bladeRF/doc/development/test_checklist.md
2018-12-21 16:42:46 -05:00

19 KiB

bladeRF Test Checklist

This file serves as a quick summary of tests that should be run regularly by developers to detect defects and regressions.

Over time these procedures should become automated, and more fine-grained tests should be introduced. At that time, this document shall be updated to reflect usage of the automated tests.

When running through the items contained in this file, developers should maintain an annotated copy of this file, containing:

  • The date when tests were run
  • The device versions used when performing tests
  • "Checks" in the "boxes" ([ ]) and any associated observations or notes

Build Tests

FPGA Bitstream

  • Perform FPGA images build with Quartus Prime (free edition) 17.1 (or latest version) for the following:

    • Board: bladeRF

      • Revision: hosted

        • Size: x40
        • Size: x115
      • Revision: atsc_tx

        • Size: x40
        • Size: x115
    • Board: bladeRF-micro

      • Revision: hosted
        • Size: A2
        • Size: A4
        • Size: A9

FX Firmware

  • Perform firmware builds for both variants:

    • Release
    • Debug
  • Ensure the build operates on the FX3 SDKs supported operating systems:

    • Windows
    • Linux
  • Flash the images to appropriate boards and verify:

    • Correct version number reported by bladeRF-cli -e 'version' (tagged builds should have no commit hash; other builds should)
    • Correct VID/PID and product descriptors are present on lsusb -v or dmesg
      • bladeRF: "Nuand bladeRF" 2cf0:5246
      • bladeRF Micro: "Nuand bladeRF 2.0" 2cf0:5250

Host software

Verify that build and installation completes from the top-level, from a clean build directory, for each of below configurations.

Keep the "treat warnings as errors" option enabled; warnings should be regarded as defects.

With a device attached, use the following commands to verify the build has installed and that the desired backend is being used: bladeRF-cli -e 'version' -e 'info'

Additionally, ensure libbladeRF_test_c and libbladeRF_test_cpp build and run.

  • Debug build: -DCMAKE_BUILD_TYPE=Debug
  • Release build: -DCMAKE_BUILD_TYPE=Release -DTAGGED_RELEASE=Yes

Perform the above builds with the following:

  • Linux (various distributions; see host/misc/docker)

    • This can be performed automatically, using Docker: bash host/misc/docker/parallel.bash | tee /tmp/output.txt
    • GCC
    • Clang
  • macOS (latest)

    • Apple LLVM
  • Windows (currently Windows 10 w/ Visual Studio Community 2017)

    • Build with support for both libusb and the Cypress backend. This should be the default behavior when both present on the build system.
    • Build with support for each of the two available backends.
    • Perform builds targeting Win32 and x64
  • Misc

    • Build with documentation (from a clean CMake directory): -DBUILD_DOCUMENTATION=ON
      • Any Doxygen warnings should be regarded as defects.
    • Clang ccc-analyzer & scan-build (with -maxloop 100 or greater):
      • CMake incantation: cmake -DBUILD_DOCUMENTATION=ON -DCMAKE_C_COMPILER=/usr/share/clang/scan-build-6.0/libexec/ccc-analyzer ..
      • Build and generate report: /usr/share/clang/scan-build-6.0/bin/scan-build -analyze-headers -maxloop 100 -o ./report make
      • Any items in the static analysis report should be regarded as defects.
        • Exception: thirdparty/

Functional Tests

Most of the functional tests listed here are system-wide integration tests.

Unit tests, Built-in self-tests (BISTs), and top-level test-benches are very much desirable additions for improving QA and testing efforts. (Patches are always welcome!)

When appropriate, repeat each test for both USB 2.0 and USB 3.0 connections, and for both the bladeRF and bladeRF Micro.


FX3 Firmware Compliance Testing

The USB IF Compliance test tools shall be used to identify missing/incorrect functionality. Run these and record any failures or warnings.


SPI Flash Access

The following items are intended to exercise SPI flash accesses.

OTP region readback

There should be no warnings, info, or debug messages reported while performing this operation.

  • bladeRF-cli -e 'info' -v debug should report the serial number.

Calibration region readback

There should be no warnings, info, or debug messages reported while performing this operation.

  • bladeRF-cli -e 'info' -v debug should report the VCTCXO DAC calibration and FPGA size.

Writing calibration data

During the following steps, there should be no error, warning, info, or debug messages other than the information about flash page read operations.

  • Back up calibration data via: bladeRF-cli -e 'flash_backup cal.bin cal' -v debug

  • Write new calibration data (using the appropriate FPGA size) via: bladeRF-cli -e 'flash_init_cal 40 0x8000' -v debug

  • Power cycle the device and verify the new values (see the previous subsection).

  • Restore the original calibration data via: bladeRF-cli -e 'flash_restore cal.bin' -v debug

  • Power cycle the device and verify the new values.

Firmware Programming

Consider any errors or warnings to be defects during these operations. Info messages are expected, and are intended to provide the user with feedback about the status of flash operations.

  • Force the device back to the FX3 bootloader via: bladeRF-cli -e 'jump_to_boot' -v debug

  • Power cycle the device. It should continue to boot into the bootloader, instead of the firmware.

  • "Recover" the device and Flash firmware via the bladeRF-cli:

bladeRF> recover <bus> <addr> bladeRF_fw_vX.Y.Z.img
bladeRF> open
bladeRF> load fx3 bladeRF_fw_vX.Y.Z.img
  • Power cycle the device and verify that it boots the desired firmware, and that the calibration data is intact: bladeRF-cli -e 'version' -e 'info' -v debug

  • Flash a different version of firmware via: bladeRF-cli -f bladeRF_fw_vA.B.C.img -v debug

  • Power cycle the device and verify that the new firmware has been loaded, and that the calibration data is correct.


FPGA Loading

FPGA Flash-based Autoloading

Consider any errors or warnings to be defects during these operations. Info messages are expected, and are intended to provide the user with feedback about the status of flash operations.

Ensure that you do not have any FPGA bitstreams on the host set up for host-based autoloading.

  • Erase any previously stored FPGA images: bladeRF-cli -L X -v debug

  • Power cycle the device, and verify that this has not corrupted the calibration data or firmware: bladeRF-cli -e 'version' -e 'info' -v debug

  • Write an FPGA bitstream to flash: bladeRF-cli -L <fpga image> -v debug

  • Power cycle the device and verify that the firmware boots, the FPGA is loaded, and that the calibration data has not been corrupted. bladeRF-cli -e 'version' -e 'info' -v debug

    • Ensure this works when powered from a USB host, from a USB charger, and from a power supply connected to the DC barrel plug.
  • Erase the FPGA and perform the associated verifications.

Host-based Autoloading

  • Place a FPGA image to an autoload directory.

  • Power cycle the device.

  • Verify that the intended FPGA is loaded: bladeRF-cli -e 'version' -e 'info' -v debug

Manual FPGA Loading

  • Verify that FPGA loading via bladeRF-cli -l <fpga> -v debug succeeds.

  • Verify that the expected version was loaded: bladeRF-cli -e 'version' -e 'info' -v debug.

Load a different version with the FPGA already loaded, and preform the verification step again.


Basic Device Control: bladeRF

These device control operations may be performed with the bladeRF-cli, or for simplicity, third party tools, as they should utilize the same underlying libbladeRF API calls.

A signal generator and spectrum analyzer are required for many of these items.

Frequency Tuning

When performing these tests, make notes of the frequencies that you are using. If issue occur, it is important to know if it is when crossing "bands" or internal device ranges.

RX

  • Verify reception of reference signals at various frequencies in 300 MHz - 3.8 GHz, without a XB-200 attached.
  • Verify reception of reference signals at various frequencies up to 3.8 GHz, with a XB-200 attached, using each filter configuration:
    • Auto 3 dB filterbank
    • Auto 1 dB filterbank
    • 6m filterbank
    • 2m filterbank
    • 1.25m filterbank
    • Custom path

TX

Suggested reference signal: W-CDMA QPSK test pattern

  • Verify that a VSA can receive and demodulate reference signals at frequencies in 300 MHz - 3.8 GHz, without a XB-200 attached.
  • Verify that a VSA can receive and demodulate reference signals at frequencies up to 3.8 GHz, with a XB-200 attached, using each filter configuration:
    • Auto 3 dB filterbank
    • Auto 1 dB filterbank
    • 6m filterbank
    • 2m filterbank
    • 1.25m filterbank
    • Custom path

Gain

RX

Supply the device with a reference signal and verify the effect of changing the gain values for:

  • Overall system gain (set gain rx <dB>)
  • RX LNA
  • RX VGA1
  • RX VGA2

TX

Transmit a reference signal to a VSA and verify the effect of changing the gain values for:

  • Overall system gain (set gain tx <dB>)
  • TX VGA1
  • TX VGA2

Bandwidth

RX

  • Supply a reference signal of a wide bandwith and verify the effect of changing the RX bandwidth.

TX

  • Transmit a reference signal of a wide bandwith to a VSA and verify the effect of changing the TX bandwidth.

DC Offset

Repeat the following steps for a variety of frequency/gain combinations and make note of the results.

  • Use the bladeRF-cli to perform an initial cal lms at a desired frequency and gain.

  • In the CLI, perform a cal dc rx operation, and note the effect on received IQ values and on an FFT plot.

  • Create and load an RX calibration table. Verify that corrections are applied over the range of the table.

  • In the CLI, perform a cal dc tx operation, and transmit a tone into a VSA. Make note of the difference in the DC offset.

  • Create and load a TX calibration table. Verify that corrections are applied over the range of the table.


Basic Device Control: bladeRF Micro

These device control operations may be performed with the bladeRF-cli, or for simplicity, third party tools, as they should utilize the same underlying libbladeRF API calls.

A signal generator and spectrum analyzer are required for many of these items.

Frequency Tuning

When performing these tests, make notes of the frequencies that you are using. If issue occur, it is important to know if it is when crossing "bands" or internal device ranges.

RX

  • Verify reception of reference signals at various frequencies in 70 MHz - 6 GHz
    • RX1
    • RX2
  • Verify that print rssi prints reasonable values that vary based on signal level.
    • RX1
    • RX2

TX

Suggested reference signal: W-CDMA QPSK test pattern

  • Verify that a VSA can receive and demodulate reference signals at frequencies in 70 MHz - 6.0 GHz
    • TX1
    • TX2

Gain

RX

Supply the device with a reference signal and verify the effect of changing the gain values for:

  • Overall system gain (set gain rx{1,2} <dB>)
    • RX1
    • RX2
  • full stage (set gain rx{1,2} full <dB>)
    • RX1
    • RX2

TX

Transmit a reference signal to a VSA and verify the effect of changing the gain values for:

  • Overall system gain (set gain tx{1,2} <dB>)
    • TX1
    • TX2
  • dsa stage (set gain tx{1,2} dsa <dB>)
    • TX1
    • TX2

Bandwidth

RX

  • Supply a reference signal of a wide bandwith and verify the effect of changing the RX bandwidth.
    • RX1
    • RX2

TX

  • Transmit a reference signal of a wide bandwith to a VSA and verify the effect of changing the TX bandwidth.
    • TX1
    • TX2

ADF4002

Connect a 10 MHz reference source (e.g. GPSDO, or 10 MHz ref output from VSA) to REFIN (J95). The signal should be [TODO: voltage here ... sine/square?]

Note: Ensure the VSA is synchronized to the same clock.

Caution: UFL connectors are fragile and have limited mating cycles.

Transmit a reference signal (e.g. QPSK constellation into EVM analysis, or a tone into spectrum analysis) into a VSA.

  • With the ADF4002 disabled (set adf_enable 0), note the frequency offset of the reference signal.
    • print adf_enable should report ADF Chip Enable: disabled
  • Enable the ADF4002 (set adf_enable 1).
    • print adf_enable should report ADF Chip Enable: enabled and ADF Chip Locked: locked
    • The frequency offset of the reference signal should be very small, within a few Hz.
    • print trimdac should report 0xc000 for current VCTCXO trim
  • Change the ADF4002's reference frequency: set adf_refclk 9M
    • print adf_enable should report ADF Chip Enable: enabled and ADF Chip Locked: unlocked
  • Change the ADF4002's reference frequency back to 10 MHz: set adf_refclk 10M
    • print adf_enable should report ADF Chip Enable: enabled and ADF Chip Locked: locked
  • Disable the reference clock while the ADF is enabled (note: don't do this on the UFL connector)
    • print adf_enable should report ADF Chip Enable: enabled and ADF Chip Locked: unlocked
  • Disable the ADF4002 (set adf_enable 0)
    • print adf_enable should report ADF Chip Enable: disabled
    • print trimdac should NOT report 0xc000 for current VCTCXO trim, but should instead return to its previous value.

Bias tees

Note: disconnect RX1/RX2/TX1/TX2 when testing the bias tees to avoid possible damage to test equipment.

  • Disable the bias tees: set biastee rx 0 ; set biastee tx 0
  • Verify 0 VDC between RX1/RX2/TX1/TX2 center pin and ground
  • Enable the RX bias tees: set biastee rx 1
  • Verify ~5 VDC between the center pins of RX1 & RX2 and ground, and 0 VDC between the center pins of TX1 & TX2 and ground
  • Enable the TX bias tees: set biastee tx 1
  • Verify ~5 VDC between the center pins of RX1/RX2/TX1/TX2 and ground
  • Disable the RX bias tees: set biastee rx 0
  • Verify ~5 VDC between the center pins of TX1 & TX2 and ground, and 0 VDC between the center pins of RX1 & RX2 and ground
  • Disable the TX bias tees: set biastee tx 0
  • Verify 0 VDC between RX1/RX2/TX1/TX2 center pin and ground

libbladeRF test programs

The libbladeRF test programs are intended to exercise specific functionalities provided by the API. However, these programs inherently exercise device control functionality not included above. A selection of the libbladeRF_test_* programs is provided here to attempt to maximize coverage with a small number of programs.

libbladeRF_test_ctrl

This test ensures that various device control calls complete without error.

Run all of its tests on bladeRF1 without an XB-200 attached.

  • sampling
  • lpf_mode
  • enable_module
  • loopback
  • rx_mux
  • correction
  • samplerate
  • bandwidth
  • gain
  • frequency
  • threads

Run all of its tests on bladeRF1 with an XB-200 attached.

  • sampling
  • lpf_mode
  • enable_module
  • loopback
  • rx_mux
  • xb200
  • correction
  • samplerate
  • bandwidth
  • gain
  • frequency
  • threads

Run all of its tests on bladeRF2 with FPGA-based tuning (default).

  • sampling
  • lpf_mode
  • enable_module
  • loopback
  • rx_mux
  • correction
  • samplerate
  • bandwidth
  • gain
  • frequency
  • threads

Run all of its tests on bladeRF2 with host-based tuning (BLADERF_DEFAULT_TUNING_MODE='host').

  • sampling
  • lpf_mode
  • enable_module
  • loopback
  • rx_mux
  • correction
  • samplerate
  • bandwidth
  • gain
  • frequency
  • threads

libbladeRF_test_rx_discont

  • Run this program over a variety of valid sample rates (for the particular USB connection). No discontinuities should occur.

  • Run this program over for sample rates that your USB controller does not support. Discontinuities should be reported.

libbladeRF_test_repeater

This program uses the async interface to immediately retransmit received samples.

Supply a reference signal from a signal generator, and verify that it can be demodulated without failures on a VSA.

Record the sample rates, bandwidths, frequencies, and stream parameters used.

  • Verify that transmitted samples are representative of the received signals.

libbladeRF_test_timestamps

This program verifies timestamp functionality. Run all available test at a variety of sample rates. Not that some tests (e.g., loopback_onoff) are not designed to support high sample rates.

  • rx_gaps
  • rx_scheduled
  • tx_onoff
  • tx_onoff_nowsched
  • tx_gmsk_bursts
  • loopback_onoff
  • format_mismatch
  • readback

libbladeRF sample programs

As these are intended to exemplify API usage, these programs must be correct.

Verify that the following run as intended:

  • libbladeRF_example_open_via_serial
  • libbladeRF_example_sync_rxtx
  • libbladeRF_example_sync_rx_meta
  • libbladeRF_example_sync_tx_meta

Third-party Tools

Testing against third-party tools with bladeRF support to ensure changes have not introduced compatibility issues is highly recommended.

Make note of the tool and its version, for future comparison, should any issues arise.