecture on which it is used by updating the SPI_DEV_PATH constant defined in file util_spectral_scan/src/loragw_fpga_spi.c.
  Note: when using same SPI device from 2 applications, we rely on the host SPI driver and OS to properly handle concurrent SPI requests. It has been tested on Raspberry Pi / Raspbian with spi_bcm2708 driver.
• Removed SPI FTDI support due to lack of performances to properly handle heavy packet traffic. Only native SPI usage is recommended.
• HAL: added a check that SX1301 firmwares have been properly loaded at startup.